fix: pipeline MACD 参数统一为标准 12/26/9（与 web/交易所一致）

LWC 折线无法画真竖线；增量刷新时用坐标采样补刷竖边，避免与横边脱节。

Co-authored-by: Cursor <cursoragent@cursor.com>

.gitignore

@@ -1,59 +1,49 @@
 # MacOS
 .DS_Store
 
-# Python
+# Python编译文件和缓存
 __pycache__/
 *.py[cod]
 *$py.class
 *.pyc
 *.pyo
-.pytest_cache/
 
-# Logs & databases
+# 策略文件的缓存
+strategies/__pycache__/
+
+# Machine Learning / AI model files
+*_model*_xgb_model.json
+*modelchan*.json
+*.libsvm
+feature_meta
+*_model_feature_data.csv
+*.pem
+
+# Log files
 *.log
+
+# Database files
 *.sqlite
 *.sqlite-shm
 *.sqlite-wal
-
+.DS_Store
-# Office documents kept alongside the repo but not part of it.
+交易记录/~$交易规则.docx
-# "~$" files are Excel's lock files, recreated every time a workbook is opened.
+/datasvc/data
-*.xlsx
+.DS_Store
-*.xls
+.DS_Store
-~$*
+/data_provider/data
-
+.DS_Store
-# Local data
+.DS_Store
-data/
+.DS_Store
-
+.DS_Store
-# Exchange history fetched by research/live/*.py. 40MB and re-fetchable from
+.DS_Store
-# the venue, so it stays local; the small result CSVs it feeds are committed.
+.DS_Store
-research/live/cache/
+data_provider/._config.json
-
-# Scratch outputs from short shakedown runs, superseded by the real collection.
-research/out/archive/
-
-# Per-trade simulation dumps from research/step*.py. 70MB+ and regenerable by
-# rerunning the step; the summaries they feed live in HANDOFF.md.
-research/out/*.feather
-
-# Virtualenvs. venv writes its own .gitignore since 3.11, but only for the
-# directory it creates — declare it here so other layouts are covered too.
-.venv/
-venv/
-
-# Local tooling
 .gstack/
-research/out/*.jsonl.gz
-research/out/penetration.csv
-research/out/shadow_*.csv
-research/out/run_meta_*.json
 
-# Telegram 凭据。**不要提交**
+# ESS gate / engineering-loop working dirs（归档进 docs/runs/）
-research/live/deploy/tg.env
+.gates/
+loop/
 
-# 生产状态与信号总线。刻意放在仓库外（LIVE_HOME / BUS_DIR），这几条只防
+# Crypto Wyckoff Screener local cache
-# 有人把它们指回仓库里：里面是日亏损累计与已处理信号键，被 git clean
+data/crypto_wyckoff/
-# 清掉等于两道闸静默失忆
-/live/deploy/live.env
-live_state.json
-live_trades.jsonl
-signals_live.jsonl

AGENTS.md

@@ -0,0 +1,33 @@
+# chan — Agent Entry
+
+本仓受 ESS 约束。不要一上来扫全库或加载全部 governance。
+
+## Boot
+
+1. `docs/PROJECT_PROFILE.md`
+2. `docs/PROJECT_RULES.md`
+3. `docs/STATE/CURRENT.md` + `docs/AGENT_MEMORY.md`
+4. 有进行中任务再读 `docs/TASKS/` / 对应 ECR / HANDOFF
+5. 角色文件：ESS 根目录 `agents/{ARCHITECT|ENGINEER|REVIEWER|RELEASE_MANAGER}.md`
+
+## Roles（选一）
+
+| 意图 | 角色 |
+|------|------|
+| 规格 / 架构 / ECR | ARCHITECT |
+| 实现 / 修 bug | ENGINEER |
+| 审阅 | REVIEWER |
+| 发版 / tag | RELEASE_MANAGER |
+
+## Never
+
+- 无 ECR 改 `config/` / `strategies/` 交易逻辑
+- 无 ADR 改缠论算法语义
+- 无 ECR 删减 `/api/analyze` 字段
+- 把聊天记录当成完成；阶段结束须落盘 `docs/`
+
+## Pointers
+
+- TRACEABILITY: `docs/TRACEABILITY.md`
+- CHANGELOG: `docs/CHANGELOG/CHANGELOG.md`
+- 人类向导：`CLAUDE.md`

CLAUDE.md

@@ -0,0 +1,116 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Project Overview
+
+缠论 (Chan Theory) technical analysis system for Freqtrade. Implements Chan Zhong Shui Chan's theory for crypto/stock trading, including fractal (分型), stroke (笔), segment (线段), pivot/center (中枢), and buy/sell point (买卖点) detection.
+
+## Governance
+
+- Agent 入口：`AGENTS.md`（boot 顺序）· `docs/PROJECT_PROFILE.md` · `docs/AGENT_MEMORY.md` · `docs/STATE/CURRENT.md`
+- ESS 文档：`docs/ECR/`、`docs/ENGINEERING_SPEC/`、`docs/TRACEABILITY.md`、`docs/CHANGELOG/`
+- **正式引擎包**：`chanlun/`；strategies / web 已用 `from chanlun import ...`
+- 根目录 `Chan*.py` / `TF_DF.py` 仍为 **兼容 shim**（旧脚本可用）
+- 变更分级：无 ECR 不改 strategies/config；无 ADR 不改缠论算法语义
+
+## Core Architecture
+
+### Chan Theory Engine (`chanlun/`)
+
+```text
+chanlun/
+  core/          # KLU KLC BI SBI SEG ZS BIZS BSP Enum CTime
+  pipeline/      # orchestrator(ChanLun) + timeframe(TF_DF) + builders/
+  indicators/    # ChanMACD*
+  analysis/      # Zone Classifier Pivot Heng PY Find_Trend ...
+```
+
+Data processing pipeline (each step feeds the next):
+
+1. **`chanlun.core.ChanKLU`** — Raw K-line unit with TA indicators and pattern recognition
+2. **`chanlun.core.ChanKLC`** — Combined K-line: inclusion + fractal; `.next`/`.pre` linked list
+3. **`chanlun.core.ChanBI`** — Stroke (笔)
+4. **`chanlun.core.ChanSBI`** — Special stroke → SEG
+5. **`chanlun.core.ChanSEG`** — Segment (线段)
+6. **`chanlun.core.ChanZS`** / **`ChanBIZS`** — Centers (中枢)
+7. **`chanlun.core.ChanBSP`** — Buy/Sell points
+8. **`chanlun.pipeline.orchestrator.ChanLun`** — Orchestrator
+9. **`chanlun.pipeline.timeframe.TF_DF`** — Timeframe facade；实现拆在 `pipeline/builders/`
+
+### Services
+
+- **外部 DATA_SERVICE** — 行情服务（env: `DATA_SERVICE_URL`）；本仓库可不含 data_provider 源码
+- **`web/`** — Flask UI：`create_app()` + `api/` blueprints + `services/`；前端 `static/js/app/`。默认端口见 `web/config.py`（`FLASK_PORT`，常见 8128）
+- **`strategies/`** — Freqtrade strategies（本 ECR 不改）
+- **`config/`** — Freqtrade configs（本 ECR 不改）
+
+### Data Flow
+
+```
+Exchange / DATA_SERVICE → Freqtrade Strategy / web → ChanLun → TF_DF
+  → KLU → KLC → BI → SBI → SEG → ZS → BSP
+```
+
+## Common Commands
+
+### Freqtrade Trading
+
+```bash
+# Live trade
+freqtrade trade -c ./user_data/Chan/config/<config>.json --strategy <StrategyName> --strategy-path ./user_data/Chan/strategies
+
+# Backtest
+freqtrade backtesting -c ./user_data/Chan/config/<config>.json --strategy <StrategyName> --strategy-path ./user_data/Chan/strategies --timerange=20251008-
+
+# Download data
+freqtrade download-data -c ./user_data/Chan/config/<config>.json -t 1m 1h 1d --pairs BTC/USDT:USDT --timerange=20240101-
+
+# Hyperopt
+freqtrade hyperopt --hyperopt-loss SharpeHyperOptLossDaily --spaces roi --strategy <StrategyName> --strategy-path ./user_data/Chan/strategies -c ./user_data/Chan/config/<config>.json -e 200 --timerange=20250201-20250901
+
+# Plot
+freqtrade plot-dataframe --strategy <StrategyName> --datadir user_data/data/binance -c ./user_data/Chan/config/<config>.json --timerange=20250721-
+```
+
+### Data Provider
+
+```bash
+# Docker
+cd data_provider && docker compose up -d
+
+# Direct
+cd data_provider && python main.py
+
+# With custom config
+CONFIG_PATH=./config.json python main.py
+```
+
+### Web UI
+
+```bash
+cd web && python app.py
+# or via gunicorn:
+gunicorn -w 4 -b 0.0.0.0:8123 app:app
+
+# Deploy scripts:
+cd web && ./deploy.sh          # standard
+cd web && ./deploy_venv.sh     # Ubuntu 22.04+ (venv)
+```
+
+### Docker (Freqtrade)
+
+```bash
+sudo docker compose run --rm chanlun_btc backtesting -c ./user_data/Chan/config/<config>.json --strategy <StrategyName> --strategy-path ./user_data/Chan/strategies --timerange=20250721-
+```
+
+## Key Conventions
+
+- All Chan theory classes are prefixed with `Chan` (e.g., `ChanBI`, `ChanZS`)
+- Strategies import `ChanLun` and add `sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))` to import from parent
+- MACD params: `MACD(26, 52, 9)` by default (slow period 52 instead of standard 26)
+- Enums in `ChanEnum.py` use `auto()` values
+- `ChanKLC` is a linked-list style data structure with `.next`/`.pre` pointers
+- The `TF_DF` class is the primary data container per timeframe
+- K-line direction uses `Chan_KLINE_DIR` (UP/DOWN/COMBINE/INCLUDED)
+- All text comments/commits are in Chinese

ChanBI.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanBI import *  # noqa: F403

ChanBIZS.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanBIZS import *  # noqa: F403

ChanBSP.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanBSP import *  # noqa: F403

ChanCTime.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanCTime import *  # noqa: F403

ChanEnum.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanEnum import *  # noqa: F403

ChanHeng.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.analysis.ChanHeng import *  # noqa: F403

ChanKLC.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanKLC import *  # noqa: F403

ChanKLU.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanKLU import *  # noqa: F403

ChanLun.py

@@ -0,0 +1,3 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.pipeline.orchestrator import ChanLun  # noqa: F401
+from chanlun.pipeline.timeframe import TF_DF  # noqa: F401

ChanLun_Classifier.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.analysis.ChanLun_Classifier import *  # noqa: F403

ChanMACD.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.indicators.ChanMACD import *  # noqa: F403

ChanMACDHistSet.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.indicators.ChanMACDHistSet import *  # noqa: F403

ChanMACDSeg.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.indicators.ChanMACDSeg import *  # noqa: F403

ChanMACDUnitTF.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.indicators.ChanMACDUnitTF import *  # noqa: F403

ChanMacro/.env.example

@@ -0,0 +1,12 @@
+# Data Provider URL (existing chan data_provider service)
+PROVIDER_URL=http://127.0.0.1:80
+
+# Database path
+DB_PATH=data/macro.db
+
+# Telegram (reuse bsp_monitor config)
+# TELEGRAM_BOT_TOKEN=your_bot_token
+# TELEGRAM_CHAT_ID=your_chat_id
+
+# AI API (for daily report, Phase 5+)
+# ANTHROPIC_API_KEY=sk-ant-...

ChanMacro/.gitignore

@@ -0,0 +1 @@
+data/

ChanMacro/__init__.py

@@ -0,0 +1,9 @@
+"""
+ChanMacro — Crypto Market Memory System (Signal Expectancy Engine).
+
+V1: 4 factors (Price Structure, Breadth, OI State, Volatility Regime)
+    3 regimes (TREND / RANGE / PANIC)
+    Factor-locked: Regime = f(Price, Breadth, Vol) — forever.
+"""
+
+__version__ = "1.0.0"

ChanMacro/chan_integration.py

@@ -0,0 +1,224 @@
+"""
+chan_integration.py — 缠论引擎集成：检测 BSP 信号并写入 signal_features。
+
+复用 bsp_monitor/engine.py 的 ChanEngine 管线，对历史日线数据批量跑缠论，
+提取 B1/B2/B3/S1/S2/S3 信号，通过 SignalTracker 记录到 signal_features。
+"""
+
+import sys
+import os
+from datetime import date as Date, timedelta
+from typing import List, Optional
+import logging
+
+# 确保 Chan 引擎在路径上（与 bsp_monitor/engine.py 相同的路径设置）
+_PARENT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+if _PARENT not in sys.path:
+    sys.path.insert(0, _PARENT)
+
+import pandas as pd
+
+from ChanEnum import Chan_BSP_TYPE, Chan_BSP_DIR
+from ChanBSP import ChanBSP
+
+logger = logging.getLogger(__name__)
+
+
+class ChanSignalDetector:
+    """
+    对历史日线数据运行缠论管线，提取所有 BSP 信号。
+
+    Usage:
+        detector = ChanSignalDetector()
+        signals = detector.detect_from_db("2026-01-01", "2026-06-24")
+        # → [{"date": Date, "signal_type": "B3", "entry_price": 96500, ...}, ...]
+    """
+
+    def __init__(self):
+        from TF_DF import TF_DF as _TF_DF_Class
+        self._TF_DF_Class = _TF_DF_Class
+
+    def detect_from_db(self, start_date: str, end_date: str) -> list[dict]:
+        """从数据库加载日线数据，跑缠论管线，提取信号。"""
+        from database import get_connection
+        conn = get_connection()
+        df = pd.read_sql_query(
+            "SELECT date, open, high, low, close, volume "
+            "FROM ohlcv_daily WHERE symbol='BTC/USDT:USDT' "
+            "AND date BETWEEN ? AND ? ORDER BY date",
+            conn, params=(start_date, end_date)
+        )
+        conn.close()
+
+        if df.empty or len(df) < 50:
+            logger.warning(f"日线数据不足: {len(df)} 根")
+            return []
+
+        return self.detect_from_df(df)
+
+    def detect_from_df(self, df: pd.DataFrame) -> list[dict]:
+        """从 DataFrame 运行缠论管线，提取 BSP 信号。"""
+        # 需要 datetime 列才能跑 TF_DF
+        df = df.copy()
+        df["timestamp"] = pd.to_datetime(df["date"])
+        df["date"] = df["timestamp"]
+
+        try:
+            engine = self._build_engine(df)
+        except Exception as e:
+            logger.error(f"缠论管线失败: {e}")
+            return []
+
+        return self._extract_signals(engine)
+
+    def _build_engine(self, df: pd.DataFrame):
+        """构建缠论管线（对齐 bsp_monitor/engine.py 的 ChanEngine）。"""
+        from TF_DF import TF_DF as _TF_DF_Class
+
+        if df.empty or len(df) < 50:
+            raise ValueError(f"数据不足: {len(df)} 根 K 线")
+
+        if "date" not in df.columns and "timestamp" in df.columns:
+            df["date"] = df["timestamp"]
+
+        # 使用 __new__ 避免触发 TF_DF.__init__
+        engine = type('ChanEngine', (), {})()  # 简单容器
+        tf = _TF_DF_Class.__new__(_TF_DF_Class)
+
+        df_with_indicators = tf.add_indicators(df.copy())
+        engine.klu_list = tf.get_klu_list(df_with_indicators)
+        engine.klc_list = tf.get_klc_list(engine.klu_list)
+        engine.bi_list = tf.cal_bi_list(engine.klc_list)
+        engine.seg_list = tf.get_seg_list(engine.bi_list)
+        engine.bi_zs_list = tf.cal_bi_zs(engine.seg_list)
+        engine.bsp_list = tf.find_all_bsp(engine.bi_list, engine.bi_zs_list)
+
+        return engine
+
+    def _extract_signals(self, engine) -> list[dict]:
+        """从 ChanEngine 输出中提取所有 BSP 信号。"""
+        signals = []
+        for bsp in engine.bsp_list:
+            if bsp.type == Chan_BSP_TYPE.NONE:
+                continue
+            if bsp.klc is None:
+                continue
+
+            signal_type = self._bsp_type_str(bsp.type)
+            entry_price = bsp.klc.close
+            signal_date = self._klc_date(bsp.klc)
+
+            if signal_date is None:
+                continue
+
+            # 信号质量：根据分型强度判断
+            strength = self._calc_strength(bsp)
+            grade = "A" if strength >= 70 else "B" if strength >= 50 else "C"
+
+            signals.append({
+                "date": signal_date,
+                "signal_type": signal_type,
+                "entry_price": float(entry_price),
+                "signal_grade": grade,
+                "signal_strength": float(strength),
+            })
+
+        details = ", ".join(f"{s['signal_type']}({s['date']})" for s in signals)
+        logger.info(f"检测到 {len(signals)} 个信号: {details}")
+        return signals
+
+    def populate_signal_features(self, start_date: str = "2024-01-01",
+                                  end_date: Optional[str] = None) -> int:
+        """
+        完整流程：检测信号 → 计算市场状态 → 写入 signal_features。
+
+        Returns: 写入的信号数量。
+        """
+        if end_date is None:
+            end_date = Date.today().isoformat()
+
+        logger.info(f"开始信号检测: {start_date} → {end_date}")
+
+        # Step 1: 检测缠论信号
+        signals = self.detect_from_db(start_date, end_date)
+        if not signals:
+            logger.warning("未检测到任何 BSP 信号")
+            return 0
+
+        # Step 2: 去重 — 跳过已存在的信号
+        from database import get_connection
+        conn = get_connection()
+        existing = set()
+        for row in conn.execute(
+            "SELECT date, signal_type FROM signal_features"
+        ).fetchall():
+            existing.add((row[0], row[1]))
+        conn.close()
+
+        new_signals = [s for s in signals
+                       if (str(s["date"]), s["signal_type"]) not in existing]
+        if not new_signals:
+            logger.info("所有信号已存在，跳过")
+            return 0
+
+        # Step 3: 写入 signal_features
+        from expectancy.tracker import SignalTracker
+        tracker = SignalTracker()
+        count = tracker.backfill_signals(new_signals)
+
+        logger.info(f"信号入库完成: {count}/{len(signals)}")
+        return count
+
+    @staticmethod
+    def _bsp_type_str(t: Chan_BSP_TYPE) -> str:
+        mapping = {
+            Chan_BSP_TYPE.B1: "B1", Chan_BSP_TYPE.B2: "B2", Chan_BSP_TYPE.B3: "B3",
+            Chan_BSP_TYPE.S1: "S1", Chan_BSP_TYPE.S2: "S2", Chan_BSP_TYPE.S3: "S3",
+        }
+        return mapping.get(t, "UNKNOWN")
+
+    @staticmethod
+    def _klc_date(klc) -> Optional[Date]:
+        """从 KLC 提取信号确认日期。"""
+        end_time = getattr(klc, "end_time", None)
+        if end_time is None:
+            start_time = getattr(klc, "start_time", None)
+            if start_time is None:
+                return None
+            end_time = start_time
+        if hasattr(end_time, "date"):
+            return end_time.date()
+        if isinstance(end_time, str):
+            return Date.fromisoformat(end_time[:10])
+        return None
+
+    @staticmethod
+    def _calc_strength(bsp: ChanBSP) -> float:
+        """根据 BSP 特征计算信号强度 0-100。"""
+        score = 50.0
+        klc = bsp.klc
+        if klc is None:
+            return score
+
+        # 分型强度
+        from ChanEnum import Chan_KLC_FX
+        fx = getattr(klc, "klc_fx_type", None)
+        if fx is not None:
+            strong_fxs = {Chan_KLC_FX.TOP2, Chan_KLC_FX.TOP3, Chan_KLC_FX.BOTTOM2, Chan_KLC_FX.BOTTOM3}
+            medium_fxs = {Chan_KLC_FX.TOP1, Chan_KLC_FX.BOTTOM1, Chan_KLC_FX.TOP4, Chan_KLC_FX.BOTTOM4}
+            if fx in strong_fxs:
+                score += 25
+            elif fx in medium_fxs:
+                score += 10
+
+        # BSP 类型
+        if bsp.type in (Chan_BSP_TYPE.B1, Chan_BSP_TYPE.S1):
+            score += 10  # 一类买卖点: 背驰确认, 额外加分
+
+        # 笔特征
+        bi = getattr(bsp, "bi", None)
+        if bi and hasattr(bi, "height") and hasattr(bi, "width"):
+            if bi.width > 3 and abs(bi.height) > 100:
+                score += 10
+
+        return min(score, 100.0)

ChanMacro/cli.py

@@ -0,0 +1,464 @@
+"""
+cli.py — Command-line interface for ChanMacro.
+"""
+
+import argparse
+import json
+import logging
+import time
+from datetime import date as Date, datetime, timedelta
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s [%(levelname)s] %(name)s: %(message)s",
+)
+logger = logging.getLogger("chanmacro")
+
+
+def parse_date(date_str: str) -> Date:
+    """Parse YYYY-MM-DD string to Date."""
+    return datetime.strptime(date_str, "%Y-%m-%d").date()
+
+
+def _build_market_state(target: Date) -> tuple:
+    """Shared helper: compute all scores → (MarketStateVector, RegimeResult)."""
+    from config import config
+    from scoring.price_structure import PriceStructureScorer
+    from scoring.breadth_scorer import BreadthScorer
+    from scoring.oi_matrix import OIMatrixScorer
+    from scoring.volatility_regime import VolatilityRegimeScorer
+    from regime_detector import RegimeDetector
+    from models import MarketStateVector
+
+    ps = PriceStructureScorer().compute(target)
+    br = BreadthScorer().compute(target)
+    oi = OIMatrixScorer().compute(target)
+    vol = VolatilityRegimeScorer().compute(target)
+
+    detector = RegimeDetector()
+    detector.load_state(config.db_path)
+    r = detector.detect(ps.score, br.breadth_top50, vol.vol_regime.value, target)
+
+    state = MarketStateVector(
+        date=target, regime=r.regime, regime_confidence=r.confidence,
+        regime_version=r.regime_version, regime_maturity_score=r.maturity_score,
+        breadth_top20=br.breadth_top20, breadth_top30=br.breadth_top30,
+        breadth_top50=br.breadth_top50, breadth_bucket=br.breadth_bucket,
+        breadth_divergence=br.breadth_divergence,
+        oi_state=oi.oi_state, volatility_regime=vol.vol_regime,
+        price_structure_score=ps, breadth_score=br,
+        oi_matrix_score=oi, volatility_regime_score=vol,
+    )
+    state.market_state_hash = state.compute_hash()
+
+    # Persist regime to DB so subsequent calls have correct state
+    from database import get_connection
+    conn = get_connection()
+    conn.execute("""
+        INSERT OR REPLACE INTO regime_history
+        (date, regime, confidence, regime_version, maturity_score, all_scores_json,
+         prior_regime, confirmation_days)
+        VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+    """, (
+        str(target), r.regime.value, r.confidence, r.regime_version,
+        r.maturity_score, json.dumps(r.all_scores),
+        r.prior_regime.value if r.prior_regime else None,
+        r.confirmation_days,
+    ))
+    conn.commit()
+    conn.close()
+
+    return state, r
+
+
+def cmd_fetch(args):
+    """Fetch raw data and store to DB."""
+    from database import init_db
+    from fetchers.ohlcv import OHLCVFetcher
+    from fetchers.breadth import BreadthFetcher
+
+    target = parse_date(args.date) if args.date else Date.today()
+    init_db()
+
+    module = args.module or "all"
+
+    if module in ("ohlcv", "all"):
+        logger.info(f"Fetching OHLCV for {target}...")
+        fetcher = OHLCVFetcher()
+        df = fetcher.fetch(target)
+        if not df.empty:
+            n = fetcher.store_df(df)
+            logger.info(f"OHLCV: stored {n} rows")
+
+    if module in ("breadth", "all"):
+        logger.info(f"Fetching Breadth for {target}...")
+        fetcher = BreadthFetcher()
+        record = fetcher.fetch(target)
+        if record:
+            fetcher.store(record=record)
+            logger.info(f"Breadth: stored (adv={record.get('advance_top50')}, "
+                        f"dec={record.get('decline_top50')}, "
+                        f"ema20={record.get('above_ema20_top50')})")
+
+    if module in ("derivatives", "all"):
+        logger.info(f"Fetching Derivatives for {target}...")
+        from fetchers.derivatives import DerivativesFetcher
+        fetcher = DerivativesFetcher()
+        records = fetcher.fetch(target)
+        if records:
+            n = fetcher.store(records=records)
+            logger.info(f"Derivatives: stored {n} records")
+
+
+def cmd_score(args):
+    """Compute all factor scores and regime for a date."""
+    from database import init_db
+
+    target = parse_date(args.date) if args.date else Date.today()
+    init_db()
+    logger.info(f"Computing scores for {target}...")
+
+    state, _ = _build_market_state(target)
+
+    # Output
+    ps = state.price_structure_score
+    br = state.breadth_score
+    oi = state.oi_matrix_score
+    vol = state.volatility_regime_score
+
+    print(f"\n{'='*60}")
+    print(f"  {target}  Market State")
+    print(f"{'='*60}")
+    print(f"  Regime:      {state.regime.value} (conf={state.regime_confidence:.2f}, "
+          f"v={state.regime_version})")
+    print(f"  Maturity:    {state.regime_maturity_score:.0f}/100")
+    print(f"  Breadth:     {state.breadth_bucket.value} "
+          f"(T20={state.breadth_top20:.0f} T30={state.breadth_top30:.0f} "
+          f"T50={state.breadth_top50:.0f} div={state.breadth_divergence:+.0f})")
+    print(f"  OI State:    {state.oi_state.value}")
+    print(f"  Volatility:  {state.volatility_regime.value}")
+    print(f"{'='*60}")
+    print(f"  Scores:")
+    print(f"    Price Structure:  {ps.score:.0f}  {ps.label}")
+    print(f"    Breadth:          {br.score:.0f}  {br.breadth_bucket.value}")
+    print(f"    OI Matrix:        {oi.score:.0f}  {oi.oi_state.value}")
+    print(f"    Volatility:       {vol.score:.0f}  {vol.vol_regime.value}")
+    print(f"{'='*60}")
+    print(f"  Market State Hash:  {state.market_state_hash}")
+    print()
+
+    return state
+
+
+def cmd_regime(args):
+    """Show regime history."""
+    from database import get_connection
+    days = args.days or 30
+    conn = get_connection()
+    rows = conn.execute(
+        "SELECT date, regime, confidence, maturity_score, confirmation_days "
+        "FROM regime_history ORDER BY date DESC LIMIT ?",
+        (days,)
+    ).fetchall()
+    conn.close()
+
+    print(f"\n{'='*50}")
+    print(f"  Regime History (last {days} days)")
+    print(f"{'='*50}")
+    for r in rows:
+        print(f"  {r['date']}  {r['regime']:7s}  conf={r['confidence']:.2f}  "
+              f"mat={r['maturity_score']:.0f}  days={r['confirmation_days']}")
+    print()
+
+
+def cmd_track(args):
+    """Record a trading signal with current market state."""
+    from database import init_db
+    from expectancy.tracker import SignalTracker
+
+    target = parse_date(args.date) if args.date else Date.today()
+    init_db()
+
+    logger.info(f"Recording {args.signal} on {target} @ {args.price}")
+
+    state, _ = _build_market_state(target)
+
+    tracker = SignalTracker()
+    rid = tracker.record(
+        date=target, signal_type=args.signal, entry_price=args.price,
+        state=state, signal_grade=args.grade, signal_strength=args.strength,
+    )
+    logger.info(f"Signal recorded: id={rid}")
+
+
+def cmd_backfill(args):
+    """Backfill historical breadth + regime scores."""
+    from datetime import date as Date, timedelta
+    from database import init_db, get_connection
+    from fetchers.ohlcv import OHLCVFetcher
+    from fetchers.breadth import BreadthFetcher
+    from config import config
+    import pandas as pd
+    import requests
+
+    start = parse_date(args.from_date)
+    end = parse_date(args.to_date) if args.to_date else Date.today()
+    init_db()
+
+    # Step 1: Ensure OHLCV data exists for the range
+    logger.info(f"Step 1/3: Fetching BTC OHLCV...")
+    OHLCVFetcher().store_df(OHLCVFetcher().fetch())
+
+    # Step 2: Backfill breadth — fetch TOP50 daily data and compute per date
+    logger.info(f"Step 2/3: Backfilling breadth {start} → {end}...")
+    provider_url = config.provider_url
+    all_symbol_data = {}
+
+    for sym in config.top50_symbols:
+        try:
+            df = pd.DataFrame(requests.get(
+                f"{provider_url}/api/candles",
+                params={"symbol": sym, "tf": "1d", "limit": 400},
+                timeout=30
+            ).json())
+            if not df.empty and "timestamp" in df.columns:
+                df["date"] = pd.to_datetime(df["timestamp"], unit="ms").dt.date
+                df["close"] = df["close"].astype(float)
+                df["high"] = df["high"].astype(float)
+                df["ema20"] = df["close"].ewm(20).mean()
+                all_symbol_data[sym] = df
+        except Exception as e:
+            logger.debug(f"  Skip {sym}: {e}")
+
+    logger.info(f"  Fetched {len(all_symbol_data)}/{len(config.top50_symbols)} symbols")
+
+    # Compute breadth for each date
+    conn = get_connection()
+    current = start
+    breadth_count = 0
+    while current <= end:
+        target_str = str(current)
+        try:
+            advances_50 = declines_50 = above_ema20_50 = new_highs_50 = 0
+            advances_30 = advances_20 = above_ema20_30 = above_ema20_20 = 0
+            new_highs_30 = new_highs_20 = 0
+
+            for rank, (sym, df) in enumerate(all_symbol_data.items()):
+                rows = df[df["date"] == current]
+                if rows.empty:
+                    continue
+                row = rows.iloc[0]
+                prev_rows = df[df["date"] < current]
+                if prev_rows.empty:
+                    continue
+                prev = prev_rows.iloc[-1]
+
+                if row["close"] > prev["close"]:
+                    if rank < 50: advances_50 += 1
+                    if rank < 30: advances_30 += 1
+                    if rank < 20: advances_20 += 1
+                elif row["close"] < prev["close"]:
+                    if rank < 50: declines_50 += 1
+
+                if not pd.isna(row.get("ema20")) and row["close"] > row["ema20"]:
+                    if rank < 50: above_ema20_50 += 1
+                    if rank < 30: above_ema20_30 += 1
+                    if rank < 20: above_ema20_20 += 1
+
+                recent_highs = df[(df["date"] < current) & (df["date"] >= current - timedelta(days=20))]
+                if not recent_highs.empty and row["high"] > recent_highs["high"].max():
+                    if rank < 50: new_highs_50 += 1
+                    if rank < 30: new_highs_30 += 1
+                    if rank < 20: new_highs_20 += 1
+
+            conn.execute("""INSERT OR REPLACE INTO breadth_daily
+                (date, total_tracked, advance_top50, decline_top50, above_ema20_top50,
+                 new_highs_20d_top50, advance_top30, advance_top20,
+                 above_ema20_top30, above_ema20_top20, new_highs_20d_top30, new_highs_20d_top20)
+                VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)""",
+                (target_str, len(all_symbol_data),
+                 advances_50, declines_50, above_ema20_50, new_highs_50,
+                 advances_30, advances_20, above_ema20_30, above_ema20_20,
+                 new_highs_30, new_highs_20))
+            breadth_count += 1
+        except Exception as e:
+            logger.debug(f"  Breadth skip {current}: {e}")
+        current += timedelta(days=1)
+
+    conn.commit()
+    logger.info(f"  Breadth backfill: {breadth_count} days")
+
+    # Step 3: Compute regime scores for each date
+    logger.info(f"Step 3/3: Computing regime scores {start} → {end}...")
+    from scoring.price_structure import PriceStructureScorer
+    from scoring.breadth_scorer import BreadthScorer
+    from scoring.oi_matrix import OIMatrixScorer
+    from scoring.volatility_regime import VolatilityRegimeScorer
+    from regime_detector import RegimeDetector
+
+    detector = RegimeDetector()
+    current = start
+    score_count = 0
+    while current <= end:
+        try:
+            ps = PriceStructureScorer().compute(current)
+            br = BreadthScorer().compute(current)
+            if br.score == 50.0 and br.label == "No Data":
+                current += timedelta(days=1)
+                continue
+            oi = OIMatrixScorer().compute(current)
+            vol = VolatilityRegimeScorer().compute(current)
+            r = detector.detect(ps.score, br.breadth_top50, vol.vol_regime.value, current)
+
+            conn.execute("""INSERT OR REPLACE INTO regime_history
+                (date, regime, confidence, regime_version, maturity_score,
+                 all_scores_json, confirmation_days)
+                VALUES (?, ?, ?, ?, ?, ?, ?)""",
+                (str(current), r.regime.value, r.confidence, r.regime_version,
+                 r.maturity_score, json.dumps(r.all_scores), r.confirmation_days))
+            score_count += 1
+            if score_count % 30 == 0:
+                conn.commit()
+                logger.info(f"  Scored {score_count} days... ({current})")
+        except Exception as e:
+            logger.debug(f"  Score skip {current}: {e}")
+        current += timedelta(days=1)
+
+    conn.commit()
+    conn.close()
+    logger.info(f"Backfill complete: {breadth_count} breadth + {score_count} regime days")
+
+
+def cmd_expectancy(args):
+    """Query signal expectancy for current market state."""
+    from database import init_db
+    from expectancy.engine import BayesianExpectancyEngine
+
+    target = parse_date(args.date) if args.date else Date.today()
+    init_db()
+
+    state, _ = _build_market_state(target)
+
+    engine = BayesianExpectancyEngine()
+    signal = args.signal or "B3"
+    report = engine.estimate(state, signal_type=signal, target_date=target)
+
+    print(f"\n{'='*60}")
+    print(f"  {target}  Signal Expectancy: {signal}")
+    print(f"{'='*60}")
+    print(f"  Regime:      {state.regime.value} (conf={state.regime_confidence:.2f})")
+    print(f"  Breadth:     {state.breadth_bucket.value} (T50={state.breadth_top50:.0f})")
+    print(f"  OI State:    {state.oi_state.value}")
+    print(f"  Volatility:  {state.volatility_regime.value}")
+    print(f"{'='*60}")
+
+    for layer in report.layers:
+        print(f"  {layer.name:15s}  N={layer.samples:4d}  eff={layer.effective_samples:.0f}  "
+              f"raw={layer.raw_winrate or 0:.1%}  post={layer.posterior_winrate:.1%}  "
+              f"ret={layer.avg_return or 0:+.1f}%")
+
+    print(f"{'='*60}")
+    print(f"  Final:  {report.final_estimate:.1%}  "
+          f"(sufficiency={report.sufficiency.value}, source={report.source})")
+    if report.profit_factor:
+        print(f"  PF={report.profit_factor}  MAE={report.max_adverse_excursion}%")
+    print()
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="ChanMacro — Crypto Market Memory System"
+    )
+    sub = parser.add_subparsers(dest="command", help="Commands")
+
+    # fetch
+    p_fetch = sub.add_parser("fetch", help="Fetch raw data")
+    p_fetch.add_argument("--date", help="Target date (YYYY-MM-DD)")
+    p_fetch.add_argument("--module", choices=["ohlcv", "breadth", "derivatives", "all"])
+
+    # score
+    p_score = sub.add_parser("score", help="Compute scores and regime")
+    p_score.add_argument("--date", help="Target date (YYYY-MM-DD)")
+
+    # regime
+    p_regime = sub.add_parser("regime", help="Show regime history")
+    p_regime.add_argument("--days", type=int, default=30)
+
+    # track
+    p_track = sub.add_parser("track", help="Record a trading signal")
+    p_track.add_argument("--date", help="Signal date (YYYY-MM-DD)")
+    p_track.add_argument("--signal", required=True, help="Signal type (B1/B2/B3/S1/S2/S3)")
+    p_track.add_argument("--price", type=float, required=True, help="Entry price")
+    p_track.add_argument("--grade", choices=["A", "B", "C"], help="Signal quality grade")
+    p_track.add_argument("--strength", type=float, help="Signal strength 0-100")
+
+    # backfill
+    p_backfill = sub.add_parser("backfill", help="Backfill historical scores")
+    p_backfill.add_argument("--from", dest="from_date", required=True)
+    p_backfill.add_argument("--to", dest="to_date")
+
+    # expectancy
+    p_expectancy = sub.add_parser("expectancy", help="Query signal expectancy")
+    p_expectancy.add_argument("--date", help="Target date (YYYY-MM-DD)")
+    p_expectancy.add_argument("--signal", default="B3", help="Signal type")
+
+    # validate
+    p_validate = sub.add_parser("validate", help="Run validation framework")
+
+    # cron
+    p_cron = sub.add_parser("cron", help="Run scheduled fetch+score loop")
+    # detect (Chan BSP signals)
+    p_detect = sub.add_parser("detect", help="Detect Chan BSP signals and populate signal_features")
+    p_detect.add_argument("--from", dest="from_date", default="2024-01-01")
+    p_detect.add_argument("--to", dest="to_date")
+    # serve
+    p_serve = sub.add_parser("serve", help="Start web dashboard")
+
+    args = parser.parse_args()
+
+    if args.command == "fetch":
+        cmd_fetch(args)
+    elif args.command == "score":
+        cmd_score(args)
+    elif args.command == "regime":
+        cmd_regime(args)
+    elif args.command == "track":
+        cmd_track(args)
+    elif args.command == "backfill":
+        cmd_backfill(args)
+    elif args.command == "expectancy":
+        cmd_expectancy(args)
+    elif args.command == "validate":
+        from validation.reporter import ValidationReporter
+        report = ValidationReporter().run_all()
+        print(report)
+    elif args.command == "detect":
+        from chan_integration import ChanSignalDetector
+        start = args.from_date
+        end = args.to_date or Date.today().isoformat()
+        detector = ChanSignalDetector()
+        count = detector.populate_signal_features(start, end)
+        logger.info(f"写入 {count} 条信号记录")
+    elif args.command == "serve":
+        from scheduler import get_scheduler
+        get_scheduler().start()
+        logger.info("启动 Web Dashboard: http://127.0.0.1:8124")
+        from web.app import app
+        app.run(host="0.0.0.0", port=8124, debug=False)
+    elif args.command == "cron":
+        from scheduler import get_scheduler
+        logger.info("启动后台调度器 (Ctrl+C 停止)")
+        s = get_scheduler()
+        s.start()
+        try:
+            while True:
+                time.sleep(60)
+        except KeyboardInterrupt:
+            s.stop()
+            logger.info("调度器已停止")
+    else:
+        parser.print_help()
+
+
+if __name__ == "__main__":
+    main()

ChanMacro/config.json

@@ -0,0 +1,23 @@
+{
+    "provider_url": "https://provider.jackyu66.com",
+    "db_path": "data/macro.db",
+    "btc_symbol": "BTC/USDT:USDT",
+    "regime_version": "v1_price_breadth_vol",
+    "half_life_days": 180,
+    "sufficiency_min_effective": 30,
+    "sufficiency_low": 50,
+    "sufficiency_medium": 100,
+    "level_min_samples": 50,
+    "knn_max_distance": 0.35,
+    "knn_k": 200,
+    "oi_price_threshold_pct": 0.5,
+    "oi_oi_threshold_pct": 0.5,
+    "vol_low_threshold": 2.0,
+    "vol_high_threshold": 5.0,
+    "vol_explosive_threshold": 10.0,
+    "regime_w_price": 0.35,
+    "regime_w_breadth": 0.50,
+    "regime_w_vol": 0.15,
+    "trend_w_price": 0.30,
+    "trend_w_breadth": 0.70
+}

ChanMacro/config.py

@@ -0,0 +1,113 @@
+"""
+config.py — Global configuration for ChanMacro.
+
+All weights, thresholds, and paths are configurable.
+V1 weights are deliberately simple; they will be tuned via Phase 0 validation.
+"""
+
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Optional
+
+
+@dataclass
+class Config:
+    """Global configuration. Override via config.json or env vars."""
+
+    # ── Paths ──────────────────────────────────────────────
+    db_path: str = "data/macro.db"
+    data_dir: str = "data"
+
+    # ── Data Provider ──────────────────────────────────────
+    provider_url: str = "https://provider.jackyu66.com"
+    btc_symbol: str = "BTC/USDT:USDT"
+    top50_symbols: list[str] = field(default_factory=lambda: [
+        "BTC/USDT:USDT", "ETH/USDT:USDT", "SOL/USDT:USDT",
+        "BNB/USDT:USDT", "XRP/USDT:USDT", "DOGE/USDT:USDT",
+        "SUI/USDT:USDT", "TON/USDT:USDT", "ZEC/USDT:USDT",
+        "1000PEPE/USDT:USDT", "SAGA/USDT:USDT",
+        "XAU/USDT:USDT", "XAG/USDT:USDT",
+        "CL/USDT:USDT", "BILL/USDT:USDT", "BZ/USDT:USDT",
+        "LAB/USDT:USDT", "CRCL/USDT:USDT", "SNDK/USDT:USDT",
+        "CHIP/USDT:USDT",
+    ])
+
+    # ── Breadth ────────────────────────────────────────────
+    breadth_top_n: list[int] = field(default_factory=lambda: [20, 30, 50])
+    breadth_ema_period: int = 20
+    breadth_new_high_window: int = 20
+
+    # ── Regime (factor-locked: Price + Breadth + Vol) ─────
+    regime_version: str = "v1_price_breadth_vol"
+    # Weights for trend_score within regime detection
+    regime_w_price: float = 0.35
+    regime_w_breadth: float = 0.50
+    regime_w_vol: float = 0.15
+    # Weights for panic_score
+    regime_panic_w_anti_trend: float = 0.60
+    regime_panic_w_vol_extreme: float = 0.40
+
+    # ── Price Structure ────────────────────────────────────
+    ps_ema_fast: int = 20
+    ps_ema_mid: int = 60
+    ps_ema_slow: int = 120
+    ps_adx_period: int = 14
+    ps_adx_threshold: int = 25
+    ps_atr_period: int = 14
+    ps_bb_period: int = 20
+    ps_roc_periods: list[int] = field(default_factory=lambda: [5, 10, 20])
+
+    # ── OI Matrix ──────────────────────────────────────────
+    oi_price_threshold_pct: float = 0.5   # min price change% to classify
+    oi_oi_threshold_pct: float = 0.5      # min OI change% to classify
+
+    # ── Volatility Regime ──────────────────────────────────
+    vol_atr_period: int = 14
+    vol_hv_short: int = 20
+    vol_hv_long: int = 60
+    # Thresholds (ATR/Close %)
+    vol_low_threshold: float = 2.0
+    vol_high_threshold: float = 5.0
+    vol_explosive_threshold: float = 10.0
+
+    # ── Trend (L2 aggregation) ─────────────────────────────
+    trend_w_price: float = 0.30
+    trend_w_breadth: float = 0.70
+
+    # ── Maturity Score ─────────────────────────────────────
+    maturity_w_trend: float = 0.50
+    maturity_w_breadth: float = 0.30
+    maturity_w_vol: float = 0.20
+
+    # ── Expectancy ─────────────────────────────────────────
+    half_life_days: int = 180
+    sufficiency_min_effective: int = 30
+    sufficiency_low: int = 50
+    sufficiency_medium: int = 100
+    level_min_samples: int = 50
+    knn_max_distance: float = 0.35
+    knn_k: int = 200
+
+    # ── Validation ─────────────────────────────────────────
+    min_history_days: int = 365
+    regime_min_avg_duration: int = 5
+    regime_max_flip_rate: float = 0.15
+
+    @classmethod
+    def from_json(cls, path: str = "config.json") -> "Config":
+        """Load config from JSON file, overriding defaults."""
+        import json
+        config = cls()
+        try:
+            with open(path) as f:
+                data = json.load(f)
+            for key, value in data.items():
+                if hasattr(config, key):
+                    setattr(config, key, value)
+        except FileNotFoundError:
+            pass
+        return config
+
+
+# Global singleton
+config = Config()

ChanMacro/database.py

@@ -0,0 +1,224 @@
+"""
+database.py — SQLite schema initialization and connection management.
+"""
+
+import sqlite3
+import os
+from pathlib import Path
+
+SCHEMA = """
+-- ═══════════════════════════════════════════════
+-- L0: Raw data tables
+-- ═══════════════════════════════════════════════
+
+CREATE TABLE IF NOT EXISTS ohlcv_daily (
+    date TEXT NOT NULL,
+    symbol TEXT NOT NULL DEFAULT 'BTC/USDT:USDT',
+    open REAL,
+    high REAL,
+    low REAL,
+    close REAL,
+    volume REAL,
+    ema20 REAL,
+    ema60 REAL,
+    ema120 REAL,
+    atr_14 REAL,
+    bb_width REAL,
+    adx_14 REAL,
+    PRIMARY KEY (date, symbol)
+);
+
+CREATE TABLE IF NOT EXISTS breadth_daily (
+    date TEXT PRIMARY KEY,
+    total_tracked INTEGER DEFAULT 50,
+    advance_top50 INTEGER DEFAULT 0,
+    decline_top50 INTEGER DEFAULT 0,
+    above_ema20_top50 INTEGER DEFAULT 0,
+    new_highs_20d_top50 INTEGER DEFAULT 0,
+    btc_dominance REAL,
+    advance_top20 INTEGER DEFAULT 0,
+    advance_top30 INTEGER DEFAULT 0,
+    above_ema20_top20 INTEGER DEFAULT 0,
+    above_ema20_top30 INTEGER DEFAULT 0,
+    new_highs_20d_top20 INTEGER DEFAULT 0,
+    new_highs_20d_top30 INTEGER DEFAULT 0,
+    fetched_at TEXT DEFAULT (datetime('now'))
+);
+
+CREATE TABLE IF NOT EXISTS derivatives (
+    date TEXT NOT NULL,
+    symbol TEXT NOT NULL DEFAULT 'BTC/USDT:USDT',
+    funding_rate REAL,
+    open_interest REAL,
+    oi_24h_change_pct REAL,
+    long_liquidations REAL,
+    short_liquidations REAL,
+    basis_annualised_pct REAL,
+    source TEXT DEFAULT 'binance',
+    fetched_at TEXT DEFAULT (datetime('now')),
+    PRIMARY KEY (date, symbol)
+);
+
+CREATE TABLE IF NOT EXISTS etf_flow (
+    date TEXT NOT NULL,
+    product TEXT NOT NULL,
+    net_flow_million REAL NOT NULL,
+    price REAL,
+    source TEXT DEFAULT 'farside',
+    fetched_at TEXT DEFAULT (datetime('now')),
+    PRIMARY KEY (date, product)
+);
+
+CREATE TABLE IF NOT EXISTS stablecoin_supply (
+    date TEXT NOT NULL,
+    token TEXT NOT NULL,
+    chain TEXT NOT NULL DEFAULT 'all',
+    supply REAL NOT NULL,
+    source TEXT DEFAULT 'defillama',
+    fetched_at TEXT DEFAULT (datetime('now')),
+    PRIMARY KEY (date, token, chain)
+);
+
+-- ═══════════════════════════════════════════════
+-- L3: Regime history
+-- ═══════════════════════════════════════════════
+
+CREATE TABLE IF NOT EXISTS regime_history (
+    date TEXT PRIMARY KEY,
+    regime TEXT NOT NULL,
+    confidence REAL,
+    regime_version TEXT NOT NULL DEFAULT 'v1_price_breadth_vol',
+    maturity_score REAL DEFAULT 50.0,
+    all_scores_json TEXT DEFAULT '{}',
+    prior_regime TEXT,
+    confirmation_days INTEGER DEFAULT 1,
+    created_at TEXT DEFAULT (datetime('now'))
+);
+
+-- ═══════════════════════════════════════════════
+-- ★ signal_features — THE moat
+-- ═══════════════════════════════════════════════
+
+CREATE TABLE IF NOT EXISTS signal_features (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    date TEXT NOT NULL,
+    signal_type TEXT NOT NULL,
+    signal_version TEXT NOT NULL DEFAULT 'b3_v1',
+    symbol TEXT DEFAULT 'BTC/USDT:USDT',
+
+    -- ★★ Version control (most important fields)
+    regime_version TEXT NOT NULL DEFAULT 'v1_price_breadth_vol',
+    signal_grade TEXT,
+    signal_strength REAL,
+
+    -- Market State Vector snapshot
+    regime TEXT NOT NULL,
+    regime_confidence REAL,
+    regime_maturity_score REAL DEFAULT 50.0,
+    market_state_hash TEXT,
+    state_embedding TEXT DEFAULT '[]',
+    breadth_top20 REAL,
+    breadth_top30 REAL,
+    breadth_top50 REAL,
+    breadth_bucket TEXT,
+    breadth_divergence REAL,
+    oi_state TEXT,
+    volatility_regime TEXT,
+    price_structure_score REAL,
+
+    -- Chan context (V5+)
+    chan_trend_direction TEXT,
+    chan_pivot_count INTEGER,
+    chan_divergence_type TEXT,
+
+    -- Outcomes
+    entry_price REAL,
+    result_1d REAL,
+    result_3d REAL,
+    result_5d REAL,
+    result_7d REAL,
+    result_14d REAL,
+    max_favorable_excursion REAL,
+    max_adverse_excursion REAL,
+    is_win_7d INTEGER,
+
+    created_at TEXT DEFAULT (datetime('now'))
+);
+
+CREATE INDEX IF NOT EXISTS idx_sf_regime ON signal_features(regime);
+CREATE INDEX IF NOT EXISTS idx_sf_signal ON signal_features(signal_type);
+CREATE INDEX IF NOT EXISTS idx_sf_oi_state ON signal_features(oi_state);
+CREATE INDEX IF NOT EXISTS idx_sf_date ON signal_features(date);
+CREATE INDEX IF NOT EXISTS idx_sf_state_hash ON signal_features(market_state_hash);
+CREATE INDEX IF NOT EXISTS idx_sf_regime_version ON signal_features(regime_version);
+CREATE INDEX IF NOT EXISTS idx_sf_signal_version ON signal_features(signal_version);
+
+-- ═══════════════════════════════════════════════
+-- Expectancy cache (raw counts, NOT posteriors)
+-- ═══════════════════════════════════════════════
+
+CREATE TABLE IF NOT EXISTS expectancy_cache (
+    state_hash TEXT NOT NULL,
+    signal_type TEXT NOT NULL,
+    wins_weighted REAL DEFAULT 0,
+    losses_weighted REAL DEFAULT 0,
+    sum_return_7d REAL DEFAULT 0,
+    sum_return_sq_7d REAL DEFAULT 0,
+    effective_samples REAL DEFAULT 0,
+    sufficiency TEXT DEFAULT 'INSUFFICIENT',
+    updated_at TEXT DEFAULT (datetime('now')),
+    PRIMARY KEY (state_hash, signal_type)
+);
+
+-- ═══════════════════════════════════════════════
+-- Similarity outcome (KNN weight learning, Phase D)
+-- ═══════════════════════════════════════════════
+
+CREATE TABLE IF NOT EXISTS similarity_outcome (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    state_a_hash TEXT,
+    state_b_hash TEXT,
+    distance REAL,
+    actual_return_gap REAL,
+    dimension_weights_json TEXT DEFAULT '{}',
+    created_at TEXT DEFAULT (datetime('now'))
+);
+
+-- ═══════════════════════════════════════════════
+-- chan_context — Chan theory integration (V1 empty)
+-- ═══════════════════════════════════════════════
+
+CREATE TABLE IF NOT EXISTS chan_context (
+    date TEXT NOT NULL,
+    timeframe TEXT NOT NULL DEFAULT '1d',
+    trend_direction TEXT,
+    trend_strength REAL,
+    pivot_count INTEGER,
+    pivot_level TEXT,
+    signal_type TEXT,
+    signal_strength REAL,
+    divergence_type TEXT,
+    chan_structure_score REAL,
+    alignment_score REAL,
+    raw_context_json TEXT DEFAULT '{}',
+    PRIMARY KEY (date, timeframe)
+);
+"""
+
+
+def init_db(db_path: str = "data/macro.db") -> sqlite3.Connection:
+    """Initialize database: create directory and all tables."""
+    Path(db_path).parent.mkdir(parents=True, exist_ok=True)
+    conn = sqlite3.connect(db_path)
+    conn.executescript(SCHEMA)
+    conn.commit()
+    return conn
+
+
+def get_connection(db_path: str = "data/macro.db") -> sqlite3.Connection:
+    """Get a database connection. Creates tables if first run."""
+    if not os.path.exists(db_path):
+        return init_db(db_path)
+    conn = sqlite3.connect(db_path)
+    conn.row_factory = sqlite3.Row
+    return conn

ChanMacro/expectancy/__init__.py

@@ -0,0 +1,4 @@
+"""Expectancy Engine — Signal tracking, Bayesian inference, time decay."""
+from .tracker import SignalTracker
+from .decay import TimeDecay
+from .engine import BayesianExpectancyEngine, SufficiencyGuard

ChanMacro/expectancy/decay.py

@@ -0,0 +1,55 @@
+"""
+expectancy/decay.py — Time-weighted sample decay.
+
+2024 market structure ≠ 2026 market structure.
+Recent samples get higher weight via exponential decay.
+"""
+
+from datetime import date as Date
+from typing import Optional
+import numpy as np
+
+
+class TimeDecay:
+    """Exponential time decay for sample weighting."""
+
+    def __init__(self, half_life_days: int = 180):
+        self.half_life = half_life_days
+        self._decay_rate = np.log(2) / half_life_days
+
+    def weight(self, sample_date: Date, reference_date: Optional[Date] = None) -> float:
+        """
+        Compute decay weight for a sample.
+        weight = exp(-days_ago * decay_rate)
+        """
+        if reference_date is None:
+            reference_date = Date.today()
+        days = (reference_date - sample_date).days
+        return np.exp(-days * self._decay_rate)
+
+    def weights(self, dates: list[Date], reference_date: Optional[Date] = None) -> np.ndarray:
+        """Compute decay weights for a list of dates."""
+        return np.array([self.weight(d, reference_date) for d in dates])
+
+    def weighted_win_rate(self, wins: np.ndarray, weights: np.ndarray) -> float:
+        """Weighted win rate: sum(wins * weights) / sum(weights)."""
+        total_weight = weights.sum()
+        if total_weight == 0:
+            return 0.0
+        return float((wins * weights).sum() / total_weight)
+
+    def weighted_mean(self, values: np.ndarray, weights: np.ndarray) -> float:
+        """Weighted mean."""
+        total_weight = weights.sum()
+        if total_weight == 0:
+            return 0.0
+        return float((values * weights).sum() / total_weight)
+
+    def effective_samples(self, weights: np.ndarray) -> float:
+        """Effective number of samples after decay weighting."""
+        return float(weights.sum())
+
+    @staticmethod
+    def weight_at_age(days_ago: int, half_life_days: int = 180) -> float:
+        """Quick weight lookup for a given age in days."""
+        return np.exp(-days_ago * np.log(2) / half_life_days)

ChanMacro/expectancy/engine.py

@@ -0,0 +1,295 @@
+"""
+expectancy/engine.py — Bayesian Expectancy Engine.
+
+Core algorithm:
+  1. LeveledExpectancy: filter layer-by-layer, stop at highest valid level
+  2. Empirical Bayes prior: prior = signal's global historical winrate
+  3. Dynamic Beta strength: adaptive to sample size
+  4. Time decay: recent samples weighted higher (half_life=180d)
+  5. SufficiencyGuard: refuse output if effective_samples < 30
+  6. KNN Fallback: similarity search when strict filtering fails (Phase D)
+"""
+
+from datetime import date as Date
+from typing import Optional
+import sqlite3
+import logging
+
+import numpy as np
+import pandas as pd
+
+from models import (
+    MarketStateVector, ExpectancyReport, ExpectancyLayer,
+    SufficiencyLevel, MarketRegime,
+)
+from config import config
+from .decay import TimeDecay
+
+logger = logging.getLogger(__name__)
+
+
+class SufficiencyGuard:
+    """Prevents trading advice from insufficient samples."""
+
+    def __init__(self, min_effective: int = 30, low: int = 50, medium: int = 100):
+        self.MIN = min_effective
+        self.LOW = low
+        self.MEDIUM = medium
+
+    def evaluate(self, effective_samples: float) -> SufficiencyLevel:
+        if effective_samples < self.MIN:
+            return SufficiencyLevel.INSUFFICIENT
+        elif effective_samples < self.LOW:
+            return SufficiencyLevel.LOW
+        elif effective_samples < self.MEDIUM:
+            return SufficiencyLevel.MEDIUM
+        return SufficiencyLevel.HIGH
+
+
+class BayesianExpectancyEngine:
+    """
+    Leveled Bayesian Expectancy Engine.
+
+    Query layers from coarse to fine. Stop when effective_samples drops below threshold.
+    Uses Empirical Bayes prior (signal's global winrate, not fixed 50%).
+    """
+
+    # Expectancy query levels: name → WHERE clause template
+    LEVELS = [
+        ("Base",            "signal_type = '{signal}'"),
+        ("+ Regime",        "signal_type = '{signal}' AND regime = '{regime}'"),
+        ("+ Breadth",       "signal_type = '{signal}' AND regime = '{regime}' AND breadth_bucket = '{breadth}'"),
+        ("+ OI State",      "signal_type = '{signal}' AND regime = '{regime}' AND breadth_bucket = '{breadth}' AND oi_state = '{oi}'"),
+        ("+ Volatility",    "signal_type = '{signal}' AND regime = '{regime}' AND breadth_bucket = '{breadth}' AND oi_state = '{oi}' AND volatility_regime = '{vol}'"),
+    ]
+
+    def __init__(self, db_path: Optional[str] = None,
+                 half_life_days: int = 180,
+                 level_min_samples: int = 50):
+        self.db_path = db_path or config.db_path
+        self.decay = TimeDecay(half_life_days)
+        self.guard = SufficiencyGuard(
+            min_effective=config.sufficiency_min_effective,
+            low=config.sufficiency_low,
+            medium=config.sufficiency_medium,
+        )
+        self.level_min = level_min_samples
+
+    def estimate(self, state: MarketStateVector,
+                 signal_type: str = "B3",
+                 target_date: Optional[Date] = None) -> ExpectancyReport:
+        """
+        Compute layered Bayesian expectancy for a signal in current market state.
+
+        Returns the estimate at the deepest level with >= level_min effective samples.
+        """
+        if target_date is None:
+            target_date = Date.today()
+
+        conn = sqlite3.connect(self.db_path)
+
+        # Get global signal winrate for Empirical Bayes prior
+        global_rate = self._global_winrate(conn, signal_type)
+
+        layers = []
+        best_result = None
+
+        for level_name, template in self.LEVELS:
+            where = template.format(
+                signal=signal_type,
+                regime=state.regime.value,
+                breadth=state.breadth_bucket.value,
+                oi=state.oi_state.value,
+                vol=state.volatility_regime.value,
+            )
+            query = f"SELECT * FROM signal_features WHERE {where}"
+            df = pd.read_sql_query(query, conn)
+
+            if df.empty:
+                layers.append(ExpectancyLayer(
+                    name=level_name, posterior_winrate=0.0,
+                    samples=0, effective_samples=0.0,
+                ))
+                continue
+
+            # Time-weighted stats
+            dates_list = [Date.fromisoformat(d) for d in df["date"]]
+            weights = self.decay.weights(dates_list, target_date)
+            eff_n = self.decay.effective_samples(weights)
+
+            wins = pd.to_numeric(df["is_win_7d"].fillna(0), errors="coerce").fillna(0).values
+            returns = pd.to_numeric(df["result_7d"].fillna(0), errors="coerce").fillna(0).values
+
+            raw_wr = float(wins.mean()) if len(wins) > 0 else 0.0
+            weighted_wr = self.decay.weighted_win_rate(wins, weights)
+            weighted_ret = self.decay.weighted_mean(returns, weights)
+
+            # Empirical Bayes posterior
+            posterior = self._bayesian_posterior(
+                global_rate=global_rate,
+                wins=wins.sum(),
+                samples=len(df),
+            )
+
+            layer = ExpectancyLayer(
+                name=level_name,
+                posterior_winrate=round(posterior, 4),
+                raw_winrate=round(raw_wr, 4),
+                samples=len(df),
+                effective_samples=round(eff_n, 1),
+                avg_return=round(weighted_ret, 2),
+            )
+            layers.append(layer)
+
+            # Level-based fallback: keep going while samples sufficient
+            if eff_n >= self.level_min:
+                best_result = layer
+
+        conn.close()
+
+        if best_result is None and layers:
+            # Fallback to the deepest layer that had any samples
+            for layer in reversed(layers):
+                if layer.samples > 0:
+                    best_result = layer
+                    break
+
+        if best_result is None:
+            return ExpectancyReport(
+                signal_type=signal_type,
+                date=target_date,
+                layers=layers,
+                final_estimate=0.0,
+                sufficiency=SufficiencyLevel.INSUFFICIENT,
+                source="insufficient",
+            )
+
+        sufficiency = self.guard.evaluate(
+            best_result.effective_samples
+        )
+
+        # Compute profit factor and MAE from the SAME level as best_result
+        profit_factor = None
+        avg_mae = None
+        if best_result and best_result.samples > 0:
+            # Re-query the level that produced best_result
+            best_level_idx = next(
+                i for i, l in enumerate(layers) if l.name == best_result.name
+            )
+            where = self.LEVELS[best_level_idx][1].format(
+                signal=signal_type, regime=state.regime.value,
+                breadth=state.breadth_bucket.value, oi=state.oi_state.value,
+                vol=state.volatility_regime.value,
+            )
+            query = f"SELECT result_7d, max_adverse_excursion FROM signal_features WHERE {where}"
+            conn2 = sqlite3.connect(self.db_path)
+            df_detail = pd.read_sql_query(query, conn2)
+            conn2.close()
+            if not df_detail.empty:
+                returns_7d = df_detail["result_7d"].dropna()
+                if len(returns_7d) > 0:
+                    gains = returns_7d[returns_7d > 0].sum()
+                    losses = abs(returns_7d[returns_7d < 0].sum())
+                    profit_factor = round(gains / losses, 2) if losses > 0 else None
+                maes = df_detail["max_adverse_excursion"].dropna()
+                if len(maes) > 0:
+                    avg_mae = round(float(maes.mean()), 2)
+
+        return ExpectancyReport(
+            signal_type=signal_type,
+            date=target_date,
+            layers=layers,
+            final_estimate=round(best_result.posterior_winrate, 4),
+            sufficiency=sufficiency,
+            prior_strength=self._prior_strength(best_result.samples),
+            half_life_days=self.decay.half_life,
+            source="bayesian",
+            avg_return_7d=best_result.avg_return,
+            profit_factor=profit_factor,
+            max_adverse_excursion=avg_mae,
+        )
+
+    def _global_winrate(self, conn: sqlite3.Connection,
+                        signal_type: str) -> float:
+        """Get global historical winrate for a signal type (Empirical Bayes prior)."""
+        row = conn.execute(
+            "SELECT AVG(is_win_7d) as wr, COUNT(*) as cnt "
+            "FROM signal_features WHERE signal_type = ? AND is_win_7d IS NOT NULL",
+            (signal_type,)
+        ).fetchone()
+        if row and row[1] and row[1] > 0:
+            return float(row[0])
+        return 0.50  # default: neutral
+
+    def _prior_strength(self, samples: int) -> int:
+        """Dynamic prior strength based on sample count."""
+        if samples < 100:
+            return 20  # Beta(10,10)
+        elif samples < 500:
+            return 40  # Beta(20,20)
+        else:
+            return 100  # Beta(50,50) — data dominates
+
+    def _bayesian_posterior(self, global_rate: float, wins: float,
+                            samples: int) -> float:
+        """
+        Empirical Bayes posterior: prior = global signal winrate.
+
+        posterior = (alpha + wins) / (alpha + beta + samples)
+        where alpha/(alpha+beta) = global_rate
+        """
+        prior_strength = self._prior_strength(samples)
+        alpha = max(global_rate * prior_strength, 1.0)  # floor at 1 to ensure shrinkage
+        beta = max((1 - global_rate) * prior_strength, 1.0)
+        return (alpha + wins) / (alpha + beta + samples)
+
+    def precompute_cache(self):
+        """
+        Precompute expectancy for all state_hashes in signal_features.
+        Populates expectancy_cache table with raw weighted counts (not posteriors).
+        """
+        conn = sqlite3.connect(self.db_path)
+        conn.row_factory = sqlite3.Row
+
+        hashes = conn.execute(
+            "SELECT DISTINCT market_state_hash, signal_type FROM signal_features"
+        ).fetchall()
+
+        today = Date.today()
+        count = 0
+
+        for row in hashes:
+            h = row["market_state_hash"]
+            sig = row["signal_type"]
+
+            df = pd.read_sql_query(
+                "SELECT date, is_win_7d, result_7d "
+                "FROM signal_features WHERE market_state_hash = ? AND signal_type = ?",
+                conn, params=(h, sig)
+            )
+
+            if df.empty:
+                continue
+
+            dates_list = [Date.fromisoformat(d) for d in df["date"]]
+            weights = self.decay.weights(dates_list, today)
+            wins_w = (df["is_win_7d"].fillna(0).values * weights).sum()
+            losses_w = ((1 - df["is_win_7d"].fillna(0)).values * weights).sum()
+            ret_sum = (df["result_7d"].fillna(0).values * weights).sum()
+            ret_sq = ((df["result_7d"].fillna(0).values ** 2) * weights).sum()
+            eff_n = weights.sum()
+
+            sufficiency = self.guard.evaluate(eff_n).value
+
+            conn.execute("""
+                INSERT OR REPLACE INTO expectancy_cache
+                (state_hash, signal_type, wins_weighted, losses_weighted,
+                 sum_return_7d, sum_return_sq_7d, effective_samples, sufficiency)
+                VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+            """, (h, sig, wins_w, losses_w, ret_sum, ret_sq, eff_n, sufficiency))
+            count += 1
+
+        conn.commit()
+        conn.close()
+        logger.info(f"Precomputed expectancy cache: {count} state×signal combos")
+        return count

ChanMacro/expectancy/tracker.py

@@ -0,0 +1,271 @@
+"""
+expectancy/tracker.py — SignalTracker: records signals with full market state
+and computes forward outcomes.
+
+This is the entry point for populating signal_features — THE moat table.
+"""
+
+from datetime import date as Date, timedelta
+from typing import Optional
+import sqlite3
+import json
+import logging
+
+import pandas as pd
+import numpy as np
+
+from models import (
+    MarketStateVector, SignalFeatureRecord, MarketRegime,
+    OIState, BreadthBucket, VolRegime, SignalGrade,
+)
+from config import config
+
+logger = logging.getLogger(__name__)
+
+
+class SignalTracker:
+    """
+    Records trading signals with full market state context.
+
+    Usage:
+        tracker = SignalTracker()
+        tracker.record(
+            date=Date(2026, 6, 24),
+            signal_type="B3",
+            entry_price=96500.0,
+            state=market_state_vector,  # from scoring pipeline
+            signal_grade="A",
+        )
+    """
+
+    def __init__(self, db_path: Optional[str] = None):
+        self.db_path = db_path or config.db_path
+
+    def record(self, date: Date, signal_type: str, entry_price: float,
+               state: MarketStateVector,
+               signal_version: str = "b3_v1",
+               signal_grade: Optional[str] = None,
+               signal_strength: Optional[float] = None) -> int:
+        """
+        Record a signal with market state snapshot and compute forward outcomes.
+
+        Returns the record ID in signal_features.
+        """
+        conn = sqlite3.connect(self.db_path)
+
+        # Compute forward outcomes
+        outcomes = self._compute_outcomes(conn, date, entry_price)
+
+        # Build embedding
+        embedding = json.dumps(state.state_embedding())
+
+        record_id = conn.execute("""
+            INSERT INTO signal_features
+            (date, signal_type, signal_version, symbol,
+             regime_version, signal_grade, signal_strength,
+             regime, regime_confidence, regime_maturity_score,
+             market_state_hash, state_embedding,
+             breadth_top20, breadth_top30, breadth_top50,
+             breadth_bucket, breadth_divergence,
+             oi_state, volatility_regime, price_structure_score,
+             entry_price,
+             result_1d, result_3d, result_5d, result_7d, result_14d,
+             max_favorable_excursion, max_adverse_excursion,
+             is_win_7d)
+            VALUES (?, ?, ?, ?, ?, ?, ?,
+                    ?, ?, ?,
+                    ?, ?,
+                    ?, ?, ?,
+                    ?, ?,
+                    ?, ?, ?,
+                    ?,
+                    ?, ?, ?, ?, ?,
+                    ?, ?,
+                    ?)
+        """, (
+            str(date), signal_type, signal_version, state.symbol,
+            state.regime_version, signal_grade, signal_strength,
+            state.regime.value, state.regime_confidence, state.regime_maturity_score,
+            state.market_state_hash, embedding,
+            state.breadth_top20, state.breadth_top30, state.breadth_top50,
+            state.breadth_bucket.value, state.breadth_divergence,
+            state.oi_state.value, state.volatility_regime.value,
+            state.price_structure_score.score,
+            entry_price,
+            outcomes.get("result_1d"), outcomes.get("result_3d"),
+            outcomes.get("result_5d"), outcomes.get("result_7d"),
+            outcomes.get("result_14d"),
+            outcomes.get("mfe"), outcomes.get("mae"),
+            outcomes.get("is_win_7d"),
+        )).lastrowid
+
+        conn.commit()
+        conn.close()
+
+        is_win = outcomes.get("is_win_7d", 0)
+        ret_7d = outcomes.get("result_7d", 0) or 0
+        logger.info(
+            f"Recorded {signal_type} on {date} @ {entry_price:.0f} "
+            f"(regime={state.regime.value}, breadth={state.breadth_bucket.value}, "
+            f"oi={state.oi_state.value}) → 7d={ret_7d:+.1f}%"
+        )
+        return record_id
+
+    def _compute_outcomes(self, conn: sqlite3.Connection, date: Date,
+                          entry_price: float) -> dict:
+        """
+        Compute forward returns, MFE, MAE from OHLCV data.
+
+        Queries future daily bars relative to the signal date.
+        """
+        # Get future OHLCV data
+        df = pd.read_sql_query(
+            "SELECT date, high, low, close FROM ohlcv_daily "
+            "WHERE date > ? AND symbol = 'BTC/USDT:USDT' "
+            "ORDER BY date ASC LIMIT 20",
+            conn, params=(str(date),)
+        )
+
+        if df.empty:
+            return {}
+
+        outcomes = {}
+        entry = entry_price
+
+        # Forward returns
+        for horizon_days, col in [(1, "result_1d"), (3, "result_3d"),
+                                   (5, "result_5d"), (7, "result_7d"),
+                                   (14, "result_14d")]:
+            if len(df) >= horizon_days:
+                exit_price = float(df.iloc[horizon_days - 1]["close"])
+                outcomes[col] = round((exit_price - entry) / entry * 100, 2)
+
+        # MFE / MAE
+        if len(df) > 0:
+            highs = df["high"].astype(float).values[:14]
+            lows = df["low"].astype(float).values[:14]
+            outcomes["mfe"] = round((max(highs) - entry) / entry * 100, 2)
+            outcomes["mae"] = round((min(lows) - entry) / entry * 100, 2)
+
+        # is_win_7d
+        outcomes["is_win_7d"] = 1 if outcomes.get("result_7d", 0) > 0 else 0
+
+        return outcomes
+
+    def backfill_signals(self, signals: list[dict]) -> int:
+        """
+        Backfill multiple signals from historical data.
+
+        Each signal dict:
+            {"date": Date, "signal_type": str, "entry_price": float,
+             "signal_grade": str (optional), "signal_strength": float (optional)}
+
+        This requires the scoring pipeline to have been run for those dates
+        (breadth_daily, ohlcv_daily, derivatives all populated).
+        """
+        from scoring.price_structure import PriceStructureScorer
+        from scoring.breadth_scorer import BreadthScorer
+        from scoring.oi_matrix import OIMatrixScorer
+        from scoring.volatility_regime import VolatilityRegimeScorer
+        from regime_detector import RegimeDetector
+
+        detector = RegimeDetector()
+        count = 0
+
+        for sig in signals:
+            target = sig["date"]
+            try:
+                # Compute market state for this date
+                ps = PriceStructureScorer(self.db_path).compute(target)
+                br = BreadthScorer(self.db_path).compute(target)
+                oi = OIMatrixScorer(self.db_path).compute(target)
+                vol = VolatilityRegimeScorer(self.db_path).compute(target)
+
+                regime_result = detector.detect(
+                    price_structure_score=ps.score,
+                    breadth_score=br.breadth_top50,
+                    volatility_regime=vol.vol_regime.value,
+                    date=target,
+                )
+
+                state = MarketStateVector(
+                    date=target,
+                    regime=regime_result.regime,
+                    regime_confidence=regime_result.confidence,
+                    regime_version=regime_result.regime_version,
+                    regime_maturity_score=regime_result.maturity_score,
+                    breadth_top20=br.breadth_top20,
+                    breadth_top30=br.breadth_top30,
+                    breadth_top50=br.breadth_top50,
+                    breadth_bucket=br.breadth_bucket,
+                    breadth_divergence=br.breadth_divergence,
+                    oi_state=oi.oi_state,
+                    volatility_regime=vol.vol_regime,
+                    price_structure_score=ps,
+                    breadth_score=br,
+                    oi_matrix_score=oi,
+                    volatility_regime_score=vol,
+                )
+                state.market_state_hash = state.compute_hash()
+
+                self.record(
+                    date=target,
+                    signal_type=sig["signal_type"],
+                    entry_price=sig["entry_price"],
+                    state=state,
+                    signal_grade=sig.get("signal_grade"),
+                    signal_strength=sig.get("signal_strength"),
+                )
+                count += 1
+            except Exception as e:
+                logger.warning(f"Failed to backfill {sig['signal_type']} on {target}: {e}")
+
+        return count
+
+    def get_samples(self, signal_type: Optional[str] = None,
+                    regime: Optional[str] = None,
+                    breadth_bucket: Optional[str] = None,
+                    oi_state: Optional[str] = None,
+                    volatility_regime: Optional[str] = None,
+                    limit: int = 5000) -> list[dict]:
+        """Query signal_features with optional filters."""
+        conn = sqlite3.connect(self.db_path)
+        conn.row_factory = sqlite3.Row
+
+        query = "SELECT * FROM signal_features WHERE 1=1"
+        params = []
+
+        if signal_type:
+            query += " AND signal_type = ?"
+            params.append(signal_type)
+        if regime:
+            query += " AND regime = ?"
+            params.append(regime)
+        if breadth_bucket:
+            query += " AND breadth_bucket = ?"
+            params.append(breadth_bucket)
+        if oi_state:
+            query += " AND oi_state = ?"
+            params.append(oi_state)
+        if volatility_regime:
+            query += " AND volatility_regime = ?"
+            params.append(volatility_regime)
+
+        query += " ORDER BY date DESC LIMIT ?"
+        params.append(limit)
+
+        rows = conn.execute(query, params).fetchall()
+        conn.close()
+        return [dict(r) for r in rows]
+
+    def count_samples(self) -> dict:
+        """Count signal_features by signal_type and regime."""
+        conn = sqlite3.connect(self.db_path)
+        rows = conn.execute("""
+            SELECT signal_type, regime, COUNT(*) as cnt
+            FROM signal_features
+            GROUP BY signal_type, regime
+            ORDER BY signal_type, regime
+        """).fetchall()
+        conn.close()
+        return {f"{r[0]}/{r[1]}": r[2] for r in rows}

ChanMacro/fetchers/__init__.py

@@ -0,0 +1,5 @@
+"""Data fetchers — L0 raw data acquisition."""
+from .base import BaseFetcher
+from .ohlcv import OHLCVFetcher
+from .breadth import BreadthFetcher
+from .derivatives import DerivativesFetcher

ChanMacro/fetchers/base.py

@@ -0,0 +1,69 @@
+"""
+fetchers/base.py — Abstract base class for all macro data fetchers.
+
+Provides retry logic, rate limiting, and a common interface.
+"""
+
+from abc import ABC, abstractmethod
+from datetime import date as Date
+from typing import Optional
+import logging
+import time
+import requests
+
+
+class BaseFetcher(ABC):
+    """Abstract base for all macro data fetchers."""
+
+    def __init__(self, timeout: int = 30, max_retries: int = 3):
+        self.timeout = timeout
+        self.max_retries = max_retries
+        self.logger = logging.getLogger(self.__class__.__name__)
+
+    def _get(self, url: str, params: Optional[dict] = None,
+             headers: Optional[dict] = None) -> dict:
+        """GET with retry and exponential backoff."""
+        for attempt in range(self.max_retries):
+            try:
+                resp = requests.get(
+                    url, params=params, headers=headers, timeout=self.timeout
+                )
+                resp.raise_for_status()
+                return resp.json()
+            except requests.RequestException as e:
+                wait = 2 ** attempt
+                self.logger.warning(
+                    f"Request failed (attempt {attempt+1}/{self.max_retries}): {e}. "
+                    f"Retrying in {wait}s"
+                )
+                if attempt < self.max_retries - 1:
+                    time.sleep(wait)
+                else:
+                    raise
+
+    def _get_raw(self, url: str, params: Optional[dict] = None,
+                 headers: Optional[dict] = None) -> bytes:
+        """GET raw bytes with retry (for non-JSON endpoints)."""
+        for attempt in range(self.max_retries):
+            try:
+                resp = requests.get(
+                    url, params=params, headers=headers, timeout=self.timeout
+                )
+                resp.raise_for_status()
+                return resp.content
+            except requests.RequestException as e:
+                wait = 2 ** attempt
+                if attempt < self.max_retries - 1:
+                    time.sleep(wait)
+                else:
+                    raise
+
+    @abstractmethod
+    def fetch(self, target_date: Optional[Date] = None) -> list[dict]:
+        """Fetch raw data. Returns list of record dicts."""
+        ...
+
+    @abstractmethod
+    def store(self, db_path: str, records: list[dict]) -> int:
+        """Store raw records into SQLite. Returns count of new rows."""
+        ...

ChanMacro/fetchers/breadth.py

@@ -0,0 +1,189 @@
+"""
+fetchers/breadth.py — Fetches TOP50 OHLCV and computes market breadth metrics.
+
+Multi-tier: Top20 / Top30 / Top50 for advance/decline, EMA20%, new highs, BTC.D.
+"""
+
+from datetime import date as Date, datetime
+from typing import Optional
+import logging
+
+import pandas as pd
+import numpy as np
+import requests
+
+from .base import BaseFetcher
+from config import config
+
+
+class BreadthFetcher(BaseFetcher):
+    """Fetches TOP50 coin OHLCV data and computes breadth metrics."""
+
+    def __init__(self, provider_url: Optional[str] = None):
+        super().__init__(timeout=60, max_retries=3)
+        self.provider_url = provider_url or config.provider_url
+        self.symbols = config.top50_symbols
+        self.ema_period = config.breadth_ema_period
+        self.new_high_window = config.breadth_new_high_window
+        self.logger = logging.getLogger(__name__)
+
+    def fetch(self, target_date: Optional[Date] = None) -> dict:
+        """
+        Fetch daily OHLCV for all TOP50 symbols and compute breadth.
+
+        Returns a dict suitable for storing in breadth_daily table.
+        """
+        if target_date is None:
+            target_date = Date.today()
+
+        # Fetch last 60 days of daily data for each symbol to compute EMAs and new highs
+        all_data = {}
+        for symbol in self.symbols:
+            try:
+                df = self._fetch_symbol(symbol)
+                if df is not None and not df.empty:
+                    all_data[symbol] = df
+            except Exception as e:
+                self.logger.debug(f"Failed to fetch {symbol}: {e}")
+
+        if not all_data:
+            self.logger.error("No symbol data fetched for breadth")
+            return {}
+
+        # Compute breadth metrics for the target date
+        breadth = self._compute_breadth(all_data, target_date)
+        return breadth
+
+    def _fetch_symbol(self, symbol: str) -> Optional[pd.DataFrame]:
+        """Fetch daily OHLCV for a single symbol."""
+        url = f"{self.provider_url}/api/candles"
+        params = {
+            "symbol": symbol,
+            "tf": "1d",
+            "limit": 100,
+        }
+        try:
+            resp = requests.get(url, params=params, timeout=15)
+            resp.raise_for_status()
+            data = resp.json()
+            if not data:
+                return None
+
+            df = pd.DataFrame(data)
+            df["timestamp"] = pd.to_datetime(df["timestamp"], unit="ms", utc=True)
+            df["date"] = df["timestamp"].dt.date
+            df = df.drop_duplicates(subset="date").sort_values("date").reset_index(drop=True)
+            df["close"] = df["close"].astype(float)
+            df["ema20"] = df["close"].ewm(span=self.ema_period, adjust=False).mean()
+            return df
+        except Exception:
+            return None
+
+    def _compute_breadth(self, all_data: dict, target_date: Date) -> dict:
+        """Compute breadth metrics for a specific date across all symbols."""
+        total = len(all_data)
+
+        advances_50 = declines_50 = 0
+        above_ema20_50 = 0
+        new_highs_50 = 0
+        advances_30 = declines_30 = 0
+        above_ema20_30 = 0
+        new_highs_30 = 0
+        advances_20 = declines_20 = 0
+        above_ema20_20 = 0
+        new_highs_20 = 0
+
+        for i, (symbol, df) in enumerate(all_data.items()):
+            # Get data for target date
+            df["date_str"] = df["date"].astype(str)
+            target_str = str(target_date)
+            idx = df[df["date_str"] == target_str].index
+
+            if len(idx) == 0:
+                continue
+
+            row_idx = idx[0]
+            if row_idx < 1:
+                continue
+
+            current_close = df.loc[row_idx, "close"]
+            prev_close = df.loc[row_idx - 1, "close"]
+
+            # Advance/Decline
+            if current_close > prev_close:
+                if i < 50: advances_50 += 1
+                if i < 30: advances_30 += 1
+                if i < 20: advances_20 += 1
+            elif current_close < prev_close:
+                if i < 50: declines_50 += 1
+                if i < 30: declines_30 += 1
+                if i < 20: declines_20 += 1
+
+            # Above EMA20
+            ema20_val = df.loc[row_idx, "ema20"]
+            if not pd.isna(ema20_val) and current_close > ema20_val:
+                if i < 50: above_ema20_50 += 1
+                if i < 30: above_ema20_30 += 1
+                if i < 20: above_ema20_20 += 1
+
+            # New 20-day highs
+            lookback_start = max(0, row_idx - self.new_high_window)
+            recent_highs = df.loc[lookback_start:row_idx - 1, "high"].astype(float)
+            current_high = df.loc[row_idx, "high"]
+            if len(recent_highs) > 0 and float(current_high) > recent_highs.max():
+                if i < 50: new_highs_50 += 1
+                if i < 30: new_highs_30 += 1
+                if i < 20: new_highs_20 += 1
+
+        return {
+            "date": str(target_date),
+            "total_tracked": total,
+            "advance_top50": advances_50,
+            "decline_top50": declines_50,
+            "above_ema20_top50": above_ema20_50,
+            "new_highs_20d_top50": new_highs_50,
+            "advance_top30": advances_30,
+            "advance_top20": advances_20,
+            "above_ema20_top30": above_ema20_30,
+            "above_ema20_top20": above_ema20_20,
+            "new_highs_20d_top30": new_highs_30,
+            "new_highs_20d_top20": new_highs_20,
+            "btc_dominance": None,  # Reserved for Coinglass API integration
+        }
+
+    def store(self, db_path: Optional[str] = None, record: Optional[dict] = None) -> int:
+        """Store a breadth record into SQLite. Returns 1 if inserted/updated."""
+        import sqlite3
+        db_path = db_path or config.db_path
+        conn = sqlite3.connect(db_path)
+
+        if record is None:
+            conn.close()
+            return 0
+
+        try:
+            conn.execute("""
+                INSERT OR REPLACE INTO breadth_daily
+                (date, total_tracked,
+                 advance_top50, decline_top50, above_ema20_top50, new_highs_20d_top50,
+                 advance_top30, advance_top20,
+                 above_ema20_top30, above_ema20_top20,
+                 new_highs_20d_top30, new_highs_20d_top20,
+                 btc_dominance)
+                VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+            """, (
+                record["date"], record.get("total_tracked", 50),
+                record.get("advance_top50", 0), record.get("decline_top50", 0),
+                record.get("above_ema20_top50", 0), record.get("new_highs_20d_top50", 0),
+                record.get("advance_top30", 0), record.get("advance_top20", 0),
+                record.get("above_ema20_top30", 0), record.get("above_ema20_top20", 0),
+                record.get("new_highs_20d_top30", 0), record.get("new_highs_20d_top20", 0),
+                record.get("btc_dominance"),
+            ))
+            conn.commit()
+            return 1
+        except Exception as e:
+            self.logger.error(f"Failed to store breadth: {e}")
+            return 0
+        finally:
+            conn.close()

ChanMacro/fetchers/derivatives.py

@@ -0,0 +1,66 @@
+"""
+fetchers/derivatives.py — Fetches derivatives data from data_provider API.
+
+Clean consumer: no direct ccxt dependency. Just HTTP GET /api/derivatives.
+"""
+
+from datetime import date as Date
+from typing import Optional
+
+import requests
+
+from .base import BaseFetcher
+from config import config
+
+
+class DerivativesFetcher(BaseFetcher):
+    """Fetches derivatives snapshot from data_provider /api/derivatives."""
+
+    def __init__(self, provider_url: Optional[str] = None):
+        super().__init__(timeout=15, max_retries=3)
+        self.provider_url = provider_url or config.provider_url
+
+    def fetch(self, target_date: Optional[Date] = None) -> list[dict]:
+        """Fetch derivatives data. Returns list with one record dict."""
+        url = f"{self.provider_url}/api/derivatives"
+        params = {"symbol": config.btc_symbol}
+        try:
+            data = self._get(url, params=params)
+            record = {
+                "date": str(target_date or Date.today()),
+                "symbol": config.btc_symbol,
+                "funding_rate": data.get("funding_rate"),
+                "open_interest": data.get("open_interest"),
+                "oi_24h_change_pct": data.get("oi_change_pct"),
+                "basis_annualised_pct": data.get("basis"),
+                "source": "data_provider",
+            }
+            return [record]
+        except Exception:
+            return []
+
+    def store(self, db_path: Optional[str] = None, records: Optional[list[dict]] = None) -> int:
+        """Store derivatives records into SQLite."""
+        import sqlite3
+        db_path = db_path or config.db_path
+        records = records or []
+        conn = sqlite3.connect(db_path)
+        count = 0
+        for r in records:
+            try:
+                conn.execute("""
+                    INSERT OR REPLACE INTO derivatives
+                    (date, symbol, funding_rate, open_interest, oi_24h_change_pct,
+                     long_liquidations, short_liquidations, basis_annualised_pct)
+                    VALUES (?, ?, ?, ?, ?, NULL, NULL, ?)
+                """, (
+                    r["date"], r.get("symbol", config.btc_symbol),
+                    r.get("funding_rate"), r.get("open_interest"),
+                    r.get("oi_24h_change_pct"), r.get("basis_annualised_pct"),
+                ))
+                count += 1
+            except Exception:
+                continue
+        conn.commit()
+        conn.close()
+        return count

ChanMacro/fetchers/ohlcv.py

@@ -0,0 +1,157 @@
+"""
+fetchers/ohlcv.py — Fetches BTC daily OHLCV from the existing data_provider service.
+
+Also pre-computes EMA20/60/120, ATR(14), BB width, ADX(14).
+"""
+
+from datetime import date as Date, datetime, timedelta
+from typing import Optional
+import logging
+
+import pandas as pd
+import numpy as np
+import requests
+
+from .base import BaseFetcher
+from config import config
+
+
+class OHLCVFetcher(BaseFetcher):
+    """Fetches BTC daily K-line data from data_provider API."""
+
+    def __init__(self, provider_url: Optional[str] = None):
+        super().__init__(timeout=30, max_retries=3)
+        self.provider_url = provider_url or config.provider_url
+        self.symbol = config.btc_symbol
+        self.logger = logging.getLogger(__name__)
+
+    def fetch(self, target_date: Optional[Date] = None) -> pd.DataFrame:
+        """
+        Fetch daily OHLCV for BTC. Returns DataFrame with computed indicators.
+
+        Fetches enough history (200 bars) to compute EMAs/ATR/BB/ADX accurately.
+        """
+        url = f"{self.provider_url}/api/candles"
+        params = {
+            "symbol": self.symbol,
+            "tf": "1d",
+            "limit": 200,
+        }
+        resp = requests.get(url, params=params, timeout=self.timeout)
+        resp.raise_for_status()
+        data = resp.json()
+
+        if not data:
+            self.logger.warning("OHLCV API returned empty data")
+            return pd.DataFrame()
+
+        df = pd.DataFrame(data)
+        df["timestamp"] = pd.to_datetime(df["timestamp"], unit="ms", utc=True)
+        df["date"] = df["timestamp"].dt.date
+        df = df.drop_duplicates(subset="date").sort_values("date").reset_index(drop=True)
+
+        # Rename columns to match expected format
+        df = df.rename(columns={
+            "open": "open", "high": "high", "low": "low", "close": "close",
+            "volume": "volume",
+        })
+
+        # Compute indicators
+        df = self._add_indicators(df)
+
+        return df
+
+    def _add_indicators(self, df: pd.DataFrame) -> pd.DataFrame:
+        """Add EMA, ATR, BB, ADX indicators."""
+        close = df["close"].astype(float)
+        high = df["high"].astype(float)
+        low = df["low"].astype(float)
+
+        # EMAs
+        df["ema20"] = close.ewm(span=20, adjust=False).mean()
+        df["ema60"] = close.ewm(span=60, adjust=False).mean()
+        df["ema120"] = close.ewm(span=120, adjust=False).mean()
+
+        # ATR(14)
+        tr1 = high - low
+        tr2 = (high - close.shift(1)).abs()
+        tr3 = (low - close.shift(1)).abs()
+        tr = pd.concat([tr1, tr2, tr3], axis=1).max(axis=1)
+        df["atr_14"] = tr.rolling(14).mean()
+
+        # Bollinger Bands width
+        sma20 = close.rolling(20).mean()
+        std20 = close.rolling(20).std()
+        df["bb_width"] = (2 * std20) / sma20 * 100  # as percentage
+
+        # ADX(14)
+        df["adx_14"] = self._compute_adx(df, period=14)
+
+        return df
+
+    @staticmethod
+    def _compute_adx(df: pd.DataFrame, period: int = 14) -> pd.Series:
+        """Compute ADX from OHLC data."""
+        high = df["high"].astype(float)
+        low = df["low"].astype(float)
+        close = df["close"].astype(float)
+
+        plus_dm = high.diff()
+        minus_dm = low.diff().abs() * -1
+        plus_dm = plus_dm.where(plus_dm > 0, 0)
+        minus_dm = minus_dm.where(minus_dm < 0, 0).abs()
+
+        tr1 = high - low
+        tr2 = (high - close.shift(1)).abs()
+        tr3 = (low - close.shift(1)).abs()
+        tr = pd.concat([tr1, tr2, tr3], axis=1).max(axis=1)
+
+        atr = tr.rolling(period).mean()
+        plus_di = 100 * (plus_dm.rolling(period).mean() / atr)
+        minus_di = 100 * (minus_dm.rolling(period).mean() / atr)
+
+        dx = (abs(plus_di - minus_di) / (plus_di + minus_di)) * 100
+        adx = dx.rolling(period).mean()
+        return adx
+
+    def store(self, db_path: str, records: list[dict]) -> int:
+        """Store OHLCV records into SQLite. Not used directly — see store_df."""
+        return 0
+
+    def store_df(self, df: pd.DataFrame, db_path: Optional[str] = None) -> int:
+        """Store the DataFrame into the ohlcv_daily table."""
+        import sqlite3
+        db_path = db_path or config.db_path
+        conn = sqlite3.connect(db_path)
+
+        count = 0
+        for _, row in df.iterrows():
+            if pd.isna(row.get("date")):
+                continue
+            date_str = str(row["date"])
+            try:
+                conn.execute("""
+                    INSERT OR REPLACE INTO ohlcv_daily
+                    (date, symbol, open, high, low, close, volume,
+                     ema20, ema60, ema120, atr_14, bb_width, adx_14)
+                    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+                """, (
+                    date_str, self.symbol,
+                    float(row["open"]), float(row["high"]),
+                    float(row["low"]), float(row["close"]),
+                    float(row.get("volume", 0)),
+                    float(row["ema20"]) if not pd.isna(row.get("ema20")) else None,
+                    float(row["ema60"]) if not pd.isna(row.get("ema60")) else None,
+                    float(row["ema120"]) if not pd.isna(row.get("ema120")) else None,
+                    float(row["atr_14"]) if not pd.isna(row.get("atr_14")) else None,
+                    float(row["bb_width"]) if not pd.isna(row.get("bb_width")) else None,
+                    float(row["adx_14"]) if not pd.isna(row.get("adx_14")) else None,
+                ))
+                count += 1
+            except Exception as e:
+                self.logger.debug(f"Skip row {date_str}: {e}")
+
+        conn.commit()
+        conn.close()
+        self.logger.info(f"Stored {count} OHLCV rows")
+        return count

ChanMacro/main.py

@@ -0,0 +1,17 @@
+#!/usr/bin/env python3
+"""
+main.py — ChanMacro entry point.
+
+CLI: python main.py fetch|score|regime|serve
+"""
+
+import sys
+import os
+
+# Ensure package root is on path
+sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
+
+from cli import main
+
+if __name__ == "__main__":
+    main()

ChanMacro/models.py

@@ -0,0 +1,370 @@
+"""
+models.py — Pydantic v2 models and enums for ChanMacro.
+
+All market state types, factor scores, and database record models.
+"""
+
+from datetime import date as Date
+from enum import Enum
+from typing import Optional
+
+from pydantic import BaseModel, Field, field_validator
+
+
+# ═══════════════════════════════════════════════════════════════
+# Shared validators
+# ═══════════════════════════════════════════════════════════════
+
+def _parse_date(v):
+    """Reusable date-string parser for field_validator."""
+    if isinstance(v, str):
+        return Date.fromisoformat(v)
+    return v
+
+
+# ═══════════════════════════════════════════════════════════════
+# Enums
+# ═══════════════════════════════════════════════════════════════
+
+class MarketRegime(str, Enum):
+    """V1: 3-state regime (factor-locked: Price + Breadth + Vol)."""
+    TREND = "TREND"
+    RANGE = "RANGE"
+    PANIC = "PANIC"
+
+
+class OIState(str, Enum):
+    """Discrete OI × Price state machine. NOT compressed into a score."""
+    NEW_LONGS = "New Longs"
+    SHORT_COVERING = "Short Covering"
+    NEW_SHORTS = "New Shorts"
+    LONG_EXIT = "Long Exit"
+    NEUTRAL = "Neutral"
+
+
+class BreadthBucket(str, Enum):
+    """Quantile-based breadth buckets — always have samples regardless of cycle."""
+    EXTREME = "EXTREME"
+    STRONG = "STRONG"
+    NORMAL = "NORMAL"
+    WEAK = "WEAK"
+    PANIC = "PANIC"
+
+
+class VolRegime(str, Enum):
+    """Volatility regime classification."""
+    LOW_VOL = "LOW_VOL"
+    NORMAL_VOL = "NORMAL_VOL"
+    HIGH_VOL = "HIGH_VOL"
+    EXPLOSIVE_VOL = "EXPLOSIVE_VOL"
+
+
+class MacroDirection(str, Enum):
+    BULLISH = "bullish"
+    NEUTRAL = "neutral"
+    BEARISH = "bearish"
+
+
+class MarketEmotion(str, Enum):
+    EXTREME_FEAR = "Extreme Fear"
+    FEAR = "Fear"
+    NEUTRAL = "Neutral"
+    GREED = "Greed"
+    EXTREME_GREED = "Extreme Greed"
+
+
+class FlowState(str, Enum):
+    STRONG_INFLOW = "Strong Inflow"
+    INFLOW = "Inflow"
+    NEUTRAL = "Neutral"
+    OUTFLOW = "Outflow"
+    STRONG_OUTFLOW = "Strong Outflow"
+
+
+class CapitalState(str, Enum):
+    ENTERING = "Entering"
+    STABLE = "Stable"
+    EXITING = "Exiting"
+
+
+class SufficiencyLevel(str, Enum):
+    HIGH = "HIGH"
+    MEDIUM = "MEDIUM"
+    LOW = "LOW"
+    INSUFFICIENT = "INSUFFICIENT"
+
+
+class SignalGrade(str, Enum):
+    A = "A"
+    B = "B"
+    C = "C"
+
+
+# ═══════════════════════════════════════════════════════════════
+# L0: Raw Data Models
+# ═══════════════════════════════════════════════════════════════
+
+class OHLCVDaily(BaseModel):
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    symbol: str
+    open: float
+    high: float
+    low: float
+    close: float
+    volume: float
+    ema20: Optional[float] = None
+    ema60: Optional[float] = None
+    ema120: Optional[float] = None
+    atr_14: Optional[float] = None
+    bb_width: Optional[float] = None
+    adx_14: Optional[float] = None
+
+
+class BreadthRecord(BaseModel):
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    total_tracked: int = 50
+    advance_top50: int = 0
+    decline_top50: int = 0
+    above_ema20_top50: int = 0
+    new_highs_20d_top50: int = 0
+    btc_dominance: Optional[float] = None
+    advance_top20: int = 0
+    advance_top30: int = 0
+    above_ema20_top20: int = 0
+    above_ema20_top30: int = 0
+    new_highs_20d_top20: int = 0
+    new_highs_20d_top30: int = 0
+
+
+class DerivativesRecord(BaseModel):
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    symbol: str = "BTC/USDT:USDT"
+    funding_rate: Optional[float] = None
+    open_interest: Optional[float] = None
+    oi_24h_change_pct: Optional[float] = None
+    long_liquidations: Optional[float] = None
+    short_liquidations: Optional[float] = None
+    basis_annualised_pct: Optional[float] = None
+    source: str = "binance"
+
+
+class ETFFlowRecord(BaseModel):
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    product: str
+    net_flow_million: float
+    price: Optional[float] = None
+    source: str = "farside"
+
+
+class StablecoinSupplyRecord(BaseModel):
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    token: str
+    chain: str = "all"
+    supply: float
+    source: str = "defillama"
+
+
+# ═══════════════════════════════════════════════════════════════
+# L1: Factor Score Models
+# ═══════════════════════════════════════════════════════════════
+
+class FactorScore(BaseModel):
+    """Single factor scoring output."""
+    name: str = ""
+    score: float = Field(default=50.0, ge=0.0, le=100.0)
+    label: str = ""
+    direction: MacroDirection = MacroDirection.NEUTRAL
+    sub_scores: dict = Field(default_factory=dict)
+    narrative: str = ""
+
+
+class PriceStructureScore(FactorScore):
+    """Price Structure — 3 sub-dimensions."""
+    trend_strength: float = 0.0
+    volatility_compression: float = 0.0
+    momentum: float = 0.0
+
+
+class BreadthScore(FactorScore):
+    """Breadth — multi-tier market diffusion."""
+    breadth_top20: float = 0.0
+    breadth_top30: float = 0.0
+    breadth_top50: float = 0.0
+    breadth_bucket: BreadthBucket = BreadthBucket.NORMAL
+    breadth_divergence: float = 0.0
+    advance_pct_top50: float = 0.0
+    above_ema20_pct_top50: float = 0.0
+    new_highs_top50: int = 0
+    btc_dominance_7d_chg: Optional[float] = None
+
+
+class OIMatrixScore(FactorScore):
+    """OI Matrix — discrete state + continuous score."""
+    oi_state: OIState = OIState.NEUTRAL
+    price_change_pct: float = 0.0
+    oi_change_pct: float = 0.0
+
+
+class VolatilityRegimeScore(FactorScore):
+    """Volatility regime classification."""
+    vol_regime: VolRegime = VolRegime.NORMAL_VOL
+    atr_pct: float = 0.0
+    hv_ratio: float = 1.0
+    bb_width_ratio: float = 1.0
+
+
+# ═══════════════════════════════════════════════════════════════
+# L4: Market State Vector (the final product)
+# ═══════════════════════════════════════════════════════════════
+
+class MarketStateVector(BaseModel):
+    """L4: Complete market state description. NOT compressed into one number."""
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+
+    date: Date
+    symbol: str = "BTC/USDT:USDT"
+
+    regime: MarketRegime
+    regime_confidence: float = Field(ge=0.0, le=1.0)
+    regime_version: str
+    regime_maturity_score: float = Field(ge=0.0, le=100.0, default=50.0)
+
+    breadth_top20: float = Field(default=50.0, ge=0.0, le=100.0)
+    breadth_top30: float = Field(default=50.0, ge=0.0, le=100.0)
+    breadth_top50: float = Field(default=50.0, ge=0.0, le=100.0)
+    breadth_bucket: BreadthBucket = BreadthBucket.NORMAL
+    breadth_divergence: float = 0.0
+
+    oi_state: OIState = OIState.NEUTRAL
+    volatility_regime: VolRegime = VolRegime.NORMAL_VOL
+
+    price_structure_score: FactorScore = Field(default_factory=FactorScore)
+    breadth_score: BreadthScore = Field(default_factory=BreadthScore)
+    oi_matrix_score: OIMatrixScore = Field(default_factory=OIMatrixScore)
+    volatility_regime_score: VolatilityRegimeScore = Field(default_factory=VolatilityRegimeScore)
+
+    market_state_hash: str = ""
+
+    def compute_hash(self) -> str:
+        import hashlib
+        key = f"{self.regime.value}|{self.breadth_bucket.value}|{self.oi_state.value}|{self.volatility_regime.value}"
+        return hashlib.md5(key.encode()).hexdigest()[:12]
+
+    def state_embedding(self) -> list[float]:
+        return [
+            self.breadth_top20,
+            self.breadth_top30,
+            self.breadth_top50,
+            self.regime_maturity_score,
+            self.price_structure_score.score,
+        ]
+
+
+# ═══════════════════════════════════════════════════════════════
+# Factor Contribution
+# ═══════════════════════════════════════════════════════════════
+
+class FactorContribution(BaseModel):
+    """How much a factor contributed to the overall score."""
+    factor: str
+    raw_score: float
+    weight: float
+    impact: float
+    direction: str  # 'bullish' / 'bearish' / 'neutral'
+
+
+# ═══════════════════════════════════════════════════════════════
+# Regime Result
+# ═══════════════════════════════════════════════════════════════
+
+class RegimeResult(BaseModel):
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    regime: MarketRegime
+    confidence: float
+    regime_version: str
+    maturity_score: float
+    all_scores: dict = Field(default_factory=dict)
+    prior_regime: Optional[MarketRegime] = None
+    confirmation_days: int = 0
+
+
+class SignalFeatureRecord(BaseModel):
+    """A single signal → market state → outcome record."""
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    signal_type: str
+    signal_version: str = "b3_v1"
+    symbol: str = "BTC/USDT:USDT"
+
+    regime_version: str
+    signal_grade: Optional[SignalGrade] = None
+    signal_strength: Optional[float] = None
+
+    regime: MarketRegime
+    regime_confidence: float
+    regime_maturity_score: float
+    market_state_hash: str
+    state_embedding: str = "[]"
+    breadth_top20: float
+    breadth_top30: float
+    breadth_top50: float
+    breadth_bucket: BreadthBucket
+    breadth_divergence: float
+    oi_state: OIState
+    volatility_regime: VolRegime
+    price_structure_score: float
+
+    chan_trend_direction: Optional[str] = None
+    chan_pivot_count: Optional[int] = None
+    chan_divergence_type: Optional[str] = None
+
+    entry_price: Optional[float] = None
+    result_1d: Optional[float] = None
+    result_3d: Optional[float] = None
+    result_5d: Optional[float] = None
+    result_7d: Optional[float] = None
+    result_14d: Optional[float] = None
+    max_favorable_excursion: Optional[float] = None
+    max_adverse_excursion: Optional[float] = None
+    is_win_7d: Optional[int] = None
+
+
+class ExpectancyLayer(BaseModel):
+    name: str
+    posterior_winrate: float
+    raw_winrate: Optional[float] = None
+    samples: int = 0
+    effective_samples: float = 0.0
+    avg_return: Optional[float] = None
+
+
+class ExpectancyReport(BaseModel):
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    signal_type: str
+    date: Date
+    layers: list[ExpectancyLayer] = Field(default_factory=list)
+    final_estimate: float
+    sufficiency: SufficiencyLevel = SufficiencyLevel.INSUFFICIENT
+    prior_strength: int = 40
+    half_life_days: int = 180
+    source: str = "bayesian"
+
+    avg_return_7d: Optional[float] = None
+    profit_factor: Optional[float] = None
+    max_adverse_excursion: Optional[float] = None
+
+
+class DailyOutput(BaseModel):
+    """Final daily output: Market State + Expectancy."""
+    _parse_date = field_validator("date", mode="before")(_parse_date)
+    date: Date
+    market_state: MarketStateVector
+    expectancy: dict[str, ExpectancyReport] = Field(default_factory=dict)
+    ai_report_en: Optional[str] = None
+    ai_report_zh: Optional[str] = None

ChanMacro/regime_detector.py

@@ -0,0 +1,213 @@
+"""
+regime_detector.py — Market regime detection (V1: 3 states).
+
+★ FACTOR-LOCKED: Regime = f(Price Structure, Breadth, Volatility) — forever.
+  Fear, Liquidation, ETF, Funding are Context, NOT regime inputs.
+  Adding new factors MUST NOT change regime definition.
+
+★ VERSIONED: regime_version = 'v1_price_breadth_vol'.
+  Weight changes → new version. Multiple versions coexist.
+  Query: WHERE regime_version = 'v1_price_breadth_vol'.
+
+★ CONFIDENCE-BASED: Each regime gets a continuous score. Highest wins.
+  No hard thresholds (prevents boundary oscillation).
+"""
+
+from datetime import date as Date
+from typing import Optional
+from collections import deque
+
+from models import MarketRegime, RegimeResult
+from config import config
+
+
+class RegimeDetector:
+    """
+    Detects market regime from Price + Breadth + Vol.
+
+    V1: 3 regimes (TREND / RANGE / PANIC)
+    V2+: Can split TREND→TREND_UP/TREND_DOWN/EUPHORIA when samples > 500/regime.
+    """
+
+    def __init__(self, regime_version: Optional[str] = None):
+        self.version = regime_version or config.regime_version
+        self.w_price = config.regime_w_price
+        self.w_breadth = config.regime_w_breadth
+        self.w_vol = config.regime_w_vol
+        self.panic_w_anti_trend = config.regime_panic_w_anti_trend
+        self.panic_w_vol_extreme = config.regime_panic_w_vol_extreme
+
+        # State persistence
+        self._current_regime: Optional[MarketRegime] = None
+        self._pending_regime: Optional[MarketRegime] = None
+        self._confirmation_count: int = 0
+        self._consecutive_days: int = 0
+        self._regime_history: deque = deque(maxlen=100)
+
+        # Confirmation: 2 days minimum
+        self.MIN_CONFIRMATION = 2
+
+    def load_state(self, db_path: str):
+        """Restore regime state from the most recent regime_history record."""
+        import sqlite3
+        try:
+            conn = sqlite3.connect(db_path)
+            conn.row_factory = sqlite3.Row
+            row = conn.execute(
+                "SELECT regime, confidence, confirmation_days, maturity_score "
+                "FROM regime_history ORDER BY date DESC LIMIT 1"
+            ).fetchone()
+            conn.close()
+
+            if row:
+                regime_str = row["regime"]
+                if regime_str in ("TREND", "RANGE", "PANIC"):
+                    self._current_regime = MarketRegime(regime_str)
+                    self._consecutive_days = row["confirmation_days"] or 1
+        except Exception:
+            pass  # DB not initialized yet, use defaults
+
+    def detect(self, price_structure_score: float, breadth_score: float,
+               volatility_regime: str, date: Date) -> RegimeResult:
+        """
+        Detect regime from the 3 locked factors.
+
+        Args:
+            price_structure_score: 0-100 from PriceStructureScorer
+            breadth_score: 0-100 from BreadthScorer
+            volatility_regime: 'LOW_VOL'/'NORMAL_VOL'/'HIGH_VOL'/'EXPLOSIVE_VOL'
+            date: Target date
+        """
+        # ── Compute regime scores ────────────────────────
+        # TREND: strong price + strong breadth + non-extreme vol
+        trend_score = (
+            price_structure_score * self.w_price +
+            breadth_score * self.w_breadth +
+            self._vol_to_trend(volatility_regime) * self.w_vol
+        )
+
+        # RANGE: neutral price + neutral breadth + low vol
+        # Score how "range-like" each dimension is
+        price_neutral = 60 - abs(price_structure_score - 50)
+        breadth_neutral = 60 - abs(breadth_score - 50)
+        vol_neutral = 80 if volatility_regime in ("LOW_VOL", "NORMAL_VOL") else 30
+        range_score = (
+            price_neutral * 0.40 +
+            breadth_neutral * 0.40 +
+            vol_neutral * 0.20
+        )
+
+        # PANIC: very weak trend + extreme vol (NO Fear/Liquidation!)
+        anti_trend = 100 - trend_score
+        vol_extreme = 100 if volatility_regime == "EXPLOSIVE_VOL" else (
+            60 if volatility_regime == "HIGH_VOL" else 20
+        )
+        panic_score = (
+            anti_trend * self.panic_w_anti_trend +
+            vol_extreme * self.panic_w_vol_extreme
+        )
+
+        scores = {
+            MarketRegime.TREND: round(trend_score, 1),
+            MarketRegime.RANGE: round(range_score, 1),
+            MarketRegime.PANIC: round(panic_score, 1),
+        }
+
+        best_regime = max(scores, key=scores.get)
+
+        # ── Persistence check ────────────────────────────
+        prior_regime = self._current_regime
+
+        if best_regime == self._current_regime:
+            self._consecutive_days += 1
+            self._pending_regime = None
+            self._confirmation_count = 0
+        elif best_regime == self._pending_regime:
+            self._confirmation_count += 1
+            if self._confirmation_count >= self.MIN_CONFIRMATION:
+                # Transition confirmed
+                prior_regime = self._current_regime
+                self._current_regime = best_regime
+                self._consecutive_days = self.MIN_CONFIRMATION
+                self._pending_regime = None
+                self._confirmation_count = 0
+        else:
+            self._pending_regime = best_regime
+            self._confirmation_count = 1
+
+        # Fallback: if no current regime yet (first run)
+        if self._current_regime is None:
+            self._current_regime = best_regime
+            self._consecutive_days = 1
+
+        # ── Confidence: for the CONFIRMED regime, not the raw best ──
+        confirmed_regime = self._current_regime
+        confidence = scores[confirmed_regime] / 100.0
+
+        # ── Maturity ─────────────────────────────────────
+        maturity = self._compute_maturity(
+            trend_score, breadth_score, volatility_regime
+        )
+
+        # Track history
+        self._regime_history.append({
+            "date": date,
+            "regime": confirmed_regime.value,
+            "confidence": round(confidence, 3),
+        })
+
+        return RegimeResult(
+            date=date,
+            regime=confirmed_regime,
+            confidence=round(confidence, 3),
+            prior_regime=prior_regime,
+            regime_version=self.version,
+            maturity_score=round(maturity, 1),
+            all_scores={k.value: v for k, v in scores.items()},
+            confirmation_days=self._consecutive_days,
+        )
+
+    @property
+    def current_regime(self) -> Optional[MarketRegime]:
+        return self._current_regime
+
+    @property
+    def pending_regime(self) -> Optional[MarketRegime]:
+        return self._pending_regime
+
+    @property
+    def confirmation_progress(self) -> tuple[int, int]:
+        """(confirmed_days, required_days) for pending transition."""
+        return (self._confirmation_count, self.MIN_CONFIRMATION)
+
+    @staticmethod
+    def _vol_to_trend(vol_regime: str) -> float:
+        """Convert volatility regime to trend-contributing score."""
+        mapping = {
+            "LOW_VOL": 50,       # Low vol: neutral for trend
+            "NORMAL_VOL": 70,    # Normal vol: good for trend
+            "HIGH_VOL": 60,      # High vol: trending but risky
+            "EXPLOSIVE_VOL": 30, # Explosive: anti-trend
+        }
+        return mapping.get(vol_regime, 50)
+
+    @staticmethod
+    def _compute_maturity(trend_score: float, breadth_score: float,
+                          vol_regime: str) -> float:
+        """
+        Compute regime maturity: 0-100 continuous.
+        0-30: EMERGING (trend accelerating, breadth expanding)
+        30-70: CONFIRMED (stable)
+        70-100: EXHAUSTING (decelerating, vol abnormal)
+        """
+        # Trend strength contribution
+        trend_contrib = trend_score * 0.50
+
+        # Breadth contribution
+        breadth_contrib = breadth_score * 0.30
+
+        # Vol contribution (inverted: low vol = early, explosive = late)
+        vol_contrib = {"LOW_VOL": 20, "NORMAL_VOL": 40, "HIGH_VOL": 60, "EXPLOSIVE_VOL": 85}
+        vol_val = vol_contrib.get(vol_regime, 50) * 0.20
+
+        return trend_contrib + breadth_contrib + vol_val

ChanMacro/requirements.txt

@@ -0,0 +1,8 @@
+ccxt>=4.0.0
+pandas>=2.0.0
+numpy>=1.21.2
+pydantic>=2.0.0
+requests>=2.31.0
+python-dotenv>=1.0.0
+scipy>=1.10.0
+flask>=3.0.0

ChanMacro/run_tests.sh

@@ -0,0 +1,11 @@
+#!/bin/bash
+# run_tests.sh — Run the ChanMacro test suite.
+#
+# Usage:
+#   ./run_tests.sh              # All tests
+#   ./run_tests.sh -v           # Verbose
+#   ./run_tests.sh -k regime    # Only regime tests
+#   ./run_tests.sh --cov        # With coverage (requires pytest-cov)
+
+cd "$(dirname "$0")"
+python -m pytest tests/ "$@" --tb=short

ChanMacro/scheduler.py

@@ -0,0 +1,153 @@
+"""
+scheduler.py — 后台自动调度：定时拉取数据 + 计算因子 + 制度判定。
+
+Python main.py cron         → 前台阻塞运行，每 N 分钟一个 tick
+Web app 启动时自动启动调度器  → 后台线程，不阻塞 Web 请求
+"""
+
+import threading
+import logging
+import time
+from datetime import datetime, timezone, timedelta
+from typing import Optional
+
+logger = logging.getLogger("chanmacro.scheduler")
+
+
+class MacroScheduler:
+    """后台调度器：定时 fetch + score。"""
+
+    def __init__(self, interval_minutes: int = 60):
+        self.interval = interval_minutes
+        self._thread: Optional[threading.Thread] = None
+        self._stop = threading.Event()
+        self._last_run: Optional[datetime] = None
+        self._running = False
+
+    def start(self) -> None:
+        """启动后台线程。"""
+        if self._running:
+            return
+        self._stop.clear()
+        self._thread = threading.Thread(target=self._loop, name="macro-scheduler", daemon=True)
+        self._thread.start()
+        self._running = True
+        logger.info(f"调度器已启动, 每 {self.interval} 分钟执行一次")
+
+    def stop(self) -> None:
+        """停止后台线程。"""
+        self._stop.set()
+        self._running = False
+        logger.info("调度器已停止")
+
+    @property
+    def last_run(self) -> Optional[datetime]:
+        return self._last_run
+
+    def _loop(self) -> None:
+        """后台循环。"""
+        # 首次启动立即跑一次
+        self._tick()
+
+        while not self._stop.wait(self.interval * 60):
+            self._tick()
+
+    def _tick(self) -> None:
+        """执行一次：fetch → score。"""
+        try:
+            from fetchers.ohlcv import OHLCVFetcher
+            from fetchers.breadth import BreadthFetcher
+            from fetchers.derivatives import DerivativesFetcher
+            from database import init_db
+            from datetime import date as Date
+
+            init_db()
+            today = Date.today()
+
+            # Fetch
+            ohlcv = OHLCVFetcher()
+            df = ohlcv.fetch()
+            if not df.empty:
+                ohlcv.store_df(df)
+
+            breadth = BreadthFetcher()
+            record = breadth.fetch()
+            if record:
+                breadth.store(record=record)
+
+            deriv = DerivativesFetcher()
+            records = deriv.fetch(today)
+            if records:
+                deriv.store(records=records)
+
+            # Score + Regime (also persisted inside _build_state)
+            from scoring.price_structure import PriceStructureScorer
+            from scoring.breadth_scorer import BreadthScorer
+            from scoring.oi_matrix import OIMatrixScorer
+            from scoring.volatility_regime import VolatilityRegimeScorer
+            from regime_detector import RegimeDetector
+            from models import MarketStateVector
+            from config import config
+            import json
+            from database import get_connection
+
+            ps = PriceStructureScorer().compute(today)
+            br = BreadthScorer().compute(today)
+            oi = OIMatrixScorer().compute(today)
+            vol = VolatilityRegimeScorer().compute(today)
+
+            detector = RegimeDetector()
+            detector.load_state(config.db_path)
+            r = detector.detect(ps.score, br.breadth_top50, vol.vol_regime.value, today)
+
+            conn = get_connection()
+            conn.execute("""
+                INSERT OR REPLACE INTO regime_history
+                (date, regime, confidence, regime_version, maturity_score,
+                 all_scores_json, prior_regime, confirmation_days)
+                VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+            """, (
+                str(today), r.regime.value, r.confidence, r.regime_version,
+                r.maturity_score, json.dumps(r.all_scores),
+                r.prior_regime.value if r.prior_regime else None,
+                r.confirmation_days,
+            ))
+            conn.commit()
+            conn.close()
+
+            # 检测新信号（每天运行一次，UTC 0 点后首次触发）
+            now = datetime.now(timezone.utc)
+            if self._last_run is None or now.date() > self._last_run.date():
+                try:
+                    from chan_integration import ChanSignalDetector
+                    detector = ChanSignalDetector()
+                    # 检测最近 90 天的 4h 信号
+                    count = detector.populate_signal_features(
+                        start_date=(today - __import__('datetime').timedelta(days=90)).isoformat(),
+                        end_date=today.isoformat(),
+                    )
+                    if count > 0:
+                        logger.info(f"新增 {count} 条信号记录")
+                except Exception as e:
+                    logger.debug(f"信号检测跳过: {e}")
+
+            self._last_run = now
+            logger.info(
+                f"Tick 完成: regime={r.regime.value} conf={r.confidence:.2f} "
+                f"breadth={br.score:.0f}({br.breadth_bucket.value}) "
+                f"price={ps.score:.0f} oi={oi.oi_state.value} vol={vol.vol_regime.value}"
+            )
+
+        except Exception as e:
+            logger.error(f"Tick 失败: {e}", exc_info=True)
+
+
+# 单例
+_scheduler: Optional[MacroScheduler] = None
+
+
+def get_scheduler() -> MacroScheduler:
+    global _scheduler
+    if _scheduler is None:
+        _scheduler = MacroScheduler(interval_minutes=60)
+    return _scheduler

ChanMacro/scoring/__init__.py

@@ -0,0 +1,6 @@
+"""Scoring engine — L1 factor computation."""
+from .base import BaseScorer
+from .price_structure import PriceStructureScorer
+from .breadth_scorer import BreadthScorer
+from .oi_matrix import OIMatrixScorer
+from .volatility_regime import VolatilityRegimeScorer

ChanMacro/scoring/base.py

@@ -0,0 +1,28 @@
+"""
+scoring/base.py — Abstract base class for all scoring modules.
+"""
+
+from abc import ABC, abstractmethod
+from datetime import date as Date
+from typing import Optional
+import sqlite3
+
+from models import FactorScore
+from config import config
+
+
+class BaseScorer(ABC):
+    """Abstract base for all factor scorers."""
+
+    def __init__(self, db_path: Optional[str] = None):
+        self.db_path = db_path or config.db_path
+
+    def get_connection(self) -> sqlite3.Connection:
+        conn = sqlite3.connect(self.db_path)
+        conn.row_factory = sqlite3.Row
+        return conn
+
+    @abstractmethod
+    def compute(self, target_date: Date) -> FactorScore:
+        """Compute factor score for a given date from database records."""
+        ...

ChanMacro/scoring/breadth_scorer.py

@@ -0,0 +1,218 @@
+"""
+scoring/breadth_scorer.py — Market Breadth Score.
+
+The first citizen of the system. Diffusion always leads price.
+
+Multi-tier: Top20 / Top30 / Top50.
+Quantile-based bucketing: EXTREME / STRONG / NORMAL / WEAK / PANIC.
+
+4 sub-indicators (equal weight):
+  1. Advance/Decline ratio (30%)
+  2. % above EMA20 (35%)
+  3. New 20d highs (20%)
+  4. BTC Dominance change (15%, inverted)
+"""
+
+from datetime import date as Date
+import sqlite3
+import numpy as np
+import pandas as pd
+
+from .base import BaseScorer
+from .constants import (
+    BREADTH_W_ADVANCE, BREADTH_W_EMA20, BREADTH_W_NEW_HIGHS, BREADTH_W_BTC_DOM,
+)
+from models import FactorScore, BreadthScore, BreadthBucket, MacroDirection
+from config import config
+
+
+class BreadthScorer(BaseScorer):
+    """Scores market breadth with quantile-based bucketing."""
+
+    def compute(self, target_date: Date) -> BreadthScore:
+        conn = self.get_connection()
+        try:
+            row = conn.execute(
+                "SELECT * FROM breadth_daily WHERE date = ?", (str(target_date),)
+            ).fetchone()
+
+            if row is None:
+                return BreadthScore(
+                    name="Breadth",
+                    score=50.0,
+                    label="No Data",
+                    breadth_bucket=BreadthBucket.NORMAL,
+                )
+
+            row = dict(row)
+            total = row.get("total_tracked", 50) or 50
+
+            # 1. Advance/Decline ratio
+            advance = row.get("advance_top50", 0) or 0
+            decline = row.get("decline_top50", 0) or 0
+            if advance + decline > 0:
+                ad_ratio = advance / (advance + decline)
+            else:
+                ad_ratio = 0.5
+            ad_score = ad_ratio * 100
+
+            # 2. % above EMA20
+            above_ema = row.get("above_ema20_top50", 0) or 0
+            ema_pct = above_ema / total if total > 0 else 0.5
+            ema_score = ema_pct * 100
+
+            # 3. New highs
+            new_highs = row.get("new_highs_20d_top50", 0) or 0
+            highs_pct = new_highs / total if total > 0 else 0
+            highs_score = highs_pct * 100
+
+            # 4. BTC Dominance (inverted: BTC.D up = bearish for alts)
+            btc_dom = row.get("btc_dominance")
+            btc_dom_score = 50.0  # neutral default
+            if btc_dom is not None:
+                # Placeholder — needs historical comparison
+                btc_dom_score = 50.0
+
+            # Weighted aggregate
+            score = (
+                ad_score * BREADTH_W_ADVANCE +
+                ema_score * BREADTH_W_EMA20 +
+                highs_score * BREADTH_W_NEW_HIGHS +
+                btc_dom_score * BREADTH_W_BTC_DOM
+            )
+
+            # Multi-tier breadth
+            b20 = self._compute_tier_breadth(row, 20, total)
+            b30 = self._compute_tier_breadth(row, 30, total)
+            b50 = score  # Top50 = full score
+
+            # Quantile bucket
+            bucket = self._assign_bucket(score)
+
+            # Divergence
+            divergence = b20 - b50
+
+            # Direction
+            if score >= 60:
+                direction = MacroDirection.BULLISH
+            elif score <= 40:
+                direction = MacroDirection.BEARISH
+            else:
+                direction = MacroDirection.NEUTRAL
+
+            # Narrative
+            narrative = self._build_narrative(bucket, divergence, ema_pct, ad_ratio)
+
+            return BreadthScore(
+                name="Breadth",
+                score=round(score, 1),
+                label=bucket.value,
+                direction=direction,
+                breadth_top20=round(b20, 1),
+                breadth_top30=round(b30, 1),
+                breadth_top50=round(b50, 1),
+                breadth_bucket=bucket,
+                breadth_divergence=round(divergence, 1),
+                advance_pct_top50=round(ad_ratio * 100, 1),
+                above_ema20_pct_top50=round(ema_pct * 100, 1),
+                new_highs_top50=new_highs,
+                sub_scores={
+                    "advance_decline": round(ad_score, 1),
+                    "above_ema20": round(ema_score, 1),
+                    "new_highs": round(highs_score, 1),
+                    "btc_dominance": round(btc_dom_score, 1),
+                },
+                narrative=narrative,
+            )
+        finally:
+            conn.close()
+
+    def _compute_tier_breadth(self, row: dict, tier: int, total: int) -> float:
+        """Compute breadth score for a specific tier (Top20 or Top30)."""
+        advance = row.get(f"advance_top{tier}", 0) or 0
+        above_ema = row.get(f"above_ema20_top{tier}", 0) or 0
+        new_highs = row.get(f"new_highs_20d_top{tier}", 0) or 0
+
+        tier_actual = min(tier, total)
+        if tier_actual == 0:
+            return 50.0
+
+        ad_ratio = advance / tier_actual if tier_actual > 0 else 0.5
+        ema_ratio = above_ema / tier_actual if tier_actual > 0 else 0.5
+        highs_ratio = new_highs / tier_actual if tier_actual > 0 else 0
+
+        return (
+            ad_ratio * 100 * BREADTH_W_ADVANCE +
+            ema_ratio * 100 * BREADTH_W_EMA20 +
+            highs_ratio * 100 * BREADTH_W_NEW_HIGHS +
+            50 * BREADTH_W_BTC_DOM  # neutral for BTC.D
+        )
+
+    def _assign_bucket(self, score: float) -> BreadthBucket:
+        """Assign quantile-based bucket. V1 uses fixed thresholds until history accumulated."""
+        # V1: fixed thresholds (will switch to quantile when enough history)
+        if score >= 80:
+            return BreadthBucket.EXTREME
+        elif score >= 60:
+            return BreadthBucket.STRONG
+        elif score >= 40:
+            return BreadthBucket.NORMAL
+        elif score >= 20:
+            return BreadthBucket.WEAK
+        else:
+            return BreadthBucket.PANIC
+
+    @staticmethod
+    def compute_quantile_boundaries(db_path: str) -> dict:
+        """Compute quantile boundaries from historical breadth data.
+
+        This should be called after accumulating enough history (> 1 year).
+        Returns boundaries for pd.qcut.
+        """
+        conn = sqlite3.connect(db_path)
+        df = pd.read_sql_query(
+            "SELECT date, advance_top50, decline_top50, above_ema20_top50 FROM breadth_daily",
+            conn
+        )
+        conn.close()
+
+        if len(df) < 100:
+            return {"boundaries": [0, 20, 40, 60, 80, 100], "is_quantile": False}
+
+        df["ad_ratio"] = df["advance_top50"] / (df["advance_top50"] + df["decline_top50"])
+        df["ema_ratio"] = df["above_ema20_top50"] / 50
+        df["breadth_raw"] = (
+            df["ad_ratio"] * BREADTH_W_ADVANCE * 100 +
+            df["ema_ratio"] * BREADTH_W_EMA20 * 100 +
+            40 * BREADTH_W_NEW_HIGHS +
+            50 * BREADTH_W_BTC_DOM
+        )
+
+        boundaries = list(np.percentile(df["breadth_raw"].dropna(), [10, 30, 70, 90]))
+        return {
+            "boundaries": [0] + boundaries + [100],
+            "is_quantile": True,
+            "n_samples": len(df),
+        }
+
+    @staticmethod
+    def _build_narrative(bucket: BreadthBucket, divergence: float,
+                         ema_pct: float, ad_ratio: float) -> str:
+        parts = []
+        if bucket == BreadthBucket.EXTREME:
+            parts.append(f"全市场极度扩散({ema_pct:.0%}站上EMA20)")
+        elif bucket == BreadthBucket.STRONG:
+            parts.append("市场广度强势")
+        elif bucket == BreadthBucket.NORMAL:
+            parts.append("市场广度中性")
+        elif bucket == BreadthBucket.WEAK:
+            parts.append("市场广度疲弱")
+        else:
+            parts.append("市场广度恐慌")
+
+        if divergence > 10:
+            parts.append("资金集中于大市值(Top20>>Top50)")
+        elif divergence < -10:
+            parts.append("垃圾币狂欢(Top50>>Top20)")
+
+        return ", ".join(parts)

ChanMacro/scoring/constants.py

@@ -0,0 +1,98 @@
+"""
+scoring/constants.py — Scoring thresholds, scale factors, and reference values.
+
+All magic numbers in one place. Tune these via Phase 0 validation.
+"""
+
+# ── Price Structure ──────────────────────────────────────────
+# ADX thresholds
+ADX_TREND_THRESHOLD = 25       # ADX > 25 = trending
+ADX_STRONG_THRESHOLD = 40      # ADX > 40 = strong trend
+
+# EMA alignment
+EMA_ALIGNMENT_BULLISH = 1.0    # EMA20 > EMA60 > EMA120
+EMA_ALIGNMENT_NEUTRAL = 0.5    # mixed
+EMA_ALIGNMENT_BEARISH = 0.0    # EMA20 < EMA60 < EMA120
+
+# Volatility compression (BB width relative to 20d average)
+BB_COMPRESSION_LOW = 0.7       # < 70% of avg = compressing
+BB_COMPRESSION_HIGH = 1.5      # > 150% of avg = expanding
+
+# Momentum (ROC annualized)
+ROC_STRONG_BULLISH = 10.0      # % over period
+ROC_STRONG_BEARISH = -10.0
+
+# Consecutive candle threshold
+CONSECUTIVE_CANDLES_SIGNAL = 4
+
+# ── Breadth ──────────────────────────────────────────────────
+# Quantile boundaries for breadth buckets
+BREADTH_QUANTILES = [0, 0.1, 0.3, 0.7, 0.9, 1.0]  # PANIC/WEAK/NORMAL/STRONG/EXTREME
+
+# Breadth score computation weights
+BREADTH_W_ADVANCE = 0.30       # advance/decline ratio
+BREADTH_W_EMA20 = 0.35         # % above EMA20
+BREADTH_W_NEW_HIGHS = 0.20     # new highs count
+BREADTH_W_BTC_DOM = 0.15       # BTC dominance change (inverted)
+
+# ── OI Matrix ────────────────────────────────────────────────
+OI_PRICE_THRESHOLD = 0.5       # min |price_change%| to classify
+OI_OI_THRESHOLD = 0.5          # min |OI_change%| to classify
+
+# Score mapping for OI states
+OI_STATE_SCORES = {
+    "New Longs": 85,
+    "Short Covering": 60,
+    "New Shorts": 20,
+    "Long Exit": 35,
+    "Neutral": 50,
+}
+
+# ── Volatility Regime ────────────────────────────────────────
+VOL_LOW = 2.0                  # ATR/Close % below this = LOW_VOL
+VOL_HIGH = 5.0                 # ATR/Close % below this = HIGH_VOL (above = EXPLOSIVE)
+HV_RATIO_LOW = 0.7             # HV(20)/HV(60) below this = compressing
+HV_RATIO_HIGH = 1.5            # HV(20)/HV(60) above this = expanding
+
+# Score mapping
+VOL_REGIME_SCORES = {
+    "LOW_VOL": 40,              # Low vol → neutral with breakout potential
+    "NORMAL_VOL": 55,
+    "HIGH_VOL": 75,
+    "EXPLOSIVE_VOL": 90,
+}
+
+# ── Regime ───────────────────────────────────────────────────
+REGIME_W_PRICE = 0.35
+REGIME_W_BREADTH = 0.50
+REGIME_W_VOL = 0.15
+
+# PANIC: anti-trend + extreme vol (NO Fear/Liquidation)
+PANIC_W_ANTI_TREND = 0.60
+PANIC_W_VOL_EXTREME = 0.40
+
+# ── Trend (L2) ───────────────────────────────────────────────
+TREND_W_PRICE = 0.30
+TREND_W_BREADTH = 0.70
+
+# ── Maturity ─────────────────────────────────────────────────
+MATURITY_W_TREND = 0.50
+MATURITY_W_BREADTH = 0.30
+MATURITY_W_VOL = 0.20
+
+# ── Expectancy ───────────────────────────────────────────────
+HALF_LIFE_DAYS = 180
+SUFFICIENCY_MIN = 30
+SUFFICIENCY_LOW = 50
+SUFFICIENCY_MEDIUM = 100
+LEVEL_MIN_SAMPLES = 50
+KNN_MAX_DISTANCE = 0.35
+KNN_K = 200
+
+# ── Validation ───────────────────────────────────────────────
+MIN_AVG_DURATION = 5
+MAX_FLIP_RATE = 0.15
+MIN_IC_THRESHOLD = 0.03
+MIN_ICIR_THRESHOLD = 0.5
+MIN_IG_THRESHOLD = 0.1      # Information Gain for regime factors
+MIN_KL_THRESHOLD = 0.5       # KL Divergence for regime separation

ChanMacro/scoring/oi_matrix.py

@@ -0,0 +1,137 @@
+"""
+scoring/oi_matrix.py — OI × Price 2×2 state machine.
+
+Discrete states, NOT a continuous score:
+  NEW_LONGS:       Price↑ OI↑  → new money entering, trend continuation
+  SHORT_COVERING:  Price↑ OI↓  → shorts covering, rally fragile
+  NEW_SHORTS:      Price↓ OI↑  → new shorts entering, trend continuation
+  LONG_EXIT:       Price↓ OI↓  → longs stopping out, panic (possible bottom)
+  NEUTRAL:         flat        → noise, don't force classification
+"""
+
+from datetime import date as Date
+import sqlite3
+
+from .base import BaseScorer
+from .constants import OI_PRICE_THRESHOLD, OI_OI_THRESHOLD, OI_STATE_SCORES
+from models import FactorScore, OIMatrixScore, OIState, MacroDirection
+from config import config
+
+
+class OIMatrixScorer(BaseScorer):
+    """Classifies OI × Price state and assigns score."""
+
+    def compute(self, target_date: Date) -> OIMatrixScore:
+        conn = self.get_connection()
+        try:
+            row = conn.execute(
+                "SELECT * FROM derivatives WHERE date = ? AND symbol = 'BTC/USDT:USDT'",
+                (str(target_date),)
+            ).fetchone()
+
+            if row is None:
+                return OIMatrixScore(
+                    name="OI Matrix",
+                    score=50.0,
+                    label="No Data",
+                    oi_state=OIState.NEUTRAL,
+                )
+
+            row = dict(row)
+            oi_change = row.get("oi_24h_change_pct") or 0
+
+            # Get price change from OHLCV
+            price_change = self._get_price_change(conn, str(target_date))
+
+            # Classify state
+            oi_state = self._classify(price_change, oi_change)
+
+            # Score from state
+            score = OI_STATE_SCORES.get(oi_state.value, 50)
+
+            # Direction
+            if oi_state == OIState.NEW_LONGS:
+                direction = MacroDirection.BULLISH
+            elif oi_state == OIState.SHORT_COVERING:
+                direction = MacroDirection.BULLISH  # bullish but fragile
+            elif oi_state == OIState.NEW_SHORTS:
+                direction = MacroDirection.BEARISH
+            elif oi_state == OIState.LONG_EXIT:
+                direction = MacroDirection.BEARISH  # bearish but possible bottom
+            else:
+                direction = MacroDirection.NEUTRAL
+
+            # Narrative
+            narrative = self._build_narrative(oi_state, price_change, oi_change)
+
+            return OIMatrixScore(
+                name="OI Matrix",
+                score=float(score),
+                label=oi_state.value,
+                direction=direction,
+                oi_state=oi_state,
+                price_change_pct=round(price_change, 2),
+                oi_change_pct=round(oi_change, 2),
+                sub_scores={
+                    "price_change_pct": round(price_change, 2),
+                    "oi_change_pct": round(oi_change, 2),
+                },
+                narrative=narrative,
+            )
+        finally:
+            conn.close()
+
+    def _get_price_change(self, conn: sqlite3.Connection, date_str: str) -> float:
+        """Get BTC 24h price change % for a given date."""
+        row = conn.execute(
+            "SELECT close FROM ohlcv_daily WHERE date = ? AND symbol = 'BTC/USDT:USDT'",
+            (date_str,)
+        ).fetchone()
+        if row is None:
+            return 0.0
+
+        # Get previous day close
+        prev = conn.execute(
+            "SELECT close FROM ohlcv_daily WHERE date < ? AND symbol = 'BTC/USDT:USDT' ORDER BY date DESC LIMIT 1",
+            (date_str,)
+        ).fetchone()
+
+        if prev is None:
+            return 0.0
+
+        current_close = float(row["close"])
+        prev_close = float(prev["close"])
+        if prev_close == 0:
+            return 0.0
+
+        return (current_close - prev_close) / prev_close * 100
+
+    @staticmethod
+    def _classify(price_change_pct: float, oi_change_pct: float) -> OIState:
+        """Classify OI × Price into discrete state."""
+        price_up = price_change_pct > OI_PRICE_THRESHOLD
+        price_down = price_change_pct < -OI_PRICE_THRESHOLD
+        oi_up = oi_change_pct > OI_OI_THRESHOLD
+        oi_down = oi_change_pct < -OI_OI_THRESHOLD
+
+        if price_up and oi_up:
+            return OIState.NEW_LONGS
+        elif price_up and oi_down:
+            return OIState.SHORT_COVERING
+        elif price_down and oi_up:
+            return OIState.NEW_SHORTS
+        elif price_down and oi_down:
+            return OIState.LONG_EXIT
+        else:
+            return OIState.NEUTRAL
+
+    @staticmethod
+    def _build_narrative(state: OIState, price_chg: float, oi_chg: float) -> str:
+        mapping = {
+            OIState.NEW_LONGS: f"新多进场: 价格+{price_chg:.1f}%, OI+{oi_chg:.1f}%, 真金白银推动",
+            OIState.SHORT_COVERING: f"空头回补: 价格+{price_chg:.1f}%, OI{oi_chg:.1f}%, 上涨脆弱",
+            OIState.NEW_SHORTS: f"新空进场: 价格{price_chg:.1f}%, OI+{oi_chg:.1f}%, 趋势延续",
+            OIState.LONG_EXIT: f"多头止损: 价格{price_chg:.1f}%, OI{oi_chg:.1f}%, 恐慌(可能见底)",
+            OIState.NEUTRAL: "OI/价格变化不显著, 噪音区",
+        }
+        return mapping.get(state, "Unknown")

ChanMacro/scoring/price_structure.py

@@ -0,0 +1,248 @@
+"""
+scoring/price_structure.py — Price Structure Score (OHLCV-only).
+
+Three sub-dimensions:
+  1. Trend Strength (40%): EMA alignment + ADX
+  2. Volatility Compression (30%): ATR + BB width
+  3. Momentum (30%): ROC + consecutive candles
+
+This module works with zero external dependencies — just OHLCV data.
+"""
+
+from datetime import date as Date
+import sqlite3
+import math
+import numpy as np
+import pandas as pd
+
+from .base import BaseScorer
+from .constants import (
+    ADX_TREND_THRESHOLD, ADX_STRONG_THRESHOLD,
+    BB_COMPRESSION_LOW, BB_COMPRESSION_HIGH,
+    ROC_STRONG_BULLISH, ROC_STRONG_BEARISH,
+    CONSECUTIVE_CANDLES_SIGNAL,
+)
+from models import FactorScore, PriceStructureScore, MacroDirection
+from config import config
+
+
+class PriceStructureScorer(BaseScorer):
+    """Scores market structure from OHLCV data alone."""
+
+    def compute(self, target_date: Date) -> PriceStructureScore:
+        conn = self.get_connection()
+        try:
+            df = self._load_ohlcv(conn, str(target_date), lookback=120)
+            if df.empty:
+                return PriceStructureScore(
+                    name="Price Structure",
+                    score=50.0,
+                    label="No Data",
+                )
+
+            trend = self._score_trend_strength(df)
+            vol_comp = self._score_volatility_compression(df)
+            momentum = self._score_momentum(df)
+
+            # Weighted aggregate
+            score = trend * 0.40 + vol_comp * 0.30 + momentum * 0.30
+
+            # Determine direction
+            if trend > 60:
+                direction = MacroDirection.BULLISH
+            elif trend < 40:
+                direction = MacroDirection.BEARISH
+            else:
+                direction = MacroDirection.NEUTRAL
+
+            # Build narrative
+            latest = df.iloc[-1]
+            narrative = self._build_narrative(trend, vol_comp, momentum, latest)
+
+            return PriceStructureScore(
+                name="Price Structure",
+                score=round(score, 1),
+                label=self._label(score),
+                direction=direction,
+                trend_strength=round(trend, 1),
+                volatility_compression=round(vol_comp, 1),
+                momentum=round(momentum, 1),
+                sub_scores={
+                    "trend_strength": round(trend, 1),
+                    "volatility_compression": round(vol_comp, 1),
+                    "momentum": round(momentum, 1),
+                },
+                narrative=narrative,
+            )
+        finally:
+            conn.close()
+
+    def _load_ohlcv(self, conn: sqlite3.Connection, date_str: str,
+                    lookback: int = 120) -> pd.DataFrame:
+        """Load OHLCV data up to target_date."""
+        df = pd.read_sql_query(
+            "SELECT * FROM ohlcv_daily WHERE date <= ? ORDER BY date DESC LIMIT ?",
+            conn, params=(date_str, lookback)
+        )
+        if df.empty:
+            return df
+        return df.sort_values("date").reset_index(drop=True)
+
+    def _score_trend_strength(self, df: pd.DataFrame) -> float:
+        """Score trend based on EMA alignment and ADX."""
+        latest = df.iloc[-1]
+
+        # EMA alignment
+        ema20 = latest.get("ema20")
+        ema60 = latest.get("ema60")
+        ema120 = latest.get("ema120")
+
+        ema_score = 50.0
+        if ema20 and ema60 and ema120 and not pd.isna(ema20) and not pd.isna(ema60) and not pd.isna(ema120):
+            alignments = 0
+            if ema20 > ema60: alignments += 1
+            if ema60 > ema120: alignments += 1
+            if ema20 > ema120: alignments += 1
+
+            # Distance from EMAs
+            close = float(latest["close"])
+            ema20_dist = abs(close - ema20) / ema20 * 100 if ema20 else 0
+
+            if alignments == 3:
+                ema_score = 80 + min(ema20_dist, 15)  # strong bullish alignment
+            elif alignments == 0:
+                ema_score = 20 - min(ema20_dist, 15)  # strong bearish alignment
+            elif alignments == 2:
+                ema_score = 65
+            else:
+                ema_score = 35
+
+        # ADX
+        adx = latest.get("adx_14")
+        adx_score = 50.0
+        if adx and not pd.isna(adx):
+            if adx > ADX_STRONG_THRESHOLD:
+                adx_score = 85
+            elif adx > ADX_TREND_THRESHOLD:
+                adx_score = 65 + (adx - ADX_TREND_THRESHOLD) / (ADX_STRONG_THRESHOLD - ADX_TREND_THRESHOLD) * 20
+            else:
+                adx_score = 50 - (ADX_TREND_THRESHOLD - adx) / ADX_TREND_THRESHOLD * 30
+
+        return ema_score * 0.55 + adx_score * 0.45
+
+    def _score_volatility_compression(self, df: pd.DataFrame) -> float:
+        """Score volatility compression — expansion = high, compression = low-mid."""
+        latest = df.iloc[-1]
+
+        bb_width = latest.get("bb_width")
+        if not bb_width or pd.isna(bb_width) or len(df) < 20:
+            return 50.0
+
+        # BB width relative to 20d average
+        recent_bb = df["bb_width"].dropna().tail(20)
+        if len(recent_bb) < 10:
+            return 50.0
+
+        bb_avg = recent_bb.mean()
+        bb_ratio = bb_width / bb_avg if bb_avg > 0 else 1.0
+
+        if bb_ratio < BB_COMPRESSION_LOW:
+            # Compression → potential breakout, neutral-bullish
+            return 45 + (BB_COMPRESSION_LOW - bb_ratio) * 30
+        elif bb_ratio > BB_COMPRESSION_HIGH:
+            # Expansion → trending or chaotic
+            return 75 + min((bb_ratio - BB_COMPRESSION_HIGH) * 20, 20)
+        else:
+            # Normal
+            return 55
+
+    def _score_momentum(self, df: pd.DataFrame) -> float:
+        """Score momentum using ROC and consecutive candles."""
+        if len(df) < 10:
+            return 50.0
+
+        closes = df["close"].astype(float)
+        latest = float(closes.iloc[-1])
+
+        # ROC (5-bar)
+        if len(closes) >= 6:
+            roc5 = (closes.iloc[-1] - closes.iloc[-6]) / closes.iloc[-6] * 100
+        else:
+            roc5 = 0
+
+        # ROC (10-bar)
+        if len(closes) >= 11:
+            roc10 = (closes.iloc[-1] - closes.iloc[-11]) / closes.iloc[-11] * 100
+        else:
+            roc10 = 0
+
+        # ROC (20-bar)
+        if len(closes) >= 21:
+            roc20 = (closes.iloc[-1] - closes.iloc[-21]) / closes.iloc[-21] * 100
+        else:
+            roc20 = 0
+
+        # Score ROC: map to 0-100
+        def roc_to_score(roc, scale=15):
+            return 50 + np.clip(roc / scale * 50, -50, 50)
+
+        roc_score = roc_to_score(roc5, 10) * 0.4 + roc_to_score(roc10, 15) * 0.35 + roc_to_score(roc20, 20) * 0.25
+
+        # Consecutive candle direction
+        consec_score = 50.0
+        consec_up = 0
+        consec_down = 0
+        for i in range(len(closes) - 1, max(0, len(closes) - 10), -1):
+            if closes.iloc[i] > closes.iloc[i - 1]:
+                consec_up += 1
+                consec_down = 0
+            elif closes.iloc[i] < closes.iloc[i - 1]:
+                consec_down += 1
+                consec_up = 0
+            else:
+                break
+
+        if consec_up >= CONSECUTIVE_CANDLES_SIGNAL:
+            consec_score = 70 + min(consec_up * 5, 25)
+        elif consec_down >= CONSECUTIVE_CANDLES_SIGNAL:
+            consec_score = 30 - min(consec_down * 5, 25)
+
+        return roc_score * 0.70 + consec_score * 0.30
+
+    def _build_narrative(self, trend: float, vol: float, momentum: float,
+                         latest: pd.Series) -> str:
+        parts = []
+        if trend > 65:
+            parts.append("EMA多头排列+ADX趋势明确")
+        elif trend > 50:
+            parts.append("趋势温和偏多")
+        elif trend < 35:
+            parts.append("EMA空头排列+ADX趋势明确")
+        elif trend < 50:
+            parts.append("趋势温和偏空")
+        else:
+            parts.append("趋势中性")
+
+        if vol > 70:
+            parts.append("波动率扩张")
+        elif vol < 45:
+            parts.append("波动率压缩(突破前兆)")
+
+        if momentum > 65:
+            parts.append("动量强劲")
+        elif momentum < 35:
+            parts.append("动量疲弱")
+
+        return ", ".join(parts) if parts else "中性"
+
+    @staticmethod
+    def _label(score: float) -> str:
+        if score >= 75:
+            return "Strong Bullish Structure"
+        elif score >= 60:
+            return "Bullish Structure"
+        elif score >= 40:
+            return "Neutral Structure"
+        elif score >= 25:
+            return "Bearish Structure"
+        return "Weak Bearish Structure"

ChanMacro/scoring/volatility_regime.py

@@ -0,0 +1,143 @@
+"""
+scoring/volatility_regime.py — Volatility Regime Classification.
+
+4 regimes from OHLCV data:
+  LOW_VOL:       ATR/Close < 2% → compression, breakout imminent
+  NORMAL_VOL:    ATR/Close 2-5% → normal trading
+  HIGH_VOL:      ATR/Close 5-10% → trend acceleration, wider stops
+  EXPLOSIVE_VOL: ATR/Close > 10% → extreme, reduce or wait
+
+Uses: ATR(14)/Close, HV(20)/HV(60) ratio, BB width ratio.
+OHLCV-only — never goes offline.
+"""
+
+from datetime import date as Date
+import sqlite3
+import numpy as np
+import pandas as pd
+
+from .base import BaseScorer
+from .constants import (
+    VOL_LOW, VOL_HIGH, VOL_REGIME_SCORES, HV_RATIO_LOW, HV_RATIO_HIGH,
+)
+from models import FactorScore, VolatilityRegimeScore, VolRegime, MacroDirection
+from config import config
+
+
+class VolatilityRegimeScorer(BaseScorer):
+    """Classifies volatility regime from OHLCV data."""
+
+    def compute(self, target_date: Date) -> VolatilityRegimeScore:
+        conn = self.get_connection()
+        try:
+            df = pd.read_sql_query(
+                "SELECT * FROM ohlcv_daily WHERE date <= ? ORDER BY date DESC LIMIT 120",
+                conn, params=(str(target_date),)
+            )
+            if df.empty:
+                return VolatilityRegimeScore(
+                    name="Volatility Regime",
+                    score=50.0,
+                    label="No Data",
+                )
+
+            df = df.sort_values("date").reset_index(drop=True)
+
+            # 1. ATR/Close %
+            latest = df.iloc[-1]
+            atr = latest.get("atr_14")
+            close = float(latest["close"])
+            atr_pct = (atr / close * 100) if atr and not pd.isna(atr) and close > 0 else 3.0
+
+            # 2. HV(20) / HV(60) ratio
+            hv_ratio = self._compute_hv_ratio(df)
+
+            # 3. BB width ratio
+            bb_ratio = self._compute_bb_ratio(df)
+
+            # Classify regime
+            regime = self._classify(atr_pct, hv_ratio, bb_ratio)
+
+            # Score
+            score = VOL_REGIME_SCORES.get(regime.value, 50)
+
+            # Narrative
+            narrative = self._build_narrative(regime, atr_pct, hv_ratio, bb_ratio)
+
+            return VolatilityRegimeScore(
+                name="Volatility Regime",
+                score=float(score),
+                label=regime.value,
+                direction=MacroDirection.NEUTRAL,
+                vol_regime=regime,
+                atr_pct=round(atr_pct, 2),
+                hv_ratio=round(hv_ratio, 2),
+                bb_width_ratio=round(bb_ratio, 2),
+                sub_scores={
+                    "atr_pct": round(atr_pct, 2),
+                    "hv_ratio": round(hv_ratio, 2),
+                    "bb_width_ratio": round(bb_ratio, 2),
+                },
+                narrative=narrative,
+            )
+        finally:
+            conn.close()
+
+    def _compute_hv_ratio(self, df: pd.DataFrame) -> float:
+        """Compute HV(20) / HV(60) ratio."""
+        closes = df["close"].astype(float)
+        returns = closes.pct_change().dropna()
+
+        if len(returns) < 60:
+            return 1.0
+
+        hv20 = returns.tail(20).std() * np.sqrt(365) * 100
+        hv60 = returns.tail(60).std() * np.sqrt(365) * 100
+
+        if hv60 == 0:
+            return 1.0
+
+        return hv20 / hv60
+
+    def _compute_bb_ratio(self, df: pd.DataFrame) -> float:
+        """Compute current BB width / 20d average BB width."""
+        bb_widths = df["bb_width"].dropna().tail(40)
+        if len(bb_widths) < 20:
+            return 1.0
+
+        current = bb_widths.iloc[-1]
+        avg = bb_widths.tail(20).mean()
+        if avg == 0:
+            return 1.0
+
+        return current / avg
+
+    @staticmethod
+    def _classify(atr_pct: float, hv_ratio: float, bb_ratio: float) -> VolRegime:
+        """Classify volatility regime from multiple indicators."""
+        # Primary: ATR/Close %
+        if atr_pct > 10.0:
+            return VolRegime.EXPLOSIVE_VOL
+        elif atr_pct > VOL_HIGH:
+            return VolRegime.HIGH_VOL
+        elif atr_pct < VOL_LOW:
+            return VolRegime.LOW_VOL
+
+        # Secondary: HV ratio and BB ratio for edge cases
+        if hv_ratio > HV_RATIO_HIGH and bb_ratio > 1.3:
+            return VolRegime.HIGH_VOL
+        elif hv_ratio < HV_RATIO_LOW and bb_ratio < 0.8:
+            return VolRegime.LOW_VOL
+
+        return VolRegime.NORMAL_VOL
+
+    @staticmethod
+    def _build_narrative(regime: VolRegime, atr_pct: float,
+                         hv_ratio: float, bb_ratio: float) -> str:
+        mapping = {
+            VolRegime.LOW_VOL: f"低波动(ATR={atr_pct:.1f}%), 布林带收窄, 突破前兆",
+            VolRegime.NORMAL_VOL: f"正常波动(ATR={atr_pct:.1f}%), 正常交易环境",
+            VolRegime.HIGH_VOL: f"高波动(ATR={atr_pct:.1f}%), 趋势加速, 放宽止损",
+            VolRegime.EXPLOSIVE_VOL: f"极端波动(ATR={atr_pct:.1f}%), 减仓或等待",
+        }
+        return mapping.get(regime, "Unknown")

ChanMacro/tests/conftest.py

@@ -0,0 +1,134 @@
+"""
+tests/conftest.py — Shared fixtures for ChanMacro tests.
+"""
+
+import os
+import sys
+import pytest
+import sqlite3
+import numpy as np
+import pandas as pd
+from datetime import date, timedelta
+from pathlib import Path
+
+# Ensure package root on path
+sys.path.insert(0, str(Path(__file__).parent.parent))
+
+
+@pytest.fixture
+def db_path(tmp_path):
+    """Create a temporary SQLite database with full mock data."""
+    db = str(tmp_path / "test_macro.db")
+    from database import init_db
+    conn = init_db(db)
+
+    np.random.seed(42)
+    base = date(2025, 9, 1)
+    n_days = 300
+
+    # Generate realistic price series with 3 regime periods
+    prices = [90000]
+    regimes = []
+    for i in range(n_days):
+        if i < 100:
+            ret = np.random.normal(0.003, 0.015)
+            regime = "TREND"
+        elif i < 200:
+            ret = np.random.normal(0.000, 0.012)
+            regime = "RANGE"
+        else:
+            ret = np.random.normal(-0.003, 0.025)
+            regime = "PANIC"
+        prices.append(prices[-1] * (1 + ret))
+        regimes.append(regime)
+
+    for i in range(n_days):
+        d = base + timedelta(days=i)
+        c = prices[i]
+        r = regimes[i]
+
+        # OHLCV
+        conn.execute("""
+            INSERT OR REPLACE INTO ohlcv_daily
+            (date,symbol,open,high,low,close,volume,ema20,ema60,ema120,atr_14,bb_width,adx_14)
+            VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)
+        """, (
+            d.strftime("%Y-%m-%d"), "BTC/USDT:USDT",
+            c * 0.99, c * 1.03, c * 0.97, c, 1000,
+            c * (0.98 if r == "TREND" else 1.02 if r == "PANIC" else 1.0),
+            c * (0.95 if r == "TREND" else 1.05 if r == "PANIC" else 1.0),
+            c * (0.90 if r == "TREND" else 1.10 if r == "PANIC" else 1.0),
+            c * (0.02 if r == "PANIC" else 0.015),
+            4.5, 28.0 if r == "TREND" else 18.0,
+        ))
+
+        # Breadth
+        adv = 42 if r == "TREND" else 25 if r == "RANGE" else 8
+        conn.execute("""
+            INSERT OR REPLACE INTO breadth_daily
+            (date,total_tracked,advance_top50,decline_top50,above_ema20_top50,
+             new_highs_20d_top50,advance_top30,advance_top20,
+             above_ema20_top30,above_ema20_top20,new_highs_20d_top30,new_highs_20d_top20)
+            VALUES (?,50,?,?,?,?,?,?,?,?,?,?)
+        """, (
+            d.strftime("%Y-%m-%d"), adv, 50 - adv, adv, min(adv, 15),
+            int(adv * 0.7), int(adv * 0.5), int(adv * 0.7), int(adv * 0.5),
+            min(int(adv * 0.7), 12), min(int(adv * 0.5), 8),
+        ))
+
+        # Derivatives
+        oi_chg = 3.5 if r == "TREND" else 0.5 if r == "RANGE" else -2.0
+        conn.execute("""
+            INSERT OR REPLACE INTO derivatives
+            (date,symbol,funding_rate,open_interest,oi_24h_change_pct,
+             long_liquidations,short_liquidations,basis_annualised_pct)
+            VALUES (?,?,?,?,?,?,?,?)
+        """, (
+            d.strftime("%Y-%m-%d"), "BTC/USDT:USDT",
+            0.0001 + np.random.normal(0, 0.0002),
+            35e9, oi_chg + np.random.normal(0, 1.0),
+            50e6 * np.random.random(), 30e6 * np.random.random(),
+            8.5 if r == "TREND" else 3.0,
+        ))
+
+        # Regime history
+        conn.execute("""
+            INSERT OR REPLACE INTO regime_history
+            (date,regime,confidence,regime_version,maturity_score,all_scores_json,confirmation_days)
+            VALUES (?,?,?,?,?,?,?)
+        """, (d.strftime("%Y-%m-%d"), r, 0.75, "v1_price_breadth_vol", 50, "{}", 1))
+
+    conn.commit()
+    conn.close()
+
+    # Override config to use test DB
+    from config import config
+    old_db = config.db_path
+    config.db_path = db
+    yield db
+    config.db_path = old_db
+
+
+@pytest.fixture
+def sample_state(db_path):
+    """Build a MarketStateVector for a known test date."""
+    from models import (
+        MarketStateVector, MarketRegime, BreadthBucket,
+        OIState, VolRegime,
+    )
+    state = MarketStateVector(
+        date=date(2026, 3, 15),
+        regime=MarketRegime.TREND,
+        regime_confidence=0.82,
+        regime_version="v1_price_breadth_vol",
+        regime_maturity_score=55.0,
+        breadth_top20=82.0,
+        breadth_top30=78.0,
+        breadth_top50=74.0,
+        breadth_bucket=BreadthBucket.STRONG,
+        breadth_divergence=8.0,
+        oi_state=OIState.NEW_LONGS,
+        volatility_regime=VolRegime.NORMAL_VOL,
+    )
+    state.market_state_hash = state.compute_hash()
+    return state

ChanMacro/tests/test_expectancy.py

@@ -0,0 +1,173 @@
+"""Test SignalTracker, TimeDecay, and BayesianExpectancyEngine."""
+import pytest
+from datetime import date, timedelta
+import numpy as np
+
+
+class TestTimeDecay:
+    def test_recent_weight_near_one(self):
+        from expectancy.decay import TimeDecay
+        d = TimeDecay(180)
+        w = d.weight(date(2026, 6, 20), date(2026, 6, 24))
+        assert 0.95 < w < 1.0
+
+    def test_old_weight_decays(self):
+        from expectancy.decay import TimeDecay
+        d = TimeDecay(180)
+        w = d.weight(date(2025, 6, 24), date(2026, 6, 24))
+        assert 0.2 < w < 0.3  # ~365 days at half_life=180
+
+    def test_effective_samples(self):
+        from expectancy.decay import TimeDecay
+        d = TimeDecay(180)
+        dates = [date(2026, 6, 24)] * 10
+        weights = d.weights(dates, date(2026, 6, 24))
+        eff = d.effective_samples(weights)
+        assert eff == pytest.approx(10.0, rel=0.01)
+
+    def test_weighted_win_rate(self):
+        from expectancy.decay import TimeDecay
+        d = TimeDecay(180)
+        wins = np.array([1, 0, 1, 0])
+        weights = np.array([1.0, 1.0, 1.0, 1.0])
+        wr = d.weighted_win_rate(wins, weights)
+        assert wr == 0.5
+
+    def test_weight_at_age(self):
+        from expectancy.decay import TimeDecay
+        w = TimeDecay.weight_at_age(180, 180)
+        assert w == pytest.approx(0.5, rel=0.01)
+
+
+class TestSignalTracker:
+    def test_record_signal(self, db_path, sample_state):
+        from expectancy.tracker import SignalTracker
+        tracker = SignalTracker()
+        rid = tracker.record(
+            date(2026, 3, 15), "B3", 98000.0, sample_state,
+            signal_grade="A", signal_strength=75.0,
+        )
+        assert rid is not None
+        assert rid > 0
+
+    def test_get_samples(self, db_path, sample_state):
+        from expectancy.tracker import SignalTracker
+        tracker = SignalTracker()
+        tracker.record(date(2026, 3, 15), "B3", 98000.0, sample_state)
+        tracker.record(date(2026, 3, 16), "B2", 98500.0, sample_state)
+
+        samples = tracker.get_samples(signal_type="B3")
+        assert len(samples) == 1
+        assert samples[0]["signal_type"] == "B3"
+
+    def test_count_samples(self, db_path, sample_state):
+        from expectancy.tracker import SignalTracker
+        tracker = SignalTracker()
+        tracker.record(date(2026, 3, 15), "B3", 98000.0, sample_state)
+        tracker.record(date(2026, 3, 16), "B3", 98500.0, sample_state)
+
+        counts = tracker.count_samples()
+        assert "B3/TREND" in counts
+        assert counts["B3/TREND"] == 2
+
+    def test_filter_by_regime(self, db_path, sample_state):
+        from expectancy.tracker import SignalTracker
+        tracker = SignalTracker()
+        tracker.record(date(2026, 3, 15), "B3", 98000.0, sample_state)
+
+        samples = tracker.get_samples(signal_type="B3", regime="TREND")
+        assert len(samples) == 1
+
+        samples = tracker.get_samples(signal_type="B3", regime="PANIC")
+        assert len(samples) == 0
+
+    def test_backfill_signals(self, db_path, sample_state):
+        from expectancy.tracker import SignalTracker
+        tracker = SignalTracker()
+        signals = [
+            {"date": date(2026, 3, 15), "signal_type": "B3", "entry_price": 98000},
+            {"date": date(2026, 3, 20), "signal_type": "B2", "entry_price": 99000},
+        ]
+        count = tracker.backfill_signals(signals)
+        assert count == 2
+
+
+class TestBayesianExpectancyEngine:
+    def test_estimate_returns_report(self, db_path, sample_state):
+        from expectancy.tracker import SignalTracker
+        from expectancy.engine import BayesianExpectancyEngine
+
+        # Record some signals first
+        tracker = SignalTracker()
+        for i in range(10):
+            tracker.record(
+                date(2026, 3, 15) + timedelta(days=i),
+                "B3", 98000.0, sample_state,
+            )
+
+        engine = BayesianExpectancyEngine(level_min_samples=3)
+        report = engine.estimate(sample_state, "B3", date(2026, 3, 25))
+        assert report.signal_type == "B3"
+        assert len(report.layers) > 0
+        assert report.source in ("bayesian", "insufficient")
+
+    def test_insufficient_with_no_samples(self, db_path, sample_state):
+        from expectancy.engine import BayesianExpectancyEngine
+        engine = BayesianExpectancyEngine(level_min_samples=10)
+        report = engine.estimate(sample_state, "B1", date(2026, 3, 25))
+        assert report.sufficiency.value in ("INSUFFICIENT", "LOW", "MEDIUM", "HIGH")
+
+    def test_empirical_bayes_shrinks_small_samples(self, db_path, sample_state):
+        """With N=3, raw=100%, posterior should be pulled toward prior."""
+        from expectancy.tracker import SignalTracker
+        from expectancy.engine import BayesianExpectancyEngine
+
+        tracker = SignalTracker()
+        for i in range(3):
+            tracker.record(
+                date(2026, 3, 15) + timedelta(days=i),
+                "B3", 98000.0, sample_state,
+            )
+
+        engine = BayesianExpectancyEngine(level_min_samples=1)
+        report = engine.estimate(sample_state, "B3", date(2026, 3, 25))
+
+        # With small N, posterior should differ from raw
+        base_layer = report.layers[0]
+        if base_layer.raw_winrate and base_layer.samples < 50:
+            # Posterior should be pulled toward prior (50% or global rate)
+            if base_layer.raw_winrate > 0.8:
+                assert base_layer.posterior_winrate < base_layer.raw_winrate
+
+    def test_leveled_fallback_stops_at_min_samples(self, db_path, sample_state):
+        from expectancy.tracker import SignalTracker
+        from expectancy.engine import BayesianExpectancyEngine
+
+        tracker = SignalTracker()
+        for i in range(20):
+            tracker.record(date(2026, 3, 15) + timedelta(days=i), "B3", 98000.0, sample_state)
+
+        engine = BayesianExpectancyEngine(level_min_samples=15)
+        report = engine.estimate(sample_state, "B3", date(2026, 3, 25))
+        # Should have stopped at a level with >= 15 effective samples
+        assert report.final_estimate >= 0
+
+
+class TestSufficiencyGuard:
+    def test_insufficient(self):
+        from expectancy.engine import SufficiencyGuard
+        from models import SufficiencyLevel
+        g = SufficiencyGuard()
+        assert g.evaluate(10) == SufficiencyLevel.INSUFFICIENT
+
+    def test_low(self):
+        from expectancy.engine import SufficiencyGuard
+        from models import SufficiencyLevel
+        g = SufficiencyGuard()
+        assert g.evaluate(40) == SufficiencyLevel.LOW
+
+    def test_high(self):
+        from expectancy.engine import SufficiencyGuard
+        from models import SufficiencyLevel
+        g = SufficiencyGuard()
+        assert g.evaluate(200) == SufficiencyLevel.HIGH

ChanMacro/tests/test_models.py

@@ -0,0 +1,130 @@
+"""Test all Pydantic models and enums."""
+import pytest
+from datetime import date
+from models import (
+    MarketRegime, OIState, BreadthBucket, VolRegime,
+    MarketStateVector, FactorScore, RegimeResult,
+    SignalFeatureRecord, ExpectancyReport, DailyOutput,
+    FactorContribution, SufficiencyLevel, SignalGrade,
+)
+
+
+class TestEnums:
+    def test_regime_values(self):
+        assert MarketRegime.TREND.value == "TREND"
+        assert MarketRegime.RANGE.value == "RANGE"
+        assert MarketRegime.PANIC.value == "PANIC"
+
+    def test_oi_state_has_neutral(self):
+        assert OIState.NEUTRAL.value == "Neutral"
+        assert len(OIState) == 5
+
+    def test_breadth_bucket_values(self):
+        assert BreadthBucket.EXTREME.value == "EXTREME"
+        assert len(BreadthBucket) == 5
+
+    def test_vol_regime_values(self):
+        assert VolRegime.LOW_VOL.value == "LOW_VOL"
+        assert VolRegime.EXPLOSIVE_VOL.value == "EXPLOSIVE_VOL"
+
+
+class TestMarketStateVector:
+    def test_minimal_construction(self):
+        sv = MarketStateVector(
+            date="2026-06-24",
+            regime=MarketRegime.TREND,
+            regime_confidence=0.82,
+            regime_version="v1_price_breadth_vol",
+        )
+        assert sv.date == date(2026, 6, 24)
+        assert sv.regime == MarketRegime.TREND
+        assert sv.breadth_top50 == 50.0  # default
+
+    def test_date_string_parsing(self):
+        sv = MarketStateVector(
+            date="2026-01-15",
+            regime=MarketRegime.RANGE,
+            regime_confidence=0.55,
+            regime_version="v1_price_breadth_vol",
+        )
+        assert sv.date == date(2026, 1, 15)
+
+    def test_compute_hash(self):
+        sv = MarketStateVector(
+            date="2026-06-24",
+            regime=MarketRegime.TREND,
+            regime_confidence=0.82,
+            regime_version="v1_price_breadth_vol",
+            breadth_bucket=BreadthBucket.EXTREME,
+            oi_state=OIState.NEW_LONGS,
+            volatility_regime=VolRegime.NORMAL_VOL,
+        )
+        h = sv.compute_hash()
+        assert len(h) == 12
+        # Same state = same hash
+        sv2 = MarketStateVector(
+            date="2026-06-25",
+            regime=MarketRegime.TREND,
+            regime_confidence=0.80,
+            regime_version="v1_price_breadth_vol",
+            breadth_bucket=BreadthBucket.EXTREME,
+            oi_state=OIState.NEW_LONGS,
+            volatility_regime=VolRegime.NORMAL_VOL,
+        )
+        assert sv2.compute_hash() == h
+
+    def test_state_embedding(self):
+        sv = MarketStateVector(
+            date="2026-06-24",
+            regime=MarketRegime.TREND,
+            regime_confidence=0.82,
+            regime_version="v1_price_breadth_vol",
+            breadth_top20=80.0,
+            breadth_top30=75.0,
+            breadth_top50=70.0,
+            regime_maturity_score=60.0,
+        )
+        emb = sv.state_embedding()
+        assert len(emb) == 5
+        assert emb[0] == 80.0
+        assert emb[3] == 60.0
+
+
+class TestRegimeResult:
+    def test_construction(self):
+        r = RegimeResult(
+            date="2026-06-24",
+            regime=MarketRegime.TREND,
+            confidence=0.82,
+            regime_version="v1_price_breadth_vol",
+            maturity_score=55.0,
+            all_scores={"TREND": 82.0, "RANGE": 45.0, "PANIC": 20.0},
+            confirmation_days=5,
+        )
+        assert r.regime == MarketRegime.TREND
+        assert r.confirmation_days == 5
+
+
+class TestExpectancyReport:
+    def test_insufficient(self):
+        r = ExpectancyReport(
+            signal_type="B3",
+            date="2026-06-24",
+            final_estimate=0.0,
+            sufficiency=SufficiencyLevel.INSUFFICIENT,
+            source="insufficient",
+        )
+        assert r.final_estimate == 0.0
+        assert r.sufficiency == SufficiencyLevel.INSUFFICIENT
+
+
+class TestFactorContribution:
+    def test_construction(self):
+        fc = FactorContribution(
+            factor="ETF Flow",
+            raw_score=85.0,
+            weight=0.1925,
+            impact=6.7,
+            direction="bullish",
+        )
+        assert fc.impact > 0

ChanMacro/tests/test_regime.py

@@ -0,0 +1,109 @@
+"""Test regime detector and validation."""
+import pytest
+from datetime import date
+import pandas as pd
+import numpy as np
+
+
+class TestRegimeDetector:
+    def test_detects_trend(self):
+        from regime_detector import RegimeDetector
+        from models import MarketRegime
+        d = RegimeDetector()
+        r = d.detect(75.0, 80.0, "NORMAL_VOL", date(2026, 6, 24))
+        assert r.regime == MarketRegime.TREND
+        assert r.confidence > 0.5
+
+    def test_detects_range(self):
+        from regime_detector import RegimeDetector
+        from models import MarketRegime
+        d = RegimeDetector()
+        r = d.detect(50.0, 50.0, "LOW_VOL", date(2026, 6, 24))
+        assert r.regime in (MarketRegime.RANGE, MarketRegime.TREND)
+
+    def test_detects_panic(self):
+        from regime_detector import RegimeDetector
+        from models import MarketRegime
+        d = RegimeDetector()
+        r = d.detect(15.0, 10.0, "EXPLOSIVE_VOL", date(2026, 6, 24))
+        assert r.regime == MarketRegime.PANIC
+
+    def test_2day_confirmation(self):
+        from regime_detector import RegimeDetector
+        from models import MarketRegime
+        d = RegimeDetector()
+        # Day 1: RANGE
+        r1 = d.detect(50.0, 50.0, "LOW_VOL", date(2026, 6, 24))
+        assert r1.regime == MarketRegime.RANGE  # first run, no confirmation needed
+        # Day 2: still RANGE
+        r2 = d.detect(50.0, 50.0, "LOW_VOL", date(2026, 6, 25))
+        assert r2.regime == MarketRegime.RANGE
+        assert r2.confirmation_days == 2
+
+    def test_transition_needs_confirmation(self):
+        from regime_detector import RegimeDetector
+        from models import MarketRegime
+        d = RegimeDetector()
+        # Establish TREND
+        d.detect(75.0, 80.0, "NORMAL_VOL", date(2026, 6, 24))
+        d.detect(75.0, 80.0, "NORMAL_VOL", date(2026, 6, 25))
+        # Day 3: weak scores → raw best = RANGE, but TREND should persist
+        r3 = d.detect(35.0, 40.0, "NORMAL_VOL", date(2026, 6, 26))
+        # First day of pending transition — should still be TREND
+        assert r3.regime == MarketRegime.TREND
+        assert d.pending_regime is not None
+
+    def test_version_is_stored(self):
+        from regime_detector import RegimeDetector
+        d = RegimeDetector(regime_version="v1_price_breadth_vol")
+        r = d.detect(75.0, 80.0, "NORMAL_VOL", date(2026, 6, 24))
+        assert r.regime_version == "v1_price_breadth_vol"
+
+    def test_load_state(self, db_path):
+        from regime_detector import RegimeDetector
+        from models import MarketRegime
+        d = RegimeDetector()
+        d.load_state(db_path)
+        # DB has TREND for first 100 days, so most recent should load
+        assert d.current_regime is not None
+
+    def test_confidence_for_confirmed_regime(self):
+        """Confidence should be for the confirmed regime, not raw best."""
+        from regime_detector import RegimeDetector
+        from models import MarketRegime
+        d = RegimeDetector()
+        # Establish TREND
+        d.detect(75.0, 80.0, "NORMAL_VOL", date(2026, 6, 24))
+        d.detect(75.0, 80.0, "NORMAL_VOL", date(2026, 6, 25))
+        # Now feed weak scores → raw best would be PANIC or RANGE
+        r = d.detect(15.0, 10.0, "EXPLOSIVE_VOL", date(2026, 6, 26))
+        # Should still report TREND (need 2 confirmations to switch)
+        assert r.regime == MarketRegime.TREND
+
+
+class TestTransitionValidator:
+    def test_stable_regime_passes(self):
+        from validation.transition_validator import TransitionValidator
+        # Create stable regime sequence: long periods
+        seq = pd.Series(
+            ["TREND"] * 50 + ["RANGE"] * 50 + ["PANIC"] * 40,
+            index=pd.date_range("2026-01-01", periods=140),
+        )
+        tv = TransitionValidator()
+        report = tv.validate(seq)
+        assert report.is_stable
+        assert report.avg_duration > 20
+        assert report.flip_rate < 0.05
+
+    def test_unstable_regime_fails(self):
+        from validation.transition_validator import TransitionValidator
+        # Create unstable sequence: flips every 2 days
+        seq = pd.Series(
+            ["TREND", "TREND", "RANGE", "RANGE", "TREND", "TREND",
+             "PANIC", "PANIC", "RANGE", "RANGE"] * 5,
+            index=pd.date_range("2026-01-01", periods=50),
+        )
+        tv = TransitionValidator()
+        report = tv.validate(seq)
+        assert not report.is_stable
+        assert report.flip_rate > 0.15

ChanMacro/tests/test_scoring.py

@@ -0,0 +1,121 @@
+"""Test all 4 core scorers."""
+import pytest
+from datetime import date
+
+
+class TestPriceStructureScorer:
+    def test_computes_score(self, db_path):
+        from scoring.price_structure import PriceStructureScorer
+        scorer = PriceStructureScorer()
+        result = scorer.compute(date(2026, 3, 15))
+        assert result.name == "Price Structure"
+        assert 0 <= result.score <= 100
+        assert result.trend_strength >= 0
+        assert result.volatility_compression >= 0
+        assert result.momentum >= 0
+        assert result.label
+
+    def test_bullish_in_trend(self, db_path):
+        from scoring.price_structure import PriceStructureScorer
+        scorer = PriceStructureScorer()
+        result = scorer.compute(date(2025, 11, 15))  # TREND period
+        assert result.score > 50  # Should be bullish in uptrend
+
+    def test_bearish_in_panic(self, db_path):
+        from scoring.price_structure import PriceStructureScorer
+        scorer = PriceStructureScorer()
+        result = scorer.compute(date(2026, 5, 15))  # PANIC period
+        # In panic period, EMA alignment should be bearish
+        assert result.trend_strength < 60
+
+    def test_no_data_handling(self, db_path):
+        from scoring.price_structure import PriceStructureScorer
+        scorer = PriceStructureScorer()
+        result = scorer.compute(date(2020, 1, 1))
+        assert result.score == 50.0
+        assert result.label == "No Data"
+
+
+class TestBreadthScorer:
+    def test_computes_score(self, db_path):
+        from scoring.breadth_scorer import BreadthScorer
+        scorer = BreadthScorer()
+        result = scorer.compute(date(2026, 3, 15))
+        assert result.name == "Breadth"
+        assert 0 <= result.score <= 100
+        assert result.breadth_bucket
+        assert result.breadth_top20 >= 0
+        assert result.breadth_top50 >= 0
+
+    def test_tier_values(self, db_path):
+        from scoring.breadth_scorer import BreadthScorer
+        scorer = BreadthScorer()
+        result = scorer.compute(date(2025, 11, 15))  # TREND period
+        # Top20 should generally be higher than Top50 (large caps lead)
+        assert result.breadth_top20 >= 0
+        assert result.breadth_top50 >= 0
+
+    def test_bucket_assignment(self, db_path):
+        from scoring.breadth_scorer import BreadthScorer, BreadthBucket
+        scorer = BreadthScorer()
+        result = scorer.compute(date(2025, 11, 15))  # TREND: adv=42/50
+        assert result.breadth_bucket in (
+            BreadthBucket.EXTREME, BreadthBucket.STRONG, BreadthBucket.NORMAL
+        )
+
+    def test_no_data(self, db_path):
+        from scoring.breadth_scorer import BreadthScorer
+        scorer = BreadthScorer()
+        result = scorer.compute(date(2020, 1, 1))
+        assert result.score == 50.0
+
+
+class TestOIMatrixScorer:
+    def test_computes_state(self, db_path):
+        from scoring.oi_matrix import OIMatrixScorer, OIState
+        scorer = OIMatrixScorer()
+        result = scorer.compute(date(2025, 11, 15))  # TREND period, oi_chg=+3.5
+        assert result.oi_state in OIState
+        assert 0 <= result.score <= 100
+
+    def test_new_longs_in_trend(self, db_path):
+        from scoring.oi_matrix import OIMatrixScorer, OIState
+        scorer = OIMatrixScorer()
+        # Test multiple dates in TREND period — at least one should be NEW_LONGS or NEUTRAL
+        found_bullish = False
+        for d in ["2025-11-15", "2025-11-20", "2025-12-01", "2025-12-15"]:
+            result = scorer.compute(date.fromisoformat(d))
+            if result.oi_state in (OIState.NEW_LONGS, OIState.SHORT_COVERING, OIState.NEUTRAL):
+                found_bullish = True
+                break
+        assert found_bullish, "No bullish OI state found in TREND period"
+
+    def test_no_data(self, db_path):
+        from scoring.oi_matrix import OIMatrixScorer
+        scorer = OIMatrixScorer()
+        result = scorer.compute(date(2020, 1, 1))
+        assert result.score == 50.0
+        assert result.label == "No Data"
+
+
+class TestVolatilityRegimeScorer:
+    def test_computes_regime(self, db_path):
+        from scoring.volatility_regime import VolatilityRegimeScorer, VolRegime
+        scorer = VolatilityRegimeScorer()
+        result = scorer.compute(date(2026, 3, 15))
+        assert result.vol_regime in VolRegime
+        assert 0 <= result.score <= 100
+
+    def test_higher_vol_in_panic(self, db_path):
+        from scoring.volatility_regime import VolatilityRegimeScorer, VolRegime
+        scorer = VolatilityRegimeScorer()
+        trend_result = scorer.compute(date(2025, 11, 15))
+        panic_result = scorer.compute(date(2026, 5, 15))
+        # PANIC period has higher ATR → higher vol regime or score
+        assert panic_result.atr_pct >= trend_result.atr_pct * 0.5  # at least comparable
+
+    def test_no_data(self, db_path):
+        from scoring.volatility_regime import VolatilityRegimeScorer
+        scorer = VolatilityRegimeScorer()
+        result = scorer.compute(date(2020, 1, 1))
+        assert result.score == 50.0

ChanMacro/trend_detector.py

@@ -0,0 +1,81 @@
+"""
+trend_detector.py — Trend strength and maturity helpers.
+
+Utility functions for computing trend alignment, acceleration, persistence.
+Used by regime_detector and price_structure scorer.
+"""
+
+import numpy as np
+import pandas as pd
+
+
+def ema_alignment_score(close: float, ema20: float, ema60: float, ema120: float) -> float:
+    """Score EMA alignment: 0=bearish, 50=neutral, 100=bullish."""
+    if any(pd.isna(x) for x in [ema20, ema60, ema120]):
+        return 50.0
+
+    alignments = 0
+    if ema20 > ema60:
+        alignments += 1
+    if ema60 > ema120:
+        alignments += 1
+    if ema20 > ema120:
+        alignments += 1
+
+    if alignments == 3:
+        return 85.0
+    elif alignments == 2:
+        return 65.0
+    elif alignments == 1:
+        return 35.0
+    else:
+        return 15.0
+
+
+def adx_trend_score(adx: float) -> float:
+    """Convert ADX value to trend score: 0-100."""
+    if pd.isna(adx):
+        return 50.0
+    if adx > 40:
+        return 90.0
+    elif adx > 25:
+        return 60.0 + (adx - 25) / 15 * 30
+    elif adx > 15:
+        return 40.0 + (adx - 15) / 10 * 20
+    else:
+        return max(10.0, adx / 15 * 40)
+
+
+def breadth_persistence(breadth_scores: list[float], window: int = 5) -> float:
+    """How consistently has breadth stayed at its current level? 0-100."""
+    if len(breadth_scores) < window:
+        return 50.0
+    recent = breadth_scores[-window:]
+    mean_val = np.mean(recent)
+    std_val = np.std(recent) if len(recent) > 1 else 0
+    # Low std = high persistence
+    persistence = 100 - min(std_val * 5, 100)
+    # Bias: higher breadth = higher persistence score
+    return persistence * 0.5 + mean_val * 0.5
+
+
+def trend_strength_composite(ema_score: float, adx_score: float,
+                              breadth_score: float) -> float:
+    """Composite trend strength 0-100."""
+    return ema_score * 0.25 + adx_score * 0.25 + breadth_score * 0.50
+
+
+def compute_maturity(trend_strength: float, breadth_persistence: float,
+                     vol_expansion: float) -> float:
+    """
+    Compute regime maturity score 0-100.
+
+    EMERGING (0-30): trend accelerating, breadth expanding
+    CONFIRMED (30-70): trend stable, breadth stable
+    EXHAUSTING (70-100): trend decelerating, breadth contracting, vol abnormal
+    """
+    return (
+        trend_strength * 0.50 +
+        breadth_persistence * 0.30 +
+        (100 - vol_expansion) * 0.20  # inverted: low vol = early stage
+    )

ChanMacro/validation/__init__.py

@@ -0,0 +1,5 @@
+"""Validation Framework — Phase 0: verify every factor before trusting it."""
+from .factor_validator import FactorValidator
+from .regime_validator import RegimeValidator
+from .transition_validator import TransitionValidator
+from .reporter import ValidationReporter

ChanMacro/validation/factor_validator.py

@@ -0,0 +1,174 @@
+"""
+validation/factor_validator.py — Validates a factor's predictive power.
+
+Tests: IC, ICIR, Hit Ratio, Quantile Spread, Lead-Lag analysis.
+Answers: "Does this factor predict future returns?"
+"""
+
+from datetime import date as Date
+from typing import Optional
+import sqlite3
+import logging
+
+import numpy as np
+import pandas as pd
+
+from config import config
+from .metrics import (
+    information_coefficient, icir, hit_ratio,
+    quantile_spread, lead_lag_ic,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class FactorReport:
+    """Structured report for a single factor's validation results."""
+
+    def __init__(self, factor_name: str):
+        self.factor_name = factor_name
+        self.ic_mean: float = 0.0
+        self.ic_std: float = 0.0
+        self.icir: float = 0.0
+        self.hit_ratio: float = 0.0
+        self.quantile_spread: float = 0.0
+        self.is_leading: bool = False
+        self.lead_days: int = 0
+        self.lead_ic: float = 0.0
+        self.n_observations: int = 0
+        self.conclusion: str = ""
+
+    def summary(self) -> str:
+        lines = [
+            f"Factor: {self.factor_name}",
+            f"  N={self.n_observations}",
+            f"  IC mean={self.ic_mean:.4f}  std={self.ic_std:.4f}  ICIR={self.icir:.2f}",
+            f"  Hit Ratio={self.hit_ratio:.1%}  Top-Bot Spread={self.quantile_spread:.4f}",
+            f"  Best Lead: {self.lead_days}d (IC={self.lead_ic:.4f})" if self.is_leading else "  Leading: No (synchronous/lagging)",
+            f"  → {self.conclusion}",
+        ]
+        return "\n".join(lines)
+
+
+class FactorValidator:
+    """
+    Validates a factor's predictive power using standard quant metrics.
+
+    For each forward horizon (1d, 3d, 5d, 7d, 14d), computes:
+      - IC (Spearman rank correlation)
+      - ICIR (IC stability)
+      - Hit Ratio (direction accuracy)
+      - Quantile spread (top vs bottom bucket)
+      - Lead-lag profile
+
+    A factor is valid if IC > 0.03 and ICIR > 0.5.
+    For regime factors, also check regime_validator.
+    """
+
+    def __init__(self, db_path: Optional[str] = None):
+        self.db_path = db_path or config.db_path
+
+    def validate(self, factor_name: str, factor_scores: pd.Series,
+                 forward_returns: dict[str, pd.Series]) -> FactorReport:
+        """
+        Args:
+            factor_name: Human-readable name
+            factor_scores: Series indexed by date, values 0-100
+            forward_returns: Dict of horizon → Series indexed by date (e.g. "1d" → returns)
+        """
+        report = FactorReport(factor_name)
+
+        # Align series to common dates
+        common_idx = factor_scores.index
+        for ret in forward_returns.values():
+            common_idx = common_idx.intersection(ret.index)
+
+        if len(common_idx) < 30:
+            report.conclusion = "INSUFFICIENT DATA (< 30 observations)"
+            return report
+
+        f = factor_scores[common_idx]
+        report.n_observations = len(common_idx)
+
+        # Test against 7d forward returns (primary horizon)
+        primary_ret = forward_returns.get("7d")
+        if primary_ret is None:
+            # Use first available
+            primary_ret = list(forward_returns.values())[0]
+
+        r = primary_ret[common_idx]
+
+        # IC
+        ic = information_coefficient(f, r)
+        report.ic_mean = round(ic, 4)
+
+        # Rolling IC for ICIR
+        rolling_ics = []
+        for i in range(30, len(f)):
+            ic_i = information_coefficient(f.iloc[:i], r.iloc[:i])
+            rolling_ics.append(ic_i)
+        ic_series = pd.Series(rolling_ics)
+        report.ic_std = round(ic_series.std(), 4)
+        report.icir = round(icir(ic_series), 2)
+
+        # Hit ratio
+        report.hit_ratio = round(hit_ratio(f, r), 4)
+
+        # Quantile spread
+        report.quantile_spread = round(quantile_spread(f, r), 4)
+
+        # Lead-lag
+        lead = lead_lag_ic(f, r, max_lag=14)
+        report.is_leading = lead["is_leading"]
+        report.lead_days = lead["lead_days"]
+        report.lead_ic = round(lead["best_ic"], 4)
+
+        # Conclusion
+        if abs(report.ic_mean) > 0.05 and report.icir > 1.0:
+            report.conclusion = "STRONG: significant predictive power"
+        elif abs(report.ic_mean) > 0.03 and report.icir > 0.5:
+            report.conclusion = "VALID: moderate predictive power"
+        elif abs(report.ic_mean) < 0.02:
+            report.conclusion = "CONFIRMING: describes current state, not predictive"
+        else:
+            report.conclusion = "WEAK: borderline, monitor or downweight"
+
+        return report
+
+    def validate_from_db(self, factor_name: str,
+                         score_query: str,
+                         horizon_days: int = 7) -> FactorReport:
+        """
+        Convenience: load scores from DB and OHLCV returns, then validate.
+
+        score_query: SQL that returns (date, score) pairs.
+        """
+        conn = sqlite3.connect(self.db_path)
+
+        scores_df = pd.read_sql_query(score_query, conn)
+        if scores_df.empty:
+            conn.close()
+            r = FactorReport(factor_name)
+            r.conclusion = "NO DATA"
+            return r
+
+        scores_df["date"] = pd.to_datetime(scores_df["date"])
+        scores = scores_df.set_index("date")["score"]
+
+        # Load forward returns from OHLCV
+        ohlcv = pd.read_sql_query(
+            "SELECT date, close FROM ohlcv_daily WHERE symbol='BTC/USDT:USDT' ORDER BY date",
+            conn
+        )
+        conn.close()
+
+        ohlcv["date"] = pd.to_datetime(ohlcv["date"])
+        ohlcv = ohlcv.set_index("date")
+        ohlcv["ret"] = ohlcv["close"].pct_change().shift(-1)  # forward 1d
+
+        # Build forward returns for multiple horizons
+        forward = {}
+        for h in [1, 3, 5, 7, 14]:
+            forward[str(h) + "d"] = ohlcv["close"].pct_change(periods=h).shift(-h)
+
+        return self.validate(factor_name, scores, forward)

ChanMacro/validation/metrics.py

@@ -0,0 +1,192 @@
+"""
+validation/metrics.py — Shared statistical metrics for factor and regime validation.
+"""
+
+import numpy as np
+import pandas as pd
+from scipy import stats
+from typing import Optional
+
+
+def information_coefficient(factor: pd.Series, forward_returns: pd.Series) -> float:
+    """Spearman rank IC between factor values and forward returns."""
+    mask = factor.notna() & forward_returns.notna()
+    if mask.sum() < 10:
+        return 0.0
+    ic, _ = stats.spearmanr(factor[mask], forward_returns[mask])
+    return float(ic) if not np.isnan(ic) else 0.0
+
+
+def icir(ic_series: pd.Series) -> float:
+    """Information Coefficient IR = mean(IC) / std(IC)."""
+    if len(ic_series) < 5 or ic_series.std() == 0:
+        return 0.0
+    return float(ic_series.mean() / ic_series.std())
+
+
+def hit_ratio(factor: pd.Series, forward_returns: pd.Series) -> float:
+    """Fraction of times factor direction matches return direction."""
+    mask = factor.notna() & forward_returns.notna()
+    if mask.sum() < 10:
+        return 0.5
+    # Compare sign of factor deviation from median vs sign of returns
+    factor_median = factor[mask].median()
+    factor_sign = np.sign(factor[mask] - factor_median)
+    return_sign = np.sign(forward_returns[mask])
+    return float((factor_sign == return_sign).mean())
+
+
+def quantile_spread(factor: pd.Series, forward_returns: pd.Series,
+                    n_quantiles: int = 5) -> float:
+    """Top vs bottom quantile return spread (分层回测)."""
+    mask = factor.notna() & forward_returns.notna()
+    if mask.sum() < n_quantiles * 3:
+        return 0.0
+    f = factor[mask]
+    r = forward_returns[mask]
+    labels = pd.qcut(f, n_quantiles, labels=False, duplicates="drop")
+    top_ret = r[labels == labels.max()].mean()
+    bot_ret = r[labels == labels.min()].mean()
+    return float(top_ret - bot_ret)
+
+
+def lead_lag_ic(factor: pd.Series, returns: pd.Series,
+                max_lag: int = 14) -> dict:
+    """Find the best leading/trailing relationship by computing IC at each lag."""
+    results = {}
+    for lag in range(-max_lag, max_lag + 1):
+        if lag < 0:
+            shifted = factor.shift(abs(lag))
+            ic = information_coefficient(shifted, returns)
+            results[f"lead_{abs(lag)}d"] = ic
+        elif lag > 0:
+            shifted = returns.shift(lag)
+            ic = information_coefficient(factor, shifted)
+            results[f"lag_{lag}d"] = ic
+        else:
+            ic = information_coefficient(factor, returns)
+            results["sync"] = ic
+
+    # Find best lead period
+    lead_ics = {k: v for k, v in results.items() if k.startswith("lead_")}
+    best_lead = max(lead_ics, key=lead_ics.get) if lead_ics else "sync"
+    best_ic = lead_ics.get(best_lead, results.get("sync", 0))
+
+    return {
+        "best_lead": best_lead,
+        "best_ic": best_ic,
+        "ic_curve": results,
+        "is_leading": best_lead.startswith("lead_") and abs(best_ic) > 0.03,
+        "lead_days": int(best_lead.split("_")[1].rstrip("d")) if best_lead.startswith("lead_") else 0,
+    }
+
+
+def mutual_information(factor: pd.Series, labels: pd.Series,
+                       n_bins: int = 10) -> float:
+    """Mutual information between factor (binned) and discrete regime labels."""
+    mask = factor.notna() & labels.notna()
+    if mask.sum() < 20:
+        return 0.0
+    f = factor[mask]
+    l = labels[mask]
+    try:
+        f_binned = pd.qcut(f, n_bins, labels=False, duplicates="drop")
+    except ValueError:
+        f_binned = pd.cut(f, n_bins, labels=False)
+    mi = 0.0
+    for fi in range(n_bins):
+        p_f = (f_binned == fi).mean()
+        if p_f == 0:
+            continue
+        for li in l.unique():
+            p_l = (l == li).mean()
+            p_joint = ((f_binned == fi) & (l == li)).mean()
+            if p_joint > 0:
+                mi += p_joint * np.log(p_joint / (p_f * p_l))
+    return float(mi)
+
+
+def kl_divergence(factor: pd.Series, labels: pd.Series,
+                  regime_a: str, regime_b: str, n_bins: int = 10) -> float:
+    """KL divergence between factor distributions in two regimes."""
+    mask_a = (labels == regime_a) & factor.notna()
+    mask_b = (labels == regime_b) & factor.notna()
+    if mask_a.sum() < 10 or mask_b.sum() < 10:
+        return 0.0
+    try:
+        hist_a, edges = np.histogram(factor[mask_a], bins=n_bins, density=True)
+        hist_b, _ = np.histogram(factor[mask_b], bins=edges, density=True)
+    except ValueError:
+        return 0.0
+    hist_a = np.clip(hist_a, 1e-10, None)
+    hist_b = np.clip(hist_b, 1e-10, None)
+    return float((hist_a * np.log(hist_a / hist_b)).sum())
+
+
+def anova_f_score(factor: pd.Series, labels: pd.Series) -> float:
+    """ANOVA F-statistic: how well factor separates different regimes."""
+    mask = factor.notna() & labels.notna()
+    if mask.sum() < 20:
+        return 0.0
+    groups = [factor[mask][labels[mask] == lbl] for lbl in labels[mask].unique()]
+    groups = [g for g in groups if len(g) > 1]
+    if len(groups) < 2:
+        return 0.0
+    f_stat, _ = stats.f_oneway(*groups)
+    return float(f_stat) if not np.isnan(f_stat) else 0.0
+
+
+def transition_matrix(labels: pd.Series) -> pd.DataFrame:
+    """Compute Markov transition matrix from regime sequence."""
+    unique = sorted(labels.dropna().unique())
+    n = len(unique)
+    matrix = np.zeros((n, n))
+    seq = labels.dropna().values
+    for i in range(len(seq) - 1):
+        from_idx = unique.index(seq[i])
+        to_idx = unique.index(seq[i + 1])
+        matrix[from_idx][to_idx] += 1
+
+    # Row-normalize
+    row_sums = matrix.sum(axis=1, keepdims=True)
+    row_sums[row_sums == 0] = 1
+    matrix = matrix / row_sums
+
+    return pd.DataFrame(matrix, index=unique, columns=unique)
+
+
+def regime_duration_stats(labels: pd.Series) -> dict:
+    """Compute average duration, flip rate, state entropy for regime sequence."""
+    seq = labels.dropna().values
+    if len(seq) < 2:
+        return {"avg_duration": 0, "flip_rate": 0, "state_entropy": 0, "n_days": len(seq)}
+
+    # Count durations
+    durations = []
+    current = seq[0]
+    count = 1
+    flips = 0
+    for i in range(1, len(seq)):
+        if seq[i] == current:
+            count += 1
+        else:
+            durations.append(count)
+            current = seq[i]
+            count = 1
+            flips += 1
+    durations.append(count)
+
+    avg_dur = float(np.mean(durations)) if durations else 0
+    flip_rate = flips / len(seq)
+
+    # State entropy
+    _, counts = np.unique(seq, return_counts=True)
+    probs = counts / counts.sum()
+    entropy = float(-(probs * np.log2(probs + 1e-10)).sum())
+
+    return {
+        "avg_duration": round(avg_dur, 1),
+        "flip_rate": round(flip_rate, 3),
+        "state_entropy": round(entropy, 3),
+        "n_days": len(seq),
+    }

ChanMacro/validation/regime_validator.py

@@ -0,0 +1,144 @@
+"""
+validation/regime_validator.py — Validates factors as regime separators.
+
+Tests: Mutual Information, KL Divergence, ANOVA F-score.
+Answers: "Does this factor distinguish different market regimes?"
+
+Key insight: a factor may have low IC (poor return predictor) but high
+regime separation (good regime classifier). Breadth is the prime example.
+"""
+
+from datetime import date as Date
+from typing import Optional
+import sqlite3
+import logging
+
+import numpy as np
+import pandas as pd
+
+from config import config
+from .metrics import (
+    mutual_information, kl_divergence, anova_f_score,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class RegimeReport:
+    """Structured report for regime separation validation."""
+
+    def __init__(self, factor_name: str):
+        self.factor_name = factor_name
+        self.mutual_info: float = 0.0
+        self.anova_f: float = 0.0
+        self.kl_pairs: dict = {}  # (regime_a, regime_b) → KL divergence
+        self.best_separates: list[str] = []
+        self.separation_score: float = 0.0
+        self.is_regime_factor: bool = False
+        self.conclusion: str = ""
+
+    def summary(self) -> str:
+        lines = [
+            f"Factor: {self.factor_name}",
+            f"  Mutual Information: {self.mutual_info:.4f}",
+            f"  ANOVA F: {self.anova_f:.1f}",
+            f"  Best separates: {', '.join(self.best_separates) if self.best_separates else 'none'}",
+            f"  Regime Factor: {'YES' if self.is_regime_factor else 'No'}",
+            f"  → {self.conclusion}",
+        ]
+        return "\n".join(lines)
+
+
+class RegimeValidator:
+    """
+    Validates a factor's ability to separate different market regimes.
+
+    A good regime factor has:
+      - Mutual Information > 0.1
+      - KL Divergence between regimes > 0.5
+      - ANOVA F-score high
+    """
+
+    def __init__(self, db_path: Optional[str] = None):
+        self.db_path = db_path or config.db_path
+
+    def validate(self, factor_name: str, factor_scores: pd.Series,
+                 regime_labels: pd.Series) -> RegimeReport:
+        """
+        Args:
+            factor_name: Human-readable name
+            factor_scores: Series indexed by date, values 0-100
+            regime_labels: Series indexed by date, values = 'TREND'/'RANGE'/'PANIC'
+        """
+        report = RegimeReport(factor_name)
+
+        # Align
+        common_idx = factor_scores.index.intersection(regime_labels.index)
+        if len(common_idx) < 30:
+            report.conclusion = "INSUFFICIENT DATA"
+            return report
+
+        f = factor_scores[common_idx]
+        labels = regime_labels[common_idx]
+
+        # Mutual Information
+        report.mutual_info = round(mutual_information(f, labels), 4)
+
+        # ANOVA
+        report.anova_f = round(anova_f_score(f, labels), 1)
+
+        # KL Divergence between each pair of regimes
+        unique_regimes = sorted(labels.unique())
+        for i, ra in enumerate(unique_regimes):
+            for rb in unique_regimes[i + 1:]:
+                kl = kl_divergence(f, labels, ra, rb)
+                report.kl_pairs[f"{ra}↔{rb}"] = round(kl, 4)
+
+        # Best separation
+        if report.kl_pairs:
+            sorted_pairs = sorted(report.kl_pairs, key=report.kl_pairs.get, reverse=True)
+            report.best_separates = sorted_pairs[:2]
+
+        # Separation score (0-1 composite)
+        mi_norm = min(report.mutual_info / 0.5, 1.0)
+        kl_avg = np.mean(list(report.kl_pairs.values())) if report.kl_pairs else 0
+        kl_norm = min(kl_avg / 1.0, 1.0)
+        report.separation_score = round(0.5 * mi_norm + 0.5 * kl_norm, 2)
+
+        # Is this a good regime factor?
+        report.is_regime_factor = (
+            report.mutual_info > 0.1 and
+            kl_avg > 0.5
+        )
+
+        if report.separation_score > 0.8:
+            report.conclusion = "EXCELLENT regime separator"
+        elif report.separation_score > 0.5:
+            report.conclusion = "GOOD regime separator"
+        elif report.separation_score > 0.3:
+            report.conclusion = "MODERATE — some regime separation"
+        else:
+            report.conclusion = "WEAK regime separator"
+
+        return report
+
+    def validate_from_db(self, factor_name: str,
+                         score_query: str) -> RegimeReport:
+        """Load scores and regime labels from DB, then validate."""
+        conn = sqlite3.connect(self.db_path)
+
+        scores_df = pd.read_sql_query(score_query, conn)
+        regimes_df = pd.read_sql_query(
+            "SELECT date, regime FROM regime_history", conn
+        )
+        conn.close()
+
+        if scores_df.empty or regimes_df.empty:
+            r = RegimeReport(factor_name)
+            r.conclusion = "NO DATA"
+            return r
+
+        scores = scores_df.set_index("date")["score"]
+        regimes = regimes_df.set_index("date")["regime"]
+
+        return self.validate(factor_name, scores, regimes)

ChanMacro/validation/reporter.py

@@ -0,0 +1,120 @@
+"""
+validation/reporter.py — Aggregates all validation reports into a unified summary.
+
+Used by: python main.py validate
+"""
+
+from datetime import date as Date
+from typing import Optional
+import logging
+
+from .factor_validator import FactorValidator, FactorReport
+from .regime_validator import RegimeValidator, RegimeReport
+from .transition_validator import TransitionValidator, TransitionReport
+
+logger = logging.getLogger(__name__)
+
+
+class ValidationReporter:
+    """
+    Orchestrates full validation pipeline:
+
+    1. Factor validation (IC, ICIR, Hit Ratio) for each factor
+    2. Regime validation (MI, KL, ANOVA) for each factor
+    3. Transition validation (stability, flip rate)
+    """
+
+    def __init__(self, db_path: Optional[str] = None):
+        from config import config
+        self.db_path = db_path or config.db_path
+        self.factor_validator = FactorValidator(self.db_path)
+        self.regime_validator = RegimeValidator(self.db_path)
+        self.transition_validator = TransitionValidator(self.db_path)
+
+    def run_all(self) -> str:
+        """Run all validations and return a formatted report string."""
+        lines = []
+        lines.append("=" * 70)
+        lines.append(f"  ChanMacro Validation Report — {Date.today()}")
+        lines.append("=" * 70)
+
+        # ── Factor Validation ──────────────────────────
+        lines.append("")
+        lines.append("─" * 50)
+        lines.append("  FACTOR VALIDATION (Predictive Power)")
+        lines.append("─" * 50)
+
+        factor_queries = {
+            "Price Structure": "SELECT date, score FROM ohlcv_daily WHERE ema20 IS NOT NULL",
+            "Breadth": """
+                SELECT bd.date,
+                    (bd.advance_top50*1.0/(bd.advance_top50+bd.decline_top50+1)*100*0.30
+                     + bd.above_ema20_top50*1.0/50*100*0.35
+                     + bd.new_highs_20d_top50*1.0/50*100*0.20
+                     + 50*0.15) as score
+                FROM breadth_daily bd
+            """,
+        }
+
+        factor_reports: list[FactorReport] = []
+        for name, query in factor_queries.items():
+            try:
+                report = self.factor_validator.validate_from_db(name, query)
+                factor_reports.append(report)
+                lines.append(report.summary())
+                lines.append("")
+            except Exception as e:
+                logger.warning(f"Factor validation failed for {name}: {e}")
+
+        # ── Regime Validation ──────────────────────────
+        lines.append("─" * 50)
+        lines.append("  REGIME VALIDATION (Regime Separation)")
+        lines.append("─" * 50)
+
+        regime_reports: list[RegimeReport] = []
+        for name, query in factor_queries.items():
+            try:
+                report = self.regime_validator.validate_from_db(name, query)
+                regime_reports.append(report)
+                lines.append(report.summary())
+                lines.append("")
+            except Exception as e:
+                logger.warning(f"Regime validation failed for {name}: {e}")
+
+        # ── Transition Validation ──────────────────────
+        lines.append("─" * 50)
+        lines.append("  TRANSITION VALIDATION (Regime Stability)")
+        lines.append("─" * 50)
+
+        try:
+            t_report = self.transition_validator.validate_from_db()
+            lines.append(t_report.summary())
+        except Exception as e:
+            logger.warning(f"Transition validation failed: {e}")
+
+        # ── Summary ────────────────────────────────────
+        lines.append("")
+        lines.append("=" * 70)
+        lines.append("  SUMMARY")
+        lines.append("=" * 70)
+
+        # Factor ranking by IC
+        if factor_reports:
+            ranked = sorted(factor_reports, key=lambda r: abs(r.ic_mean), reverse=True)
+            lines.append("  Factor Ranking (by |IC|):")
+            for i, r in enumerate(ranked):
+                tag = "★★★" if abs(r.ic_mean) > 0.05 else "★★" if abs(r.ic_mean) > 0.03 else "★"
+                lines.append(f"    {i+1}. {r.factor_name:20s} IC={r.ic_mean:+.4f} {tag}  {r.conclusion}")
+
+        # Regime factor ranking
+        if regime_reports:
+            ranked_r = sorted(regime_reports, key=lambda r: r.separation_score, reverse=True)
+            lines.append("")
+            lines.append("  Regime Factor Ranking (by Separation Score):")
+            for i, r in enumerate(ranked_r):
+                lines.append(f"    {i+1}. {r.factor_name:20s} Score={r.separation_score:.2f}  {r.conclusion}")
+
+        lines.append("")
+        lines.append("=" * 70)
+
+        return "\n".join(lines)

ChanMacro/validation/transition_validator.py

@@ -0,0 +1,131 @@
+"""
+validation/transition_validator.py — Validates regime stability.
+
+Tests: Transition matrix, average duration, flip rate, state entropy.
+Answers: "Does the regime design produce stable, persistent states?"
+
+Hard requirements:
+  - avg_duration > 5 days
+  - flip_rate < 15%
+  - Fails → regime definition needs redesign.
+"""
+
+from typing import Optional
+import sqlite3
+import logging
+
+import numpy as np
+import pandas as pd
+
+from config import config
+from .metrics import transition_matrix, regime_duration_stats
+
+logger = logging.getLogger(__name__)
+
+
+class TransitionReport:
+    """Structured report for regime stability validation."""
+
+    def __init__(self):
+        self.avg_duration: float = 0.0
+        self.flip_rate: float = 0.0
+        self.state_entropy: float = 0.0
+        self.n_days: int = 0
+        self.transition_matrix: Optional[pd.DataFrame] = None
+        self.persistence_score: float = 0.0
+        self.is_stable: bool = False
+        self.conclusion: str = ""
+        self.warnings: list[str] = []
+
+    def summary(self) -> str:
+        lines = [
+            f"Regime Stability (N={self.n_days} days)",
+            f"  Avg Duration: {self.avg_duration:.1f} days (need > {config.regime_min_avg_duration})",
+            f"  Flip Rate: {self.flip_rate:.1%} (need < {config.regime_max_flip_rate:.0%})",
+            f"  State Entropy: {self.state_entropy:.3f}",
+            f"  Persistence Score: {self.persistence_score:.2f}",
+            f"  Stable: {'YES' if self.is_stable else 'NO — redesign needed'}",
+        ]
+        if self.warnings:
+            lines.append(f"  Warnings: {'; '.join(self.warnings)}")
+        if self.transition_matrix is not None:
+            lines.append(f"  Transition Matrix:\n{self.transition_matrix.to_string()}")
+        lines.append(f"  → {self.conclusion}")
+        return "\n".join(lines)
+
+
+class TransitionValidator:
+    """
+    Validates regime temporal stability.
+
+    Regime must persist — not flip daily.
+    If flip_rate > 20% or avg_duration < 3 days → regime definition failed.
+    """
+
+    def __init__(self, db_path: Optional[str] = None):
+        self.db_path = db_path or config.db_path
+
+    def validate(self, regime_labels: pd.Series) -> TransitionReport:
+        """Validate a regime sequence for stability."""
+        report = TransitionReport()
+        report.n_days = len(regime_labels)
+
+        if len(regime_labels) < 30:
+            report.conclusion = "INSUFFICIENT DATA (< 30 days)"
+            return report
+
+        # Duration stats
+        stats = regime_duration_stats(regime_labels)
+        report.avg_duration = stats["avg_duration"]
+        report.flip_rate = stats["flip_rate"]
+        report.state_entropy = stats["state_entropy"]
+
+        # Transition matrix
+        report.transition_matrix = transition_matrix(regime_labels)
+
+        # Persistence: how often does regime stay the same?
+        diag = np.diag(report.transition_matrix.values)
+        report.persistence_score = round(float(np.mean(diag)), 2)
+
+        # Stability check
+        report.is_stable = (
+            report.avg_duration >= config.regime_min_avg_duration and
+            report.flip_rate <= config.regime_max_flip_rate
+        )
+
+        # Warnings
+        if report.avg_duration < 3:
+            report.warnings.append(f"CRITICAL: avg duration={report.avg_duration:.1f}d — regime flips too fast")
+        elif report.avg_duration < config.regime_min_avg_duration:
+            report.warnings.append(f"WARNING: avg duration={report.avg_duration:.1f}d < {config.regime_min_avg_duration}")
+
+        if report.flip_rate > 0.20:
+            report.warnings.append(f"CRITICAL: flip rate={report.flip_rate:.1%} — regime unstable")
+        elif report.flip_rate > config.regime_max_flip_rate:
+            report.warnings.append(f"WARNING: flip rate={report.flip_rate:.1%} > {config.regime_max_flip_rate:.0%}")
+
+        if report.state_entropy > 2.0:
+            report.warnings.append(f"NOTE: high state entropy={report.state_entropy:.2f}, regimes may be too fine-grained")
+
+        if report.is_stable:
+            report.conclusion = "PASS: regime design is stable"
+        else:
+            report.conclusion = "FAIL: regime definition needs adjustment"
+
+        return report
+
+    def validate_from_db(self) -> TransitionReport:
+        """Load regime history from DB and validate stability."""
+        conn = sqlite3.connect(self.db_path)
+        df = pd.read_sql_query(
+            "SELECT date, regime FROM regime_history ORDER BY date", conn
+        )
+        conn.close()
+
+        if df.empty:
+            r = TransitionReport()
+            r.conclusion = "NO DATA"
+            return r
+
+        regimes = df.set_index("date")["regime"]
+        return self.validate(regimes)

ChanMacro/web/app.py

@@ -0,0 +1,178 @@
+"""
+web/app.py — ChanMacro dashboard (Flask, port 8124).
+"""
+
+import sys
+import os
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+import json
+from datetime import date as Date
+from flask import Flask, render_template, jsonify, request
+
+from database import get_connection
+from config import config
+from scoring.price_structure import PriceStructureScorer
+from scoring.breadth_scorer import BreadthScorer
+from scoring.oi_matrix import OIMatrixScorer
+from scoring.volatility_regime import VolatilityRegimeScorer
+from regime_detector import RegimeDetector
+from models import MarketStateVector
+from expectancy.engine import BayesianExpectancyEngine
+
+app = Flask(__name__)
+
+
+def _build_state(target: Date):
+    """Build MarketStateVector and persist regime to DB."""
+    ps = PriceStructureScorer().compute(target)
+    br = BreadthScorer().compute(target)
+    oi = OIMatrixScorer().compute(target)
+    vol = VolatilityRegimeScorer().compute(target)
+
+    detector = RegimeDetector()
+    detector.load_state(config.db_path)
+    r = detector.detect(ps.score, br.breadth_top50, vol.vol_regime.value, target)
+
+    state = MarketStateVector(
+        date=target, regime=r.regime, regime_confidence=r.confidence,
+        regime_version=r.regime_version, regime_maturity_score=r.maturity_score,
+        breadth_top20=br.breadth_top20, breadth_top30=br.breadth_top30,
+        breadth_top50=br.breadth_top50, breadth_bucket=br.breadth_bucket,
+        breadth_divergence=br.breadth_divergence,
+        oi_state=oi.oi_state, volatility_regime=vol.vol_regime,
+        price_structure_score=ps, breadth_score=br,
+        oi_matrix_score=oi, volatility_regime_score=vol,
+    )
+    state.market_state_hash = state.compute_hash()
+
+    # Persist regime to DB so load_state() works across requests
+    conn = get_connection()
+    conn.execute("""
+        INSERT OR REPLACE INTO regime_history
+        (date, regime, confidence, regime_version, maturity_score, all_scores_json,
+         prior_regime, confirmation_days)
+        VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+    """, (
+        str(target), r.regime.value, r.confidence, r.regime_version,
+        r.maturity_score, json.dumps(r.all_scores),
+        r.prior_regime.value if r.prior_regime else None,
+        r.confirmation_days,
+    ))
+    conn.commit()
+    conn.close()
+
+    return state
+
+
+@app.route("/")
+def dashboard():
+    return render_template("index.html")
+
+
+@app.route("/api/state")
+def api_state():
+    """Current market state with all factor scores."""
+    try:
+        target = Date.today()
+        state = _build_state(target)
+        return jsonify({
+            "date": str(state.date),
+            "regime": state.regime.value,
+            "regime_confidence": state.regime_confidence,
+            "regime_maturity": state.regime_maturity_score,
+            "breadth": {
+                "score": state.breadth_score.score,
+                "bucket": state.breadth_bucket.value,
+                "top20": state.breadth_top20,
+                "top30": state.breadth_top30,
+                "top50": state.breadth_top50,
+                "divergence": state.breadth_divergence,
+                "narrative": state.breadth_score.narrative,
+            },
+            "oi_state": state.oi_state.value,
+            "oi_score": state.oi_matrix_score.score,
+            "oi_narrative": state.oi_matrix_score.narrative,
+            "volatility": state.volatility_regime.value,
+            "price_structure": {
+                "score": state.price_structure_score.score,
+                "trend": state.price_structure_score.trend_strength,
+                "vol_comp": state.price_structure_score.volatility_compression,
+                "momentum": state.price_structure_score.momentum,
+                "label": state.price_structure_score.label,
+                "narrative": state.price_structure_score.narrative,
+            },
+        })
+    except Exception as e:
+        return jsonify({"error": str(e)}), 500
+
+
+@app.route("/api/history")
+def api_history():
+    """Regime and factor score history."""
+    days = request.args.get("days", 60, type=int)
+    conn = get_connection()
+
+    # Regime history
+    regimes = conn.execute(
+        "SELECT date, regime, confidence, maturity_score FROM regime_history ORDER BY date DESC LIMIT ?",
+        (days,)
+    ).fetchall()
+
+    # Breadth history
+    breadth = conn.execute(
+        "SELECT date, advance_top50, decline_top50, above_ema20_top50 FROM breadth_daily ORDER BY date DESC LIMIT ?",
+        (days,)
+    ).fetchall()
+
+    conn.close()
+
+    return jsonify({
+        "regimes": [{"date": r["date"], "regime": r["regime"],
+                      "confidence": r["confidence"], "maturity": r["maturity_score"]}
+                    for r in reversed(regimes)],
+        "breadth": [{"date": b["date"], "advance": b["advance_top50"],
+                      "decline": b["decline_top50"], "above_ema20": b["above_ema20_top50"]}
+                    for b in reversed(breadth)],
+    })
+
+
+@app.route("/api/expectancy")
+def api_expectancy():
+    """Query signal expectancy."""
+    signal = request.args.get("signal", "B3")
+    try:
+        target = Date.today()
+        state = _build_state(target)
+        engine = BayesianExpectancyEngine(level_min_samples=5)
+        report = engine.estimate(state, signal_type=signal, target_date=target)
+
+        layers = []
+        for l in report.layers:
+            layers.append({
+                "name": l.name,
+                "samples": l.samples,
+                "effective_samples": l.effective_samples,
+                "raw_winrate": l.raw_winrate,
+                "posterior_winrate": l.posterior_winrate,
+                "avg_return": l.avg_return,
+            })
+
+        return jsonify({
+            "signal": signal,
+            "final_estimate": report.final_estimate,
+            "sufficiency": report.sufficiency.value,
+            "source": report.source,
+            "avg_return_7d": report.avg_return_7d,
+            "profit_factor": report.profit_factor,
+            "max_adverse": report.max_adverse_excursion,
+            "layers": layers,
+        })
+    except Exception as e:
+        return jsonify({"error": str(e)}), 500
+
+
+if __name__ == "__main__":
+    from scheduler import get_scheduler
+    get_scheduler().start()
+    app.run(host="0.0.0.0", port=8124, debug=True)

ChanMacro/web/static/js/dashboard.js

@@ -0,0 +1,160 @@
+// dashboard.js — ChanMacro
+
+const C = { TREND: "#3fb950", RANGE: "#d29922", PANIC: "#f85149" };
+let regimeChart = null, breadthChart = null;
+
+async function loadState() {
+  try {
+    const r = await fetch("/api/state");
+    const d = await r.json();
+    if (d.error) { document.getElementById("update-time").textContent = d.error; return; }
+
+    document.getElementById("update-time").textContent = d.date;
+
+    // Hero
+    const regime = d.regime;
+    const names = { TREND: "TREND", RANGE: "RANGE", PANIC: "PANIC" };
+    document.getElementById("hero-regime").textContent = names[regime] || regime;
+    document.getElementById("hero-regime").className = "regime-name " + regime.toLowerCase();
+    document.getElementById("hero-badge").textContent = regime;
+    document.getElementById("hero-badge").className = "regime-badge " + regime.toLowerCase();
+    document.getElementById("hero-conf").textContent = (d.regime_confidence * 100).toFixed(0) + "%";
+    document.getElementById("hero-maturity").textContent = d.regime_maturity.toFixed(0);
+    document.getElementById("hero-ps").textContent = d.price_structure.score.toFixed(0);
+    document.getElementById("hero-ps").style.color =
+      d.price_structure.score >= 60 ? "#3fb950" : d.price_structure.score >= 40 ? "#d29922" : "#f85149";
+    document.getElementById("hero-br").textContent = d.breadth.score.toFixed(0);
+    document.getElementById("hero-br").style.color =
+      d.breadth.bucket === "EXTREME" || d.breadth.bucket === "STRONG" ? "#3fb950" :
+      d.breadth.bucket === "WEAK" || d.breadth.bucket === "PANIC" ? "#f85149" : "#d29922";
+
+    // Factor cards
+    const ps = d.price_structure;
+    document.getElementById("f-price").textContent = ps.score.toFixed(0);
+    document.getElementById("f-price").style.color =
+      ps.score >= 60 ? "#3fb950" : ps.score >= 40 ? "#d29922" : "#f85149";
+    document.getElementById("f-price-sub").textContent =
+      `趋势 ${ps.trend.toFixed(0)} · 波动 ${ps.vol_comp.toFixed(0)} · 动量 ${ps.momentum.toFixed(0)}`;
+    document.getElementById("bar-price").style.width = ps.score + "%";
+    document.getElementById("bar-price").style.background =
+      ps.score >= 60 ? "#3fb950" : ps.score >= 40 ? "#d29922" : "#f85149";
+
+    const br = d.breadth;
+    document.getElementById("f-breadth").textContent = br.score.toFixed(0);
+    document.getElementById("f-breadth").style.color =
+      br.bucket === "EXTREME" || br.bucket === "STRONG" ? "#3fb950" :
+      br.bucket === "WEAK" || br.bucket === "PANIC" ? "#f85149" : "#d29922";
+    document.getElementById("f-breadth-sub").textContent =
+      `${br.bucket} · T20=${br.top20.toFixed(0)} T50=${br.top50.toFixed(0)}`;
+    document.getElementById("bar-breadth").style.width = br.score + "%";
+    document.getElementById("bar-breadth").style.background =
+      br.bucket === "EXTREME" || br.bucket === "STRONG" ? "#3fb950" :
+      br.bucket === "WEAK" || br.bucket === "PANIC" ? "#f85149" : "#d29922";
+
+    document.getElementById("f-oi").textContent = d.oi_state.toUpperCase().replace(" ", "\n");
+    document.getElementById("f-oi").style.color =
+      d.oi_state === "New Longs" ? "#3fb950" : d.oi_state.includes("Short") || d.oi_state === "Long Exit" ? "#f85149" : "#8b949e";
+    document.getElementById("f-oi-sub").textContent = d.oi_narrative;
+
+    const vm = { LOW_VOL: "低波动", NORMAL_VOL: "正常", HIGH_VOL: "高波动", EXPLOSIVE_VOL: "极端" };
+    document.getElementById("f-vol").textContent = vm[d.volatility] || d.volatility;
+    document.getElementById("f-vol").style.color =
+      d.volatility === "LOW_VOL" ? "#58a6ff" : d.volatility === "NORMAL_VOL" ? "#8b949e" :
+      d.volatility === "HIGH_VOL" ? "#d29922" : "#f85149";
+    document.getElementById("f-vol-sub").textContent = d.volatility;
+    document.getElementById("bar-vol").style.width =
+      (d.volatility === "EXPLOSIVE_VOL" ? 95 : d.volatility === "HIGH_VOL" ? 70 :
+       d.volatility === "NORMAL_VOL" ? 40 : 20) + "%";
+    document.getElementById("bar-vol").style.background =
+      d.volatility === "EXPLOSIVE_VOL" ? "#f85149" : d.volatility === "HIGH_VOL" ? "#d29922" :
+      d.volatility === "NORMAL_VOL" ? "#8b949e" : "#58a6ff";
+  } catch (e) {
+    document.getElementById("update-time").textContent = "连接失败";
+  }
+}
+
+async function loadHistory() {
+  try {
+    const r = await fetch("/api/history?days=60");
+    const d = await r.json();
+
+    const dates = d.regimes.map(x => x.date);
+    const colors = d.regimes.map(x => C[x.regime] || "#5c6675");
+
+    if (regimeChart) regimeChart.destroy();
+    regimeChart = new Chart(document.getElementById("chart-regime").getContext("2d"), {
+      type: "bar",
+      data: { labels: dates, datasets: [{ data: d.regimes.map(x => x.confidence * 100),
+        backgroundColor: colors, borderWidth: 0, borderRadius: 2 }] },
+      options: {
+        responsive: true, maintainAspectRatio: false,
+        plugins: { legend: { display: false } },
+        scales: {
+          x: { ticks: { color: "#5c6675", maxTicksLimit: 15, maxRotation: 45, font: { size: 10 } },
+               grid: { color: "#151a23" } },
+          y: { max: 100, ticks: { color: "#5c6675", font: { size: 10 } }, grid: { color: "#151a23" } }
+        }
+      }
+    });
+
+    if (breadthChart) breadthChart.destroy();
+    breadthChart = new Chart(document.getElementById("chart-breadth").getContext("2d"), {
+      type: "line",
+      data: {
+        labels: d.breadth.map(x => x.date),
+        datasets: [
+          { label: "上涨", data: d.breadth.map(x => x.advance), borderColor: "#3fb950",
+            backgroundColor: "rgba(63,185,80,0.08)", fill: true, tension: 0.3, pointRadius: 0 },
+          { label: "下跌", data: d.breadth.map(x => x.decline), borderColor: "#f85149",
+            backgroundColor: "rgba(248,81,73,0.06)", fill: true, tension: 0.3, pointRadius: 0 },
+          { label: ">EMA20", data: d.breadth.map(x => x.above_ema20), borderColor: "#58a6ff",
+            borderDash: [3, 3], tension: 0.3, pointRadius: 0 },
+        ]
+      },
+      options: {
+        responsive: true, maintainAspectRatio: false,
+        plugins: { legend: { labels: { color: "#5c6675", usePointStyle: true, boxWidth: 6, font: { size: 10 } } } },
+        scales: {
+          x: { ticks: { color: "#5c6675", maxTicksLimit: 15, maxRotation: 45, font: { size: 10 } },
+               grid: { color: "#151a23" } },
+          y: { ticks: { color: "#5c6675", font: { size: 10 } }, grid: { color: "#151a23" } }
+        }
+      }
+    });
+  } catch (e) { console.error(e); }
+}
+
+async function loadExpectancy() {
+  const signal = document.getElementById("exp-signal").value;
+  try {
+    const r = await fetch(`/api/expectancy?signal=${signal}`);
+    const d = await r.json();
+    if (d.error) { document.getElementById("exp-layers").innerHTML =
+      `<tr><td colspan="6" style="color:#f85149">${d.error}</td></tr>`; return; }
+
+    const el = document.getElementById("exp-sufficiency");
+    el.textContent = d.sufficiency;
+    el.className = "suff suff-" + d.sufficiency;
+
+    let html = "";
+    for (const l of d.layers) {
+      html += `<tr>
+        <td>${l.name}</td><td>${l.samples}</td><td>${l.effective_samples.toFixed(0)}</td>
+        <td>${l.raw_winrate ? (l.raw_winrate * 100).toFixed(1) + "%" : "—"}</td>
+        <td><strong>${(l.posterior_winrate * 100).toFixed(1)}%</strong></td>
+        <td style="color:${l.avg_return > 0 ? '#3fb950' : l.avg_return < 0 ? '#f85149' : '#8b949e'}">${l.avg_return ? (l.avg_return > 0 ? "+" : "") + l.avg_return.toFixed(2) + "%" : "—"}</td>
+      </tr>`;
+    }
+    document.getElementById("exp-layers").innerHTML = html;
+
+    let s = `后验胜率 <strong style="color:#58a6ff">${(d.final_estimate * 100).toFixed(1)}%</strong>`;
+    if (d.avg_return_7d) s += ` · 平均收益 <strong>${d.avg_return_7d > 0 ? "+" : ""}${d.avg_return_7d.toFixed(2)}%</strong>`;
+    if (d.profit_factor) s += ` · 盈亏比 <strong>${d.profit_factor}</strong>`;
+    if (d.max_adverse) s += ` · MAE <strong>${d.max_adverse.toFixed(1)}%</strong>`;
+    document.getElementById("exp-summary").innerHTML = s;
+  } catch (e) { console.error(e); }
+}
+
+loadState();
+loadHistory();
+loadExpectancy();

ChanMacro/web/templates/index.html

@@ -0,0 +1,163 @@
+<!DOCTYPE html>
+<html lang="zh-CN">
+<head>
+<meta charset="UTF-8">
+<meta name="viewport" content="width=device-width, initial-scale=1.0">
+<title>ChanMacro — 市场状态</title>
+<script src="https://cdn.jsdelivr.net/npm/chart.js@4.4.0/dist/chart.umd.min.js"></script>
+<style>
+  * { margin: 0; padding: 0; box-sizing: border-box; }
+  body { background: #0a0e14; color: #c9d1d9; font-family: -apple-system, BlinkMacSystemFont, "SF Mono", monospace; }
+  .app { max-width: 1200px; margin: 0 auto; padding: 20px 24px; }
+
+  /* Header */
+  .header { display: flex; justify-content: space-between; align-items: flex-end; padding: 20px 0 28px;
+            border-bottom: 1px solid #1c2333; margin-bottom: 24px; }
+  .header h1 { font-size: 22px; font-weight: 600; letter-spacing: 1px; }
+  .header h1 span { color: #58a6ff; }
+  .header .time { color: #5c6675; font-size: 13px; }
+  .dot { display: inline-block; width: 7px; height: 7px; border-radius: 50%; background: #3fb950;
+         margin-right: 6px; animation: pulse 2s infinite; }
+  @keyframes pulse { 0%,100%{opacity:1} 50%{opacity:0.4} }
+
+  /* Regime Hero */
+  .hero { display: flex; gap: 16px; margin-bottom: 24px; }
+  .hero-card { flex: 1; background: #11161e; border: 1px solid #1c2333; border-radius: 8px; padding: 20px 24px; }
+  .hero-card.main { flex: 2; display: flex; align-items: center; gap: 28px; }
+  .regime-badge { display: inline-block; padding: 5px 16px; border-radius: 4px; font-size: 13px;
+                  font-weight: 600; letter-spacing: 2px; }
+  .regime-badge.trend { background: rgba(63,185,80,0.12); color: #3fb950; border: 1px solid rgba(63,185,80,0.3); }
+  .regime-badge.range { background: rgba(210,153,34,0.12); color: #d29922; border: 1px solid rgba(210,153,34,0.3); }
+  .regime-badge.panic { background: rgba(248,81,73,0.12); color: #f85149; border: 1px solid rgba(248,81,73,0.3); }
+  .regime-name { font-size: 42px; font-weight: 700; letter-spacing: 2px; }
+  .regime-name.trend { color: #3fb950; }
+  .regime-name.range { color: #d29922; }
+  .regime-name.panic { color: #f85149; }
+  .hero-stat { text-align: center; }
+  .hero-stat .val { font-size: 28px; font-weight: 600; color: #e6edf3; }
+  .hero-stat .lbl { font-size: 11px; color: #5c6675; letter-spacing: 1px; margin-top: 4px; }
+
+  /* Factor Grid */
+  .grid { display: grid; grid-template-columns: repeat(4, 1fr); gap: 12px; margin-bottom: 24px; }
+  .fcard { background: #11161e; border: 1px solid #1c2333; border-radius: 8px; padding: 18px 20px; }
+  .fcard .title { font-size: 11px; color: #5c6675; letter-spacing: 1.5px; margin-bottom: 10px; }
+  .fcard .score { font-size: 38px; font-weight: 700; margin-bottom: 4px; }
+  .fcard .sub { font-size: 12px; color: #5c6675; }
+  .fcard .bar-wrap { height: 3px; background: #1c2333; border-radius: 2px; margin-top: 12px; }
+  .fcard .bar { height: 100%; border-radius: 2px; transition: width 0.6s; }
+
+  /* Charts */
+  .charts { display: grid; grid-template-columns: 1fr 1fr; gap: 12px; margin-bottom: 24px; }
+  .chart-box { background: #11161e; border: 1px solid #1c2333; border-radius: 8px; padding: 18px 20px; }
+  .chart-box h3 { font-size: 12px; color: #5c6675; letter-spacing: 1.5px; margin-bottom: 14px; }
+  .chart-box canvas { max-height: 260px; }
+
+  /* Expectancy */
+  .exp { background: #11161e; border: 1px solid #1c2333; border-radius: 8px; padding: 18px 20px; }
+  .exp h3 { font-size: 12px; color: #5c6675; letter-spacing: 1.5px; margin-bottom: 14px; }
+  .exp-row { display: flex; gap: 12px; align-items: center; margin-bottom: 14px; }
+  .exp select { background: #0a0e14; color: #c9d1d9; border: 1px solid #1c2333; padding: 6px 12px;
+                border-radius: 4px; font-size: 13px; }
+  .exp button { background: #1c3a5c; color: #58a6ff; border: 1px solid #2d4f7c; padding: 6px 18px;
+                border-radius: 4px; cursor: pointer; font-size: 13px; }
+  .exp button:hover { background: #254d7a; }
+  .exp .suff { font-size: 11px; padding: 3px 10px; border-radius: 3px; }
+  .suff-HIGH { background: rgba(63,185,80,0.12); color: #3fb950; }
+  .suff-MEDIUM { background: rgba(210,153,34,0.12); color: #d29922; }
+  .suff-LOW { background: rgba(248,81,73,0.12); color: #f85149; }
+  .suff-INSUFFICIENT { background: rgba(92,102,117,0.12); color: #5c6675; }
+  table { width: 100%; border-collapse: collapse; font-size: 13px; }
+  th { text-align: left; color: #5c6675; font-weight: 500; padding: 8px 10px; border-bottom: 1px solid #1c2333; }
+  td { padding: 7px 10px; border-bottom: 1px solid #0e1219; color: #8b949e; }
+  td strong { color: #e6edf3; }
+  .exp-summary { margin-top: 14px; font-size: 13px; color: #8b949e; padding: 10px 14px;
+                 background: #0d1117; border-radius: 6px; border-left: 3px solid #58a6ff; }
+  .exp-summary strong { color: #e6edf3; }
+</style>
+</head>
+<body>
+<div class="app">
+
+  <!-- Header -->
+  <div class="header">
+    <div>
+      <h1><span>Chan</span>Macro</h1>
+    </div>
+    <div class="time"><span class="dot"></span> <span id="update-time">加载中...</span></div>
+  </div>
+
+  <!-- Regime Hero -->
+  <div class="hero">
+    <div class="hero-card main">
+      <div>
+        <div class="regime-badge" id="hero-badge">—</div>
+        <div class="regime-name" id="hero-regime">—</div>
+      </div>
+      <div style="display:flex; gap:32px; margin-left:auto;">
+        <div class="hero-stat"><div class="val" id="hero-conf">—</div><div class="lbl">置信度</div></div>
+        <div class="hero-stat"><div class="val" id="hero-maturity">—</div><div class="lbl">成熟度</div></div>
+      </div>
+    </div>
+    <div class="hero-card" style="flex:1">
+      <div class="hero-stat"><div class="val" id="hero-ps">—</div><div class="lbl">价格结构</div></div>
+    </div>
+    <div class="hero-card" style="flex:1">
+      <div class="hero-stat"><div class="val" id="hero-br">—</div><div class="lbl">市场广度</div></div>
+    </div>
+  </div>
+
+  <!-- 4 Factor Cards -->
+  <div class="grid">
+    <div class="fcard">
+      <div class="title">价格结构 PRICE STRUCTURE</div>
+      <div class="score" id="f-price">—</div>
+      <div class="sub" id="f-price-sub"></div>
+      <div class="bar-wrap"><div class="bar" id="bar-price"></div></div>
+    </div>
+    <div class="fcard">
+      <div class="title">市场广度 BREADTH</div>
+      <div class="score" id="f-breadth">—</div>
+      <div class="sub" id="f-breadth-sub"></div>
+      <div class="bar-wrap"><div class="bar" id="bar-breadth"></div></div>
+    </div>
+    <div class="fcard">
+      <div class="title">持仓状态 OI MATRIX</div>
+      <div class="score" id="f-oi" style="font-size:24px">—</div>
+      <div class="sub" id="f-oi-sub"></div>
+    </div>
+    <div class="fcard">
+      <div class="title">波动率 VOLATILITY</div>
+      <div class="score" id="f-vol">—</div>
+      <div class="sub" id="f-vol-sub"></div>
+      <div class="bar-wrap"><div class="bar" id="bar-vol"></div></div>
+    </div>
+  </div>
+
+  <!-- Charts -->
+  <div class="charts">
+    <div class="chart-box"><h3>制度历史 REGIME HISTORY</h3><canvas id="chart-regime"></canvas></div>
+    <div class="chart-box"><h3>市场广度 BREADTH</h3><canvas id="chart-breadth"></canvas></div>
+  </div>
+
+  <!-- Expectancy -->
+  <div class="exp">
+    <h3>信号期望 SIGNAL EXPECTANCY</h3>
+    <div class="exp-row">
+      <select id="exp-signal">
+        <option value="B3">B3 · 三买</option><option value="B2">B2 · 二买</option><option value="B1">B1 · 一买</option>
+        <option value="S3">S3 · 三卖</option><option value="S2">S2 · 二卖</option><option value="S1">S1 · 一卖</option>
+      </select>
+      <button onclick="loadExpectancy()">查询</button>
+      <span class="suff" id="exp-sufficiency">—</span>
+    </div>
+    <table>
+      <thead><tr><th>层级</th><th>样本</th><th>有效样本</th><th>原始胜率</th><th>后验胜率</th><th>平均收益</th></tr></thead>
+      <tbody id="exp-layers"></tbody>
+    </table>
+    <div class="exp-summary" id="exp-summary"></div>
+  </div>
+</div>
+
+<script src="/static/js/dashboard.js"></script>
+</body>
+</html>

ChanPY.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.analysis.ChanPY import *  # noqa: F403

ChanPivotClassifier.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.analysis.ChanPivotClassifier import *  # noqa: F403

ChanPivotMonitor.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.analysis.ChanPivotMonitor import *  # noqa: F403

ChanSBI.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanSBI import *  # noqa: F403

ChanSEG.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanSEG import *  # noqa: F403

ChanZS.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.ChanZS import *  # noqa: F403

ChanZone.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.analysis.ChanZone import *  # noqa: F403

Chan_FX_Box.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.core.Chan_FX_Box import *  # noqa: F403

Find_Trend.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.analysis.Find_Trend import *  # noqa: F403

README.md

@@ -1,130 +0,0 @@
-# Chan — 缠论分析引擎
-
-把 OHLCV K 线拆成缠论结构（K 线单元 → 合并 K 线 → 分型 → 笔 → 线段 → 中枢 → 买卖点），
-配一个 TradingView Charting Library 的 Web 界面，外加一套验证信号有效性的回测脚本。
-
-标的不限：加密永续（ccxt）与 A 股（akshare）都走同一条分析链路。
-
-```
-chanlun/      缠论引擎，纯 pandas/numpy，无外部指标库依赖
-web/          Flask API + 图表界面
-research/     信号有效性验证脚本（step1 ~ step30）
-data/         本地 K 线（freqtrade 的 feather 格式）
-```
-
-## 安装
-
-需要 Python ≥ 3.11（pandas 3.x / numpy 2.x 的要求，不是本项目代码的限制）。
-
-```bash
-python -m venv .venv
-.venv/bin/pip install -r requirements.txt          # 核心运行时，8 个包
-.venv/bin/pip install -r requirements-dev.txt      # 另加测试与 research/ 所需
-```
-
-## 跑 Web
-
-```bash
-cd web
-../.venv/bin/python app.py        # 默认 http://0.0.0.0:8128
-```
-
-从仓库根跑 `.venv/bin/python web/app.py` 也可以——Python 会把脚本所在目录放进 `sys.path`。
-但**不能用 `python -m web.app`**，也不能 `import web.app`：`web/` 内部是无前缀导入
-（`import config`、`from api.analyze import bp`），`-m` 方式下 `sys.path` 里是仓库根而不是
-`web/`，会 `ModuleNotFoundError: No module named 'config'`。
-
-配置全部走环境变量，见 `web/config.py`：
-
-| 变量 | 默认值 | 用途 |
-|------|--------|------|
-| `FLASK_HOST` / `FLASK_PORT` | `0.0.0.0` / `8128` | 监听地址 |
-| `DATA_SERVICE_URL` | `https://provider.jackyu66.com` | 行情 REST 源 |
-| `DATA_SERVICE_WS_URL` | `wss://jackyu66.com/ws` | 行情 WebSocket 源 |
-| `ASHARE_DP_URL` | `http://103.179.242.166:8000` | A 股数据源 |
-| `CHAN_HTTP_PROXY` | 未设置则不走代理 | ccxt / HTTP 代理 |
-| `MACD_FACTOR` / `MACD_SMOOTH` | `1` / `1` | MACD 周期倍数，默认 12/26/9 |
-
-主要接口：`GET /api/analyze` 返回某标的某周期的完整缠论结构，`/api/klines/recent`
-取最新 K 线，`/api/trend_filter` 与 `/api/trend_detail` 做多周期趋势筛选，
-`/api/symbols`、`/api/search_stock`、`/api/sectors` 等负责标的检索。页面在 `/` 与 `/chan_tv`。
-
-## 作为库使用
-
-```python
-import pandas as pd
-from chanlun import TF_DF
-
-# 需要 date/open/high/low/close/volume 六列，date 为 datetime
-df = pd.read_feather("data/binance/futures/BTC_USDT_USDT-1h-futures.feather")
-
-tf = TF_DF(df, interval=1, timeframe="1h")
-
-len(tf.klu_list)    # K 线单元
-len(tf.klc_list)    # 合并 K 线（处理包含关系后）
-len(tf.bi_list)     # 笔
-len(tf.seg_list)    # 线段
-len(tf.zs_list)     # 中枢
-tf.bi_list[-1].dir  # Chan_BI_DIR.DOWN
-tf.chanmacd         # MACD 结构分析（背驰判定用）
-```
-
-**`interval` 的单位是分钟**，对传入的 df 做重采样；`interval=1` 是特例，表示原样使用、
-不重采样。所以拿 1h 的 feather 要传 `interval=1`，拿 1m 数据想看 1h 才传 `interval=60`：
-
-```python
-df1m = pd.read_feather("data/binance/futures/BTC_USDT_USDT-1m-futures.feather")
-TF_DF(df1m, interval=5,  timeframe="5m")
-TF_DF(df1m, interval=60, timeframe="1h")
-```
-
-传错不会报错，只会静默给出错误周期的结构——1h 数据配 `interval=4` 相当于按 4 分钟
-重采样，结果与 `interval=1` 完全相同。
-
-## 分析流程
-
-`TF_DF.init_TF_DF()` 按顺序做这几步，每步的实现在 `chanlun/pipeline/builders/` 下同名文件：
-
-1. `resample_to_interval` — 重采样（`interval != 1` 时）
-2. `add_indicators` — 追加 33 列指标（MACD / BBANDS / EMA / RSI / ATR 等）
-3. `cal_kl_data` → `klu_list` — K 线单元
-4. `get_klc_list` → `klc_list` — 按包含关系合并 K 线，并标记分型
-5. `cal_bi_list` → `bi_list` — 笔
-6. `cal_bi_zs_list_pure` → `bi_zs_list` — 笔中枢
-7. `get_seg_list` → `seg_list` — 线段
-8. `get_zs_list` / `get_big_zs_list` — 中枢与大级别中枢
-9. `ChanMACD(klu_list)` — MACD 段 / 柱堆结构，供背驰判定
-
-## 数据
-
-`data/<交易所>/futures/<SYMBOL>-<周期>-futures.feather`，即 freqtrade 的下载格式，
-如 `data/binance/futures/BTC_USDT_USDT-1h-futures.feather`。
-`research/lib/data.py` 负责定位：`BTC/USDT:USDT` + `1h` 会解析到上面这个路径，
-找不到本地文件则回落到远端拉取。
-
-## 测试
-
-```bash
-.venv/bin/python -m pytest chanlun/tests web/tests -q
-```
-
-`chanlun/tests/test_ta_compat.py` 有个**需要注意的陷阱**：它把 `chanlun/indicators/ta.py`
-的输出逐 bar 钉在 TA-Lib 上，但 **TA-Lib 不存在时会静默跳过**。也就是说改了 `ta.py`
-之后在没装 TA-Lib 的环境里跑，测试会显示通过，其实一项都没验证。改动那个文件时请先装：
-
-```bash
-sudo apt-get install -y libta-lib0 ta-lib-dev
-.venv/bin/pip install TA-Lib technical
-```
-
-## 已知问题
-
-- **`web/DEPLOY_GUIDE.md` 已失效**：它引用的 `deploy_venv.sh`、`stop_venv.sh`、
-  `status_venv.sh` 等 6 个脚本都在 `7f393b9` 精简提交里删掉了，目前没有部署脚本。
-  两个 systemd unit 文件（`web/chanlun-web*.service`）仍可参考，但它们用 gunicorn
-  且写死端口 8123，与 `config.py` 默认的 8128 不一致，gunicorn 也不在依赖清单里。
-- **`web/README.txt` 已过时**：它说的 `web/requirements.txt` 不存在，依赖清单在仓库根目录。
-- **四个零引用的死文件**：`chanlun/analysis/` 下的 `ChanPY.py`、`ChanLun_Classifier.py`、
-  `Find_Trend.py`、`ChanHeng.py` 全项目无人引用。`ChanPY.py` 依赖未安装的外部 chan.py 库，
-  另外三个需要 matplotlib / mplfinance / xgboost / scikit-learn——这些都**不在**依赖清单里，
-  是有意为之。要用得自行安装。

TF_DF.py

@@ -0,0 +1,2 @@
+"""兼容 shim — 请优先 from chanlun import ..."""
+from chanlun.pipeline.timeframe import TF_DF  # noqa: F401

chanlun/analysis/ChanLun_Classifier.py

@@ -14,10 +14,12 @@ from chanlun.core.ChanSBI import ChanSBI
 from chanlun.core.ChanSEG import ChanSEG
 from chanlun.core.ChanZS import ChanZS
 from chanlun.core.ChanBSP import ChanBSP
+import talib.abstract as ta
 import pandas as pd
 import matplotlib.pyplot as plt
 from matplotlib.dates import DateFormatter, date2num
 import matplotlib.patches as patches
+from technical.util import resample_to_interval
 from decimal import Decimal
 from chanlun.pipeline.orchestrator import ChanLun
 import xgboost as xgb

chanlun/analysis/Find_Trend.py

@@ -2,7 +2,7 @@ import ccxt
 import pandas as pd
 import numpy as np
 import mplfinance as mpf
-from chanlun.indicators import ta
+from talib import MACD, SMA
 from datetime import datetime, timedelta
 import logging
 import datetime as dt
@@ -249,9 +249,8 @@ def analyze_higher_timeframe(df_30m):
 # 8. Back-divergence detection (enhanced)
 def detect_back_divergence(df, strokes, higher_trend):
     try:
-        macd_df = ta.MACD(df['Close'], fastperiod=12, slowperiod=26, signalperiod=9)
+        macd, signal, hist = MACD(df['Close'], fastperiod=12, slowperiod=26, signalperiod=9)
-        macd, hist = macd_df['macd'], macd_df['macdhist']
+        sma20 = SMA(df['Close'], timeperiod=20)
-        sma20 = ta.SMA(df['Close'], timeperiod=20)
         df['macd'] = macd
         df['hist'] = hist
         df['sma20'] = sma20

chanlun/analysis/fast_bsp.py

@@ -1,384 +0,0 @@
-"""快速三类买卖点（引擎内称第四类，B4/S4）。
-
-原本长在 `research/lib/fast_bsp3.py`，现移入引擎作为唯一实现，研究脚本改为转发导入，
-这样回测口径与 web 图表永远一致。
-
-命名说明：缠论原文里没有「第四类买卖点」，但 B4/S4 也不只是「B3/S3 提前几根」——
-两者的选样口径不同，统计性质符号相反，故单独立类。形态条件是实时可判的：
-
-    中枢已成 -> 收盘突破 zg -> 收盘未跌回中枢 -> 重新上行
-
-最后一步发生的当根就能下单，滞后约 2 根，而引擎 B3/S3 要等 pullback_bi.sure_time，
-滞后 9~10 根。但差别不止滞后：
-
-  判据   引擎用笔端点事后判（回拉笔低点 >= zg），本函数用收盘价实时判
-  方向   引擎由离开笔方向决定，本函数由收盘从哪一侧突破决定
-  口径   627 个中枢里引擎发 625 个信号（几乎不筛），本函数只认 212 个（34%）；
-         被拒的多数是「中枢确认时价格早已离开、此后再没回来」的历史区间
-
-step30 同条件对拍（同一套 pure 笔中枢、同一组过滤器、同样的 1.5/3.0/48 出场）：
-
-                 原始 PF   +大级别同向   +同向+阶梯   胜率    t值
-    引擎 B3/S3     0.66      0.66         0.71      27.4%  -18.76
-    本函数 B4/S4   1.59      1.85         2.26      47.1%  +10.09
-
-同一组过滤器对 B4 有效、对 B3 无效，滞后差解释不了这一点（入场后移 1~4 根只是
-PF 3.18->2.53 的平滑衰减）。且 SL1.5/TP3.0 下随机入场胜率约 33%，引擎那 27.4%
-低于随机——它选中的是一批系统性反向的样本，不是「晚了所以差」。
-
-require_touch 控制是否强求回抽碰到中枢边界。step32/33 的实测表明强求反而更差：
-这等于排除掉「突破后一去不回头」的强势段，而那正是缠论里最强的趋势形态。
-故默认 False（笔数 +21%、滞后 -1 根、PF 2.26→2.37）。
-
-判据本身（收盘越过前一根极值）没有独立预测力：裸用 15 万笔样本 PF 0.95，
-把中枢换成「近20根高点」这类伪阻力位后 PF 0.90。alpha 全部来自中枢结构，
-判据只负责在这个已知价位上确认动能恢复。它也不能用于识别笔端点——
-入场前的回抽极值命中笔端点±2根的比例 19.3%，低于随机基准 22%。
-
-全部判定只使用当根及之前的数据，无未来函数。
-"""
-from __future__ import annotations
-
-import numpy as np
-import pandas as pd
-
-from chanlun.core.ChanEnum import Chan_FX_TYPE
-
-# 中枢的「可用时刻」取第几笔的确认时间。
-#   -1  = 最后一笔（历史默认）。中枢每吸收一根K线，最后一笔就可能后移，
-#         于是 available_ts 跟着漂——这是右边缘重画的根因（见 HANDOFF §5.41）。
-#    2  = 第三笔。三笔重叠即中枢成立，此后不再变，理论上更早也更稳。
-# ⛔ step47 已判定 bis[2] 作废（毛 R 0.933 → −0.000，见 HANDOFF §5.42）。
-# 开关保留只为可复现那次 A/B，**不要改默认值**。
-import os as _os
-
-AVAIL_BI_INDEX = -1
-
-
-def _resolve_avail_bi(avail_bi: int | None) -> int:
-    """优先级：显式入参 > 环境变量 CHAN_AVAIL_BI > 模块默认。
-
-    环境变量在调用时读取而非 import 时——ProcessPoolExecutor 在 fork 启动方式下
-    子进程会继承已 import 的模块，import 时读就固化成父进程的值了。
-    """
-    if avail_bi is not None:
-        return avail_bi
-    return int(_os.environ.get("CHAN_AVAIL_BI", AVAIL_BI_INDEX))
-
-
-def timestamps_ms(src: pd.DataFrame) -> np.ndarray:
-    """取毫秒时间戳。研究侧的 df 自带 timestamp，web 侧的不一定，故按 date 回退。
-
-    回退写法不能用 `date.astype("int64") // 10**6`：该值单位取决于列精度，
-    对毫秒精度的列会把时间戳砸平。
-    """
-    if "timestamp" in src.columns:
-        return src["timestamp"].to_numpy()
-    d = pd.to_datetime(src["date"])
-    if getattr(d.dt, "tz", None) is None:
-        d = d.dt.tz_localize("UTC")
-    return (d.dt.tz_convert("UTC").dt.tz_localize(None)
-             .astype("datetime64[ms]").astype("int64").to_numpy())
-
-
-def ensure_timestamp(df: pd.DataFrame) -> pd.DataFrame:
-    """保证 df 带 timestamp 列，缺失时补一份副本，不改动调用方的对象。"""
-    if "timestamp" in df.columns:
-        return df
-    out = df.copy()
-    out["timestamp"] = timestamps_ms(out)
-    return out
-
-
-def build_htf_zones(df_htf: pd.DataFrame, tf: str, chan=None, avail_bi: int | None = None) -> pd.DataFrame:
-    """算 pure 笔中枢，返回带生效时间的区间表。
-
-    available_ts —— 该中枢最早可被使用的时间戳（其确认时刻）。
-    传入已构建好的 chan 可避免重复跑一遍 pipeline（大数据集上省一半时间）。
-    """
-    if chan is None:
-        from chanlun import TF_DF
-
-        chan = TF_DF(df_htf, 1, tf)
-    zs_list = chan.cal_bi_zs_list_pure(chan.bi_list)
-    # 用引擎自己的 dataframe 对齐，避免调用方传入的 df 与引擎内部行数不一致
-    src = chan.dataframe if getattr(chan, "dataframe", None) is not None else df_htf
-    return zones_from_zs_list(zs_list, src, avail_bi=avail_bi)
-
-
-def zones_from_zs_list(zs_list, src: pd.DataFrame, avail_bi: int | None = None) -> pd.DataFrame:
-    """把已算好的 pure 笔中枢转成区间表。
-
-    调用方手工跑过 cal_bi_zs_list_pure 时走这里，免得再算一遍（web 的 analyze_chan
-    就是这种用法）。
-    """
-    ts_of = dict(zip(src["date"].dt.strftime("%Y-%m-%d %H:%M:%S"), timestamps_ms(src)))
-    # 中枢可用时刻取第几笔。在循环外解析一次，别让每个中枢都去读一遍环境变量。
-    i = _resolve_avail_bi(avail_bi)
-
-    rows = []
-    for zs in zs_list:
-        bis = getattr(zs, "bi_list", [])
-        if not bis:
-            continue
-        # 笔数不够时退回最后一笔（i=2 需要至少 3 笔才成立）
-        key_bi = bis[i] if (i == -1 or len(bis) > i) else bis[-1]
-        sure_key = str(getattr(key_bi, "sure_time", "") or "")
-        end_key = str(getattr(key_bi, "end_time", "") or "")
-        avail = ts_of.get(sure_key) or ts_of.get(end_key)
-        if avail is None:
-            continue
-        start_key = str(bis[0].start_time)
-        rows.append({
-            "zg": float(zs.zg), "zd": float(zs.zd),
-            "gg": float(getattr(zs, "gg", zs.zg)), "dd": float(getattr(zs, "dd", zs.zd)),
-            "start_ts": ts_of.get(start_key, avail),
-            "available_ts": int(avail),
-        })
-    out = pd.DataFrame(rows)
-    return out.sort_values("available_ts").reset_index(drop=True) if not out.empty else out
-
-
-def add_zone_ladder(zones: pd.DataFrame) -> pd.DataFrame:
-    """标注每个中枢相对前一个中枢是否同向推进（缠论「趋势 vs 盘整」）。
-
-    z_above / z_below 分别对应向上、向下推进。买信号要求 z_above、卖信号要求 z_below
-    时，30m/2h 的 PF 从 2.72 升到 3.41。
-    """
-    out = zones.copy()
-    if out.empty:
-        out["z_above"], out["z_below"] = pd.Series(dtype=bool), pd.Series(dtype=bool)
-        return out
-    pg, pdn = out["zg"].shift(), out["zd"].shift()
-    out["z_above"] = (out["zd"] > pg).fillna(False)
-    out["z_below"] = (out["zg"] < pdn).fillna(False)
-    return out
-
-
-def htf_fx_timeline(chan_htf, df_htf: pd.DataFrame | None = None) -> pd.DataFrame:
-    """把大级别分型压成一条按确认时间排序的时间线。
-
-    confirm_ts 是该分型最早可被使用的时刻。timestamp 是K线开盘时刻，而分型要等这根K线
-    收盘才算数，所以整体后移一个大级别周期；否则小级别会提前一整根大级别K线拿到信号。
-
-    只取方向与确认时刻——同向过滤用不到背驰强度，省掉 MACD 面积计算。
-    """
-    src = chan_htf.dataframe if getattr(chan_htf, "dataframe", None) is not None else df_htf
-    if src is None or len(src) == 0:
-        return pd.DataFrame(columns=["confirm_ts", "fx_ts", "direction", "price"])
-    ts = timestamps_ms(src)
-    idx_of = {t: i for i, t in enumerate(src["date"].dt.strftime("%Y-%m-%d %H:%M:%S"))}
-    period = int(np.median(np.diff(ts))) if len(ts) > 1 else 0
-
-    rows = []
-    for klc in getattr(chan_htf, "klc_list", []):
-        if klc.fx not in (Chan_FX_TYPE.TOP, Chan_FX_TYPE.BOTTOM):
-            continue
-        if klc.next is None or klc.next.end_klu is None:
-            continue
-        e_key, c_key = str(klc.end_time), str(klc.next.end_klu.time)
-        if e_key not in idx_of or c_key not in idx_of:
-            continue
-        fx_idx, confirm_idx = idx_of[e_key], idx_of[c_key]
-        if confirm_idx <= fx_idx:
-            continue
-        d = 1 if klc.fx == Chan_FX_TYPE.BOTTOM else -1
-        rows.append({
-            "confirm_ts": int(ts[confirm_idx]) + period,
-            "fx_ts": int(ts[fx_idx]),
-            "direction": d,
-            "price": float(klc.low if d == 1 else klc.high),
-        })
-    out = pd.DataFrame(rows)
-    if out.empty:
-        return pd.DataFrame(columns=["confirm_ts", "fx_ts", "direction", "price"])
-    return out.sort_values("confirm_ts").reset_index(drop=True)
-
-
-def attach_htf_agree(sig: pd.DataFrame, df_ltf: pd.DataFrame, tl: pd.DataFrame) -> pd.DataFrame:
-    """给每个小级别信号挂上「入场时刻之前最近的大级别分型」是否同向。
-
-    产出 htf_dir（+1 底 / -1 顶）与 htf_agree（1 同向 / 0 反向 / NaN 无可用分型）。
-    """
-    out = sig.copy()
-    if sig.empty or tl.empty:
-        out["htf_dir"] = np.nan
-        out["htf_agree"] = np.nan
-        return out
-
-    ts_ltf = timestamps_ms(df_ltf)
-    entry_ts = ts_ltf[out["entry_idx"].to_numpy().astype(int)]
-    k = np.searchsorted(tl["confirm_ts"].to_numpy(), entry_ts, side="right") - 1
-    valid = k >= 0
-    k_safe = np.clip(k, 0, len(tl) - 1)
-
-    fx_dir = tl["direction"].to_numpy()[k_safe].astype(float)
-    out["htf_dir"] = np.where(valid, fx_dir, np.nan)
-    out["htf_agree"] = np.where(
-        valid, (fx_dir == out["direction"].to_numpy()).astype(float), np.nan
-    )
-    return out
-
-
-def attach_zone_ladder(sig: pd.DataFrame, zones: pd.DataFrame) -> pd.DataFrame:
-    """按信号方向取该中枢的阶梯标记：买看 z_above、卖看 z_below。
-
-    zone_i 是 find_fast_bsp3 内 enumerate 出的位置序号，故用 iloc 定位。
-    """
-    out = sig.copy()
-    if sig.empty:
-        out["ladder_ok"] = pd.Series(dtype=bool)
-        return out
-    z = zones if "z_above" in zones.columns else add_zone_ladder(zones)
-    above = z["z_above"].to_numpy()
-    below = z["z_below"].to_numpy()
-    zi = out["zone_i"].to_numpy().astype(int)
-    ok = np.where(out["direction"].to_numpy() == 1, above[zi], below[zi])
-    out["ladder_ok"] = ok.astype(bool)
-    return out
-
-
-def find_fast_bsp3(
-    df: pd.DataFrame,
-    zones: pd.DataFrame,
-    scan: int = 200,
-    pullback_win: int = 30,
-    tol: float = -1.0,
-    max_per_zone: int = 1,
-    diag: dict | None = None,
-    require_touch: bool = False,
-) -> pd.DataFrame:
-    """扫描每个中枢，找突破后回抽不回中枢的入场点。
-
-    zones 需含 zg / zd / available_ts，且 available_ts 已是可用时刻。
-    max_per_zone > 1 时，同一中枢在首次入场后继续往后找二次、三次突破回抽，
-    用来检验「趋势里同一中枢反复给机会」是否值得做。
-
-    tol 是「算作回抽中」的边界容差。require_touch=False 时它不再是入场门槛，
-    仅决定哪些K线被视为回抽中而跳过转强判定，故 tol 越大入场越晚。
-    默认负值 = 完全禁用该跳过，突破后每根都检查转强，滞后压到 2.2 根。
-    step34 实测滞后与收益严格单调：9.9根 PF1.88 / 5.8根 2.37 / 2.2根 2.86。
-
-    返回列：
-      entry_idx   实时可下单的K线
-      direction   +1 三买 / -1 三卖
-      bo_idx      突破根
-      pb_idx      回抽极值根
-      lag         entry_idx - bo_idx
-      depth       回抽深度（相对中枢边界，负值表示曾插入中枢）
-      occ         这是该中枢的第几次入场
-    """
-    if zones.empty:
-        return pd.DataFrame()
-
-    ts = df["timestamp"].to_numpy()
-    close = df["close"].to_numpy(dtype=float)
-    high = df["high"].to_numpy(dtype=float)
-    low = df["low"].to_numpy(dtype=float)
-    n = len(df)
-    rows = []
-
-    def note(key: str) -> None:
-        if diag is not None:
-            diag[key] = diag.get(key, 0) + 1
-
-    for zone_i, (_, z) in enumerate(zones.iterrows()):
-        note("中枢总数")
-        zg, zd = float(z["zg"]), float(z["zd"])
-        if zg <= zd:
-            note("×无效中枢")
-            continue
-        start = int(np.searchsorted(ts, z["available_ts"], side="left"))
-        if start >= n - 2:
-            note("×中枢太靠后")
-            continue
-        # 允许多次入场时按比例放宽扫描窗口，否则后几次机会会被窗口截断
-        scan_end = min(start + scan * max_per_zone, n)
-        cursor = start
-
-        for occ in range(1, max_per_zone + 1):
-            if cursor >= n - 2:
-                break
-            # 第一步：找突破。要求突破前确实待在中枢内，避免把远处的价格当突破。
-            was_inside = False
-            bo_idx, d = None, 0
-            for j in range(cursor, scan_end):
-                c = close[j]
-                if zd <= c <= zg:
-                    was_inside = True
-                    continue
-                if not was_inside:
-                    continue
-                bo_idx, d = j, (1 if c > zg else -1)
-                break
-            if bo_idx is None:
-                if occ == 1:
-                    note("×窗口内未突破")
-                break
-
-            edge = zg if d == 1 else zd
-            # 第二步：突破后监控回抽，回抽不跌回中枢且重新顺势 -> 入场
-            touched = False
-            pb_idx = None
-            pb_ext = None
-            entry_idx = None
-            fell_back = False
-            for j in range(bo_idx + 1, min(bo_idx + pullback_win + 1, n)):
-                # 收盘跌回中枢 -> 突破失效
-                if zd <= close[j] <= zg:
-                    fell_back = True
-                    break
-                # 回抽触及边界附近（允许 tol 的毛刺）
-                near = (low[j] <= edge * (1 + tol)) if d == 1 else (high[j] >= edge * (1 - tol))
-                if near:
-                    touched = True
-                    ext = low[j] if d == 1 else high[j]
-                    if pb_ext is None or ((ext < pb_ext) if d == 1 else (ext > pb_ext)):
-                        pb_ext, pb_idx = ext, j
-                    continue
-                # 回抽后重新顺势：收盘创出前一根之上（三买）/ 之下（三卖）
-                # require_touch=False 时不强求回抽碰到中枢边界，
-                # 这样「突破后一去不回头」的强势段也能收进来。
-                if (touched and pb_idx is not None) or not require_touch:
-                    go = close[j] > high[j - 1] if d == 1 else close[j] < low[j - 1]
-                    if go:
-                        entry_idx = j
-                        break
-            if entry_idx is None:
-                if occ == 1:
-                    note("×突破后跌回中枢" if fell_back
-                         else "×回抽未触及边界" if not touched
-                         else "×触及边界但未转强")
-                # 这次突破没走成，从突破点之后继续找下一次
-                cursor = bo_idx + 1
-                continue
-            if pb_ext is None:
-                # 未触及边界就转强（require_touch=False）：取突破至入场间的实际极值
-                seg = slice(bo_idx + 1, entry_idx + 1)
-                pb_ext = float(low[seg].min() if d == 1 else high[seg].max())
-                pb_idx = int((low[seg].argmin() if d == 1 else high[seg].argmax())
-                             + bo_idx + 1)
-            if occ == 1:
-                note("√成交")
-
-            # 回抽深度：>0 表示未插入中枢，越大表示回抽越浅
-            depth = (pb_ext - zg) / zg if d == 1 else (zd - pb_ext) / zd
-            rows.append({
-                "entry_idx": entry_idx, "direction": d,
-                "bo_idx": bo_idx, "pb_idx": pb_idx,
-                "lag": entry_idx - bo_idx,
-                "depth": depth,
-                "zg": zg, "zd": zd,
-                "width_pct": (zg - zd) / close[bo_idx],
-                "occ": occ,
-                "zone_i": zone_i,
-            })
-            cursor = entry_idx + 1
-
-    out = pd.DataFrame(rows)
-    if out.empty:
-        return out
-    # 相邻中枢可能突破到同一根K线，同一时刻只能有一个仓位，保留最早成型的那个
-    return (out.sort_values(["entry_idx", "occ", "zone_i"])
-               .drop_duplicates("entry_idx", keep="first")
-               .reset_index(drop=True))

chanlun/analysis/wyckoff/__init__.py

@@ -0,0 +1,7 @@
+"""威科夫分析（启发式）：交易区间 / 阶段 / 事件 / Volume Profile / Live。"""
+from __future__ import annotations
+
+from .engine import analyze_wyckoff
+from .live import execution_signal_from_wyckoff
+
+__all__ = ["analyze_wyckoff", "execution_signal_from_wyckoff"]

chanlun/analysis/wyckoff/engine.py

@@ -0,0 +1,196 @@
+"""威科夫分析入口：Cycle → Phase → Event → VP + Live（MULTI-CYCLE / LIVE-STRUCTURE）。
+
+range.py 只产 TradingRange；Confirmed 走 events.py；Live 走 live.py。
+cycles[0]=ACTIVE；禁止 cycles[-1] 取 active。
+Execution 只消费 Confirmed（见 live.execution_signal_from_wyckoff）。
+"""
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional
+
+import pandas as pd
+
+from .events import build_phases, detect_bias_and_events
+from .live import analyze_live_structure
+from .range import detect_trading_ranges
+from .volume_profile import compute_volume_profile
+
+
+def _fmt_time(v) -> Optional[str]:
+	if v is None:
+		return None
+	if hasattr(v, "isoformat"):
+		try:
+			return v.isoformat()
+		except Exception:
+			pass
+	return str(v)
+
+
+def _empty(vp_bins: int) -> Dict[str, Any]:
+	return {
+		"cycles": [],
+		"trading_range": None,
+		"bias": "unknown",
+		"phases": [],
+		"events": [],
+		"volume_profile": {"bins": [], "poc": None, "vah": None, "val": None, "bin_count": vp_bins},
+		"volume_confirm": {"avg_volume": 0.0, "event_checks": {}},
+		"live": None,
+	}
+
+
+def _confidence_for_confirmed(
+	tr: Dict[str, Any],
+	phases: List[Dict[str, Any]],
+	events: List[Dict[str, Any]],
+) -> Dict[str, float]:
+	range_c = float(tr.get("range_confidence") or 0.5)
+	labels = {p.get("phase") for p in phases}
+	phase_c = 0.35
+	if "A" in labels and "B" in labels:
+		phase_c += 0.15
+	if "C" in labels:
+		phase_c += 0.2
+	if "D" in labels or "E" in labels:
+		phase_c += 0.15
+	phase_c = min(0.95, phase_c)
+	types = {e.get("type") for e in events}
+	event_c = 0.25
+	for t in ("Spring", "UTAD", "SOS", "SOW", "LPS", "LPSY"):
+		if t in types:
+			event_c += 0.12
+	event_c = min(0.95, event_c)
+	overall = 0.4 * range_c + 0.3 * phase_c + 0.3 * event_c
+	return {
+		"range": round(range_c, 3),
+		"phase": round(phase_c, 3),
+		"event": round(event_c, 3),
+		"overall": round(overall, 3),
+	}
+
+
+def _build_cycle(
+	work: pd.DataFrame,
+	tr: Dict[str, Any],
+	cycle_id: int,
+	vp_bins: int,
+) -> Dict[str, Any]:
+	bias, events, volume_confirm = detect_bias_and_events(work, tr)
+	phases = build_phases(work, tr, bias, events)
+	vp = compute_volume_profile(
+		work,
+		int(tr["abs_start_idx"]),
+		int(tr["abs_end_idx"]),
+		bin_count=vp_bins,
+	)
+	for ev in events:
+		ev["time"] = _fmt_time(ev.get("time"))
+	for ph in phases:
+		ph["start_time"] = _fmt_time(ph.get("start_time"))
+		ph["end_time"] = _fmt_time(ph.get("end_time"))
+
+	is_active = cycle_id == 0
+	trading_range = {
+		"start_time": _fmt_time(tr.get("start_time")),
+		"end_time": _fmt_time(tr.get("end_time")),
+		"high": float(tr["high"]),
+		"low": float(tr["low"]),
+		"mid": float(tr["mid"]),
+		"active": bool(is_active),
+		"bars": int(tr.get("bars", 0)),
+	}
+	conf = _confidence_for_confirmed(tr, phases, events)
+
+	# Live 层：仅 ACTIVE 周期做推演；历史周期归档为 COMPLETED
+	if is_active:
+		live = analyze_live_structure(
+			work, tr, confirmed_events=events, confirmed_phases=phases, bias=bias,
+		)
+		lifecycle = live.get("lifecycle") or "FORMING"
+	else:
+		live = None
+		lifecycle = "COMPLETED"
+
+	return {
+		"id": int(cycle_id),
+		"role": "latest" if is_active else "historical",
+		# MULTI-CYCLE：时间线角色
+		"status": "ACTIVE" if is_active else "HISTORICAL",
+		# LIVE-STRUCTURE：生命周期
+		"lifecycle": lifecycle,
+		"direction": "latest" if is_active else "historical",
+		"period": {
+			"start_time": _fmt_time(tr.get("start_time")),
+			"end_time": _fmt_time(tr.get("end_time")),
+			"bars": int(tr.get("bars", 0)),
+		},
+		"confidence": conf,
+		"trading_range": trading_range,
+		"bias": bias,
+		# 兼容旧读法：顶层 phases/events = confirmed
+		"phases": phases,
+		"events": events,
+		"confirmed": {
+			"phases": phases,
+			"events": events,
+			"volume_confirm": volume_confirm,
+		},
+		"live": live,
+		"volume_profile": vp,
+		"volume_confirm": volume_confirm,
+	}
+
+
+def analyze_wyckoff(
+	df: pd.DataFrame,
+	lookback: int = 120,
+	vp_bins: int = 50,
+	min_bars: int = 24,
+	atr_mult: float = 1.2,
+	range_start_time=None,
+	prefer_start_time=None,
+	max_cycles: int = 8,
+) -> Dict[str, Any]:
+	"""
+	多周期威科夫分析。
+	cycles[0] = ACTIVE；顶层 phases/events 只镜像 Confirmed。
+	顶层 live 镜像 cycles[0].live。
+	"""
+	empty = _empty(vp_bins)
+	if df is None or len(df) < 30:
+		return empty
+	if not all(c in df.columns for c in ("open", "high", "low", "close")):
+		return empty
+	work = df.copy()
+	if "volume" not in work.columns:
+		work["volume"] = 1.0
+
+	trs = detect_trading_ranges(
+		work,
+		lookback=lookback,
+		min_bars=max(8, int(min_bars)),
+		atr_mult=atr_mult,
+		max_cycles=max(1, min(8, int(max_cycles))),
+		prefer_start_time=prefer_start_time,
+		range_start_time=range_start_time,
+	)
+	if not trs:
+		return empty
+
+	cycles: List[Dict[str, Any]] = []
+	for i, tr in enumerate(trs):
+		cycles.append(_build_cycle(work, tr, cycle_id=i, vp_bins=vp_bins))
+
+	active = cycles[0]
+	return {
+		"cycles": cycles,
+		"trading_range": active["trading_range"],
+		"bias": active["bias"],
+		"phases": active["confirmed"]["phases"],
+		"events": active["confirmed"]["events"],
+		"volume_profile": active["volume_profile"],
+		"volume_confirm": active["volume_confirm"],
+		"live": active.get("live"),
+		"lifecycle": active.get("lifecycle"),
+	}

chanlun/analysis/wyckoff/events.py

@@ -0,0 +1,369 @@
+"""威科夫阶段与事件（启发式）。"""
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional, Tuple
+
+import numpy as np
+import pandas as pd
+
+
+def _bar_time(df: pd.DataFrame, i: int):
+	row = df.iloc[i]
+	if "date" in df.columns and pd.notna(row["date"]):
+		return row["date"]
+	if "timestamp" in df.columns:
+		return row["timestamp"]
+	return i
+
+
+def _avg_vol(df: pd.DataFrame, i: int, win: int = 20) -> float:
+	a = max(0, i - win + 1)
+	v = df["volume"].astype(float).iloc[a : i + 1]
+	m = float(v.mean()) if len(v) else 0.0
+	return m if m > 0 else 1.0
+
+
+def detect_bias_and_events(
+	df: pd.DataFrame,
+	tr: Dict[str, Any],
+) -> Tuple[str, List[Dict[str, Any]], Dict[str, Any]]:
+	"""
+	返回 bias、events、volume_confirm。
+
+	Spring/UTAD 相对「结构高低」判定：取区间内次低/次高（剔除单根极值），
+	避免箱体把假破低点吃进 lo 后永远刺不破、从而无 C 阶段。
+	"""
+	hi = float(tr["high"])
+	lo = float(tr["low"])
+	mid = float(tr["mid"])
+	tol = float(tr.get("tol") or (hi - lo) * 0.05)
+	s = int(tr["abs_start_idx"])
+	e = int(tr["abs_end_idx"])
+	events: List[Dict[str, Any]] = []
+
+	# 结构边界：用次低/次高作假破参照（至少 8 根才启用）
+	seg = df.iloc[s : e + 1]
+	event_lo, event_hi = lo, hi
+	if len(seg) >= 8:
+		lows = seg["low"].astype(float)
+		highs = seg["high"].astype(float)
+		# nsmallest(2) 的较大者 = 次低；nlargest(2) 的较小者 = 次高
+		event_lo = float(lows.nsmallest(min(2, len(lows))).iloc[-1])
+		event_hi = float(highs.nlargest(min(2, len(highs))).iloc[-1])
+		# 勿比公布箱沿更「松」：结构带应在箱内
+		event_lo = max(event_lo, lo)
+		event_hi = min(event_hi, hi)
+		# 若次低仍等于极值（多根同价），略抬参照便于识别收回
+		if abs(event_lo - lo) < 1e-12:
+			event_lo = lo + max(tol * 0.35, (hi - lo) * 0.02)
+		if abs(event_hi - hi) < 1e-12:
+			event_hi = hi - max(tol * 0.35, (hi - lo) * 0.02)
+
+	# 扫描区间内及之后（含 tail_reserve）
+	scan_end = int(tr.get("abs_scan_end_idx", min(len(df) - 1, e + 15)))
+	scan_end = min(len(df) - 1, max(scan_end, e))
+	spring = None
+	utad = None
+	sos = None
+	sod = None  # sign of weakness / distribution breakdown
+	lps = None
+	lpsy = None
+
+	for i in range(s + 2, scan_end + 1):
+		row = df.iloc[i]
+		low = float(row["low"])
+		high = float(row["high"])
+		close = float(row["close"])
+		vol = float(row["volume"]) if "volume" in df.columns else 0.0
+		avg_v = _avg_vol(df, i)
+		ratio = vol / avg_v if avg_v else 0.0
+
+		# Spring: pierce below structural support then close back
+		if spring is None and low < event_lo - tol * 0.35 and close >= event_lo - tol * 0.35:
+			vol_ok = ratio <= 1.35 or (i + 1 <= scan_end and float(df.iloc[min(i + 1, scan_end)]["volume"]) / avg_v < 1.2)
+			spring = {
+				"type": "Spring",
+				"time": _bar_time(df, i),
+				"price": low,
+				"note": "假破下沿后收回",
+				"volume_ratio": round(ratio, 3),
+				"volume_ok": bool(vol_ok),
+				"idx": i,
+			}
+
+		# UTAD: pierce above structural resistance then close back
+		if utad is None and high > event_hi + tol * 0.35 and close <= event_hi + tol * 0.35:
+			vol_ok = ratio >= 0.8
+			utad = {
+				"type": "UTAD",
+				"time": _bar_time(df, i),
+				"price": high,
+				"note": "假破上沿后跌回",
+				"volume_ratio": round(ratio, 3),
+				"volume_ok": bool(vol_ok),
+				"idx": i,
+			}
+
+		# SOS: close above high with volume
+		if sos is None and close > hi + tol * 0.15:
+			vol_ok = ratio >= 1.15
+			sos = {
+				"type": "SOS",
+				"time": _bar_time(df, i),
+				"price": close,
+				"note": "放量上破交易区间",
+				"volume_ratio": round(ratio, 3),
+				"volume_ok": bool(vol_ok),
+				"idx": i,
+			}
+
+		# SOW / breakdown
+		if sod is None and close < lo - tol * 0.15:
+			vol_ok = ratio >= 1.15
+			sod = {
+				"type": "SOW",
+				"time": _bar_time(df, i),
+				"price": close,
+				"note": "放量下破交易区间",
+				"volume_ratio": round(ratio, 3),
+				"volume_ok": bool(vol_ok),
+				"idx": i,
+			}
+
+	# LPS after SOS: pullback that holds above mid/high-band with lighter volume
+	if sos is not None:
+		si = int(sos["idx"])
+		for i in range(si + 1, min(len(df), si + 25)):
+			row = df.iloc[i]
+			low = float(row["low"])
+			close = float(row["close"])
+			vol = float(row["volume"]) if "volume" in df.columns else 0.0
+			avg_v = _avg_vol(df, i)
+			ratio = vol / avg_v if avg_v else 0.0
+			if low >= mid - tol and close >= hi - tol * 2:
+				vol_ok = ratio <= 1.05
+				lps = {
+					"type": "LPS",
+					"time": _bar_time(df, i),
+					"price": low,
+					"note": "突破后缩量回踩不破",
+					"volume_ratio": round(ratio, 3),
+					"volume_ok": bool(vol_ok),
+					"idx": i,
+				}
+				break
+
+	if sod is not None:
+		si = int(sod["idx"])
+		for i in range(si + 1, min(len(df), si + 25)):
+			row = df.iloc[i]
+			high = float(row["high"])
+			close = float(row["close"])
+			vol = float(row["volume"]) if "volume" in df.columns else 0.0
+			avg_v = _avg_vol(df, i)
+			ratio = vol / avg_v if avg_v else 0.0
+			if high <= mid + tol and close <= lo + tol * 2:
+				vol_ok = ratio <= 1.05
+				lpsy = {
+					"type": "LPSY",
+					"time": _bar_time(df, i),
+					"price": high,
+					"note": "下跌突破后缩量反抽不过",
+					"volume_ratio": round(ratio, 3),
+					"volume_ok": bool(vol_ok),
+					"idx": i,
+				}
+				break
+
+	# 冲突清理：已判定吸筹且有 SOS 时，丢弃更早的 UTAD（避免阶段/图面误导）
+	# 派发且有 SOW 时，丢弃更晚才合理的 Spring 假信号同理在偏置后再滤
+	keep = []
+	for ev in (spring, sos, lps, utad, sod, lpsy):
+		if not ev:
+			continue
+		keep.append(ev)
+
+	# bias（先算）
+	last_c = float(df["close"].iloc[-1])
+	bias = "unknown"
+	if sos and (not sod or int(sos.get("idx", 0)) >= int(sod.get("idx", 0))):
+		bias = "accumulation"
+	elif sod and (not sos or int(sod.get("idx", 0)) > int(sos.get("idx", 0))):
+		bias = "distribution"
+	elif spring and not utad:
+		bias = "accumulation"
+	elif utad and not spring:
+		bias = "distribution"
+	elif last_c >= mid:
+		bias = "accumulation"
+	else:
+		bias = "distribution"
+
+	filtered = []
+	for ev in keep:
+		if bias == "accumulation" and ev["type"] == "UTAD" and sos and int(ev["idx"]) <= int(sos["idx"]):
+			continue
+		if bias == "distribution" and ev["type"] == "Spring" and sod and int(ev["idx"]) <= int(sod["idx"]):
+			continue
+		filtered.append(ev)
+
+	events = [{k: v for k, v in ev.items() if k != "idx"} for ev in filtered]
+
+	avg_volume = float(df["volume"].astype(float).iloc[max(0, e - 20) : e + 1].mean()) if "volume" in df.columns else 0.0
+	volume_confirm = {
+		"avg_volume": avg_volume,
+		"event_checks": {ev["type"]: {"volume_ok": ev.get("volume_ok"), "volume_ratio": ev.get("volume_ratio")} for ev in events},
+	}
+	return bias, events, volume_confirm
+
+
+def build_phases(
+	df: pd.DataFrame,
+	tr: Dict[str, Any],
+	bias: str,
+	events: List[Dict[str, Any]],
+	min_bars: int = 3,
+) -> List[Dict[str, Any]]:
+	"""
+	按威科夫事件锚点切分 A–E（启发式）。
+
+	吸筹：A停止 → B筑底 → C测试(Spring) → D拉升(SOS…LPS) → E离开
+	派发：A停止 → B筑顶 → C测试(UTAD) → D派发(SOW…LPSY) → E离开
+
+	无 Spring/UTAD 时：若已有 SOS/SOW，用突破前末次沿带测试补 C；仍无则省略 C。
+	"""
+	s = int(tr["abs_start_idx"])
+	e = int(tr["abs_end_idx"])
+	hi = float(tr["high"])
+	lo = float(tr["low"])
+	n_last = len(df) - 1
+	min_span = max(2, min_bars - 1)
+	range_len = max(1, e - s)
+
+	def _match_idx(t) -> Optional[int]:
+		if t is None:
+			return None
+		lo = max(0, s - 2)
+		hi = min(len(df), e + 40)
+		for i in range(lo, hi):
+			if _bar_time(df, i) == t:
+				return i
+		try:
+			tt = pd.Timestamp(t)
+			sample = None
+			if "date" in df.columns and len(df):
+				sample = df["date"].iloc[min(s, n_last)]
+			if sample is not None and getattr(sample, "tzinfo", None) is not None and tt.tzinfo is None:
+				tt = tt.tz_localize(sample.tzinfo)
+			for i in range(lo, hi):
+				bt = _bar_time(df, i)
+				try:
+					if abs((pd.Timestamp(bt) - tt).total_seconds()) <= 1:
+						return i
+				except Exception:
+					continue
+		except Exception:
+			pass
+		return None
+
+	event_idx: Dict[str, int] = {}
+	for ev in events:
+		idx = _match_idx(ev.get("time"))
+		if idx is not None:
+			event_idx[str(ev.get("type"))] = idx
+
+	accum = bias != "distribution"
+	if accum:
+		c_ev = event_idx.get("Spring")
+		d_ev = event_idx.get("SOS")
+		d_tail = event_idx.get("LPS") or d_ev
+	else:
+		c_ev = event_idx.get("UTAD")
+		d_ev = event_idx.get("SOW")
+		d_tail = event_idx.get("LPSY") or d_ev
+
+	# 有 D 无明确测试事件时：用突破前最后一次触及下/上沿作为 C（次级测试）
+	if c_ev is None and d_ev is not None:
+		band = lo + (hi - lo) * 0.28 if accum else hi - (hi - lo) * 0.28
+		for i in range(int(d_ev) - 1, s + 1, -1):
+			row = df.iloc[i]
+			if accum and float(row["low"]) <= band:
+				c_ev = i
+				break
+			if not accum and float(row["high"]) >= band:
+				c_ev = i
+				break
+
+	def _lab(phase: str) -> str:
+		if accum:
+			m = {"A": "A停止下跌", "B": "B筑底", "C": "C测试", "D": "D拉升", "E": "E离开"}
+		else:
+			m = {"A": "A停止上涨", "B": "B筑顶", "C": "C测试", "D": "D派发", "E": "E离开"}
+		return m.get(phase, phase)
+
+	a_end = s + max(min_bars, range_len // 5)
+
+	c_start = c_end = None
+	if c_ev is not None:
+		c_start = max(s, int(c_ev) - 1)
+		c_end = min(n_last, int(c_ev) + 1)
+
+	if d_ev is not None:
+		d_start = int(d_ev)
+		d_end = min(n_last, max(int(d_tail or d_ev), d_start) + max(min_bars, range_len // 8))
+		if d_tail is not None:
+			d_end = max(d_end, min(n_last, int(d_tail) + 1))
+	else:
+		d_start = d_end = None
+
+	if c_start is not None:
+		b_end = max(a_end + 1, c_start)
+	elif d_start is not None:
+		b_end = max(a_end + 1, d_start)
+	else:
+		b_end = max(a_end + 1, e)
+
+	if d_end is not None:
+		e_start = min(n_last, d_end)
+		e_end = n_last
+	else:
+		e_start = e_end = None
+
+	raw = [("A", s, a_end), ("B", a_end, b_end)]
+	if c_start is not None and c_end is not None:
+		raw.append(("C", c_start, c_end))
+	if d_start is not None and d_end is not None:
+		raw.append(("D", d_start, d_end))
+	if e_start is not None and e_end is not None and e_end > e_start:
+		raw.append(("E", e_start, e_end))
+
+	phases: List[Dict[str, Any]] = []
+	cursor = s
+	for phase, _a, _b in raw:
+		if cursor >= n_last:
+			break
+		a = max(int(_a), cursor)
+		b = int(max(int(_b), a))
+		need = 1 if phase == "C" else min_span
+		if b < a + need:
+			b = min(n_last, a + need)
+		b = int(np.clip(b, a, n_last))
+		if b < a:
+			continue
+		if phases and phases[-1].get("_a") == a and phases[-1].get("_b") == b:
+			continue
+		phases.append(
+			{
+				"phase": phase,
+				"label": _lab(phase),
+				"start_time": _bar_time(df, a),
+				"end_time": _bar_time(df, b),
+				"_a": a,
+				"_b": b,
+			}
+		)
+		cursor = b
+	for p in phases:
+		p.pop("_a", None)
+		p.pop("_b", None)
+	return phases

chanlun/analysis/wyckoff/live.py

@@ -0,0 +1,258 @@
+"""威科夫 Live / Developing 层（WYCKOFF-LIVE-STRUCTURE-001）。
+
+独立于 Confirmed Engine：不修改 events 确认条件，不写入 confirmed.events。
+Execution 不得消费本模块输出。
+"""
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional, Set
+
+import numpy as np
+import pandas as pd
+
+
+def _avg_vol(df: pd.DataFrame, i: int, win: int = 20) -> float:
+	a = max(0, i - win + 1)
+	v = df["volume"].astype(float).iloc[a : i + 1]
+	m = float(v.mean()) if len(v) else 0.0
+	return m if m > 0 else 1.0
+
+
+def _empty_live() -> Dict[str, Any]:
+	return {
+		"lifecycle": "UNKNOWN",
+		"range_formation": None,
+		"phase_candidate": None,
+		"event_candidates": [],
+		"next_expected": None,
+		"confidence": {
+			"cycle": 0.0,
+			"phase": 0.0,
+			"event": 0.0,
+			"structure": 0.0,
+			"volume": 0.0,
+			"overall": 0.0,
+		},
+		"note": "",
+	}
+
+
+def analyze_live_structure(
+	df: pd.DataFrame,
+	tr: Optional[Dict[str, Any]],
+	confirmed_events: Optional[List[Dict[str, Any]]] = None,
+	confirmed_phases: Optional[List[Dict[str, Any]]] = None,
+	bias: str = "unknown",
+) -> Dict[str, Any]:
+	"""
+	基于当前 TradingRange 与已确认事件，推演 Live candidates。
+	confirmed_* 只读，用于避免重复提示已确认事件，不修改之。
+	"""
+	out = _empty_live()
+	if df is None or len(df) < 20 or tr is None:
+		out["note"] = "insufficient structure"
+		return out
+
+	confirmed_events = confirmed_events or []
+	confirmed_phases = confirmed_phases or []
+	confirmed_types: Set[str] = {str(e.get("type")) for e in confirmed_events if e.get("type")}
+
+	s = int(tr["abs_start_idx"])
+	e = int(tr["abs_end_idx"])
+	scan_end = int(tr.get("abs_scan_end_idx", len(df) - 1))
+	scan_end = min(len(df) - 1, max(scan_end, e))
+	hi = float(tr["high"])
+	lo = float(tr["low"])
+	mid = float(tr["mid"])
+	tol = float(tr.get("tol") or (hi - lo) * 0.05)
+	atr = float(tr.get("atr") or max((hi - lo) * 0.2, 1e-9))
+
+	seg = df.iloc[s : e + 1]
+	if len(seg) < 8:
+		out["note"] = "range too short"
+		return out
+
+	# —— Range Formation（横盘 / 波动收敛）——
+	closes = seg["close"].astype(float)
+	highs = seg["high"].astype(float)
+	lows = seg["low"].astype(float)
+	vols = seg["volume"].astype(float) if "volume" in seg.columns else pd.Series([1.0] * len(seg))
+	half = max(4, len(seg) // 2)
+	vol_early = float(np.std(closes.iloc[:half])) if half > 1 else 0.0
+	vol_late = float(np.std(closes.iloc[-half:])) if half > 1 else 0.0
+	width = hi - lo
+	width_atr = width / atr if atr > 0 else 99.0
+	converging = vol_early > 1e-12 and vol_late < vol_early * 0.85
+	range_ok = 1.2 <= width_atr <= 10.0 and len(seg) >= 16
+	structure_score = 0.35
+	if range_ok:
+		structure_score += 0.25
+	if converging:
+		structure_score += 0.2
+	if width_atr <= 6.0:
+		structure_score += 0.1
+	structure_score = float(min(0.95, structure_score))
+
+	out["range_formation"] = {
+		"potential_trading_range": bool(range_ok),
+		"converging": bool(converging),
+		"width_atr": round(width_atr, 3),
+		"bars": int(len(seg)),
+	}
+
+	# —— 最近 K 形态（Phase C / Event candidates）——
+	i = scan_end
+	row = df.iloc[i]
+	o = float(row["open"])
+	h = float(row["high"])
+	l = float(row["low"])
+	c = float(row["close"])
+	rng = max(h - l, 1e-9)
+	lower_wick = min(o, c) - l
+	upper_wick = h - max(o, c)
+	avg_v = _avg_vol(df, i)
+	vol = float(row["volume"]) if "volume" in df.columns else avg_v
+	vol_ratio = vol / avg_v if avg_v else 1.0
+	volume_score = float(np.clip(1.1 - abs(vol_ratio - 1.0) * 0.35, 0.2, 0.95))
+
+	phase_candidate = None
+	phase_conf = 0.0
+	# Phase C：测低 + 下影 + 缩量（吸筹语境）
+	near_lo = l <= lo + tol * 1.2
+	test_low = l < mid and lower_wick >= rng * 0.35
+	vol_contract = vol_ratio <= 1.05
+	if bias != "distribution" and near_lo and test_low and vol_contract:
+		phase_candidate = "C"
+		phase_conf = 0.55 + (0.1 if lower_wick >= rng * 0.5 else 0) + (0.08 if vol_ratio < 0.9 else 0)
+	# Phase D 候选：价格在箱上半、有上破意图但未确认 SOS
+	elif c >= mid and (h >= hi - tol or c > hi - tol * 0.5):
+		phase_candidate = "D"
+		phase_conf = 0.5 + (0.1 if c > mid else 0)
+	elif c < mid and (l <= lo + tol):
+		phase_candidate = "B"
+		phase_conf = 0.45
+
+	# 已有 confirmed phase 时，candidate 取「下一阶段」提示，不覆盖事实
+	confirmed_phase_set = {str(p.get("phase")) for p in confirmed_phases}
+	if "E" in confirmed_phase_set:
+		phase_candidate = phase_candidate or "E"
+		phase_conf = max(phase_conf, 0.7)
+	elif "D" in confirmed_phase_set and phase_candidate is None:
+		phase_candidate = "D"
+		phase_conf = max(phase_conf, 0.65)
+
+	out["phase_candidate"] = phase_candidate
+	phase_conf = float(min(0.92, phase_conf))
+
+	# —— Event candidates（仅 Spring / SOS / LPS / UTAD）——
+	candidates: List[Dict[str, Any]] = []
+
+	def _add(typ: str, conf: float, note: str) -> None:
+		if typ in confirmed_types:
+			return  # 已确认则不再作为 candidate
+		candidates.append(
+			{
+				"type": typ,
+				"confidence": round(float(min(0.9, conf)), 3),
+				"confirmed": False,
+				"note": note,
+			}
+		)
+
+	# Spring candidate：刺破或贴近下沿，收盘收回，但未达 Confirmed 规则（或不在 confirmed）
+	pierce_lo = l < lo - tol * 0.15
+	close_back = c >= lo - tol * 0.5
+	if pierce_lo and close_back:
+		_add("Spring", 0.5 + (0.12 if vol_ratio <= 1.2 else 0) + (0.08 if close_back else 0), "假破下沿收回（未确认）")
+	elif l <= lo + tol * 0.35 and close_back and lower_wick >= rng * 0.4:
+		_add("Spring", 0.45 + (0.1 if vol_contract else 0), "测下沿长下影（未确认）")
+
+	# UTAD candidate
+	pierce_hi = h > hi + tol * 0.15
+	close_back_dn = c <= hi + tol * 0.5
+	if pierce_hi and close_back_dn:
+		_add("UTAD", 0.5 + (0.1 if vol_ratio >= 0.9 else 0), "假破上沿跌回（未确认）")
+
+	# SOS candidate：接近/轻破上沿，量能一般，未确认
+	if c > hi - tol * 0.4 or h >= hi:
+		sos_conf = 0.48 + (0.12 if c > hi else 0) + (0.1 if vol_ratio >= 1.05 else 0)
+		_add("SOS", sos_conf, "上破/逼近箱顶（未确认）")
+
+	# LPS candidate：站上 mid/上沿带后回踩
+	if c >= mid and l >= mid - tol * 1.5 and l > lo + (hi - lo) * 0.25:
+		_add("LPS", 0.46 + (0.1 if vol_ratio <= 1.0 else 0), "箱内上沿带回踩（未确认）")
+
+	candidates.sort(key=lambda x: x["confidence"], reverse=True)
+	out["event_candidates"] = candidates[:4]
+
+	event_score = float(candidates[0]["confidence"]) if candidates else 0.25
+
+	# next_expected（简规则）
+	next_exp = None
+	if "Spring" in confirmed_types and "SOS" not in confirmed_types:
+		next_exp = "SOS"
+	elif "SOS" in confirmed_types and "LPS" not in confirmed_types:
+		next_exp = "LPS"
+	elif "UTAD" in confirmed_types and "SOW" not in confirmed_types:
+		next_exp = "SOW"
+	elif any(c["type"] == "Spring" for c in candidates):
+		next_exp = "Test"
+	elif any(c["type"] == "SOS" for c in candidates):
+		next_exp = "LPS"
+	out["next_expected"] = next_exp
+
+	# —— lifecycle ——
+	key_confirmed = confirmed_types & {"Spring", "SOS", "UTAD", "SOW", "LPS", "LPSY"}
+	if key_confirmed:
+		lifecycle = "CONFIRMED"
+	elif range_ok or phase_candidate or candidates:
+		lifecycle = "FORMING"
+	else:
+		lifecycle = "UNKNOWN"
+	out["lifecycle"] = lifecycle
+
+	cycle_c = structure_score
+	overall = 0.35 * cycle_c + 0.25 * phase_conf + 0.25 * event_score + 0.15 * volume_score
+	out["confidence"] = {
+		"cycle": round(cycle_c, 3),
+		"phase": round(phase_conf, 3),
+		"event": round(event_score, 3),
+		"structure": round(structure_score, 3),
+		"volume": round(volume_score, 3),
+		"overall": round(float(overall), 3),
+	}
+	parts = []
+	if out["range_formation"]["potential_trading_range"]:
+		parts.append("Potential TR")
+	if phase_candidate:
+		parts.append(f"Phase {phase_candidate} candidate")
+	if candidates:
+		parts.append(f"{candidates[0]['type']} candidate")
+	out["note"] = "; ".join(parts) if parts else "observing"
+	return out
+
+
+def execution_signal_from_wyckoff(payload: Dict[str, Any]) -> Optional[Dict[str, Any]]:
+	"""
+	Execution 边界：只允许 Confirmed。
+	返回 source='confirmed' 的信号描述；Live-only 时返回 None。
+	"""
+	if not payload:
+		return None
+	cycles = payload.get("cycles") or []
+	active = cycles[0] if cycles else None
+	events = []
+	if active and isinstance(active.get("confirmed"), dict):
+		events = list(active["confirmed"].get("events") or [])
+	if not events:
+		# 兼容旧顶层 events（均为 confirmed 镜像）
+		events = list(payload.get("events") or [])
+	if not events:
+		return None
+	last = events[-1]
+	return {
+		"source": "confirmed",
+		"type": last.get("type"),
+		"time": last.get("time"),
+		"lifecycle": (active or {}).get("lifecycle") or "CONFIRMED",
+	}

chanlun/analysis/wyckoff/range.py

@@ -0,0 +1,442 @@
+"""交易区间检测：仅负责 TradingRange（起止/高低/结构分）。
+
+WYCKOFF-MULTI-CYCLE-001：Phase/Event/VP 不得进入本模块。
+过滤顺序固定：detect → quality → trend → overlap(<0.2) → accept → mask。
+"""
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional, Tuple
+
+import numpy as np
+import pandas as pd
+
+MAX_CYCLES = 8
+OVERLAP_RATIO_MAX = 0.2
+
+
+def _atr(df: pd.DataFrame, period: int = 14) -> pd.Series:
+	high = df["high"].astype(float)
+	low = df["low"].astype(float)
+	close = df["close"].astype(float)
+	prev_close = close.shift(1)
+	tr = pd.concat(
+		[
+			(high - low).abs(),
+			(high - prev_close).abs(),
+			(low - prev_close).abs(),
+		],
+		axis=1,
+	).max(axis=1)
+	return tr.rolling(period, min_periods=max(3, period // 2)).mean()
+
+
+def _robust_width(seg: pd.DataFrame) -> float:
+	"""用 90/10 分位估宽，避免单根影线把长窗卡死。"""
+	h = seg["high"].astype(float)
+	l = seg["low"].astype(float)
+	if len(seg) < 6:
+		return float(h.max() - l.min())
+	return float(np.nanpercentile(h, 90) - np.nanpercentile(l, 10))
+
+
+def _score_segment(
+	length: int,
+	near_hi: int,
+	near_lo: int,
+	inside: float,
+	width: float,
+	atr: float,
+) -> float:
+	"""结构质量分（非 Phase/Event）。"""
+	touch = min(near_hi, 6) + min(near_lo, 6)
+	width_pen = (width / atr) if atr > 0 else width
+	return float(touch) * 4.0 + float(inside) * 25.0 - width_pen * 3.0 + min(length / 40.0, 2.0)
+
+
+def _time_col(df: pd.DataFrame) -> Optional[str]:
+	if "date" in df.columns:
+		return "date"
+	if "timestamp" in df.columns:
+		return "timestamp"
+	return None
+
+
+def _bar_index_at_or_after(work: pd.DataFrame, ts: Any) -> Optional[int]:
+	col = _time_col(work)
+	if col is None or ts is None:
+		return None
+	try:
+		target = pd.Timestamp(ts)
+	except Exception:
+		return None
+	series = pd.to_datetime(work[col], utc=True, errors="coerce")
+	if target.tzinfo is None:
+		target = target.tz_localize("UTC")
+	else:
+		target = target.tz_convert("UTC")
+	if series.isna().all():
+		return None
+	ge = series >= target
+	if ge.any():
+		return int(np.flatnonzero(ge.to_numpy())[0])
+	return 0
+
+
+def _pack_range(
+	work: pd.DataFrame,
+	df: pd.DataFrame,
+	start_i: int,
+	end_i: int,
+	hi: float,
+	lo: float,
+	tol: float,
+	last_atr: float,
+	score: float,
+	n: int,
+	window_offset: int = 0,
+) -> Dict[str, Any]:
+	"""组装 TradingRange（仅结构字段）。"""
+	mid = (hi + lo) / 2.0
+	last_c = float(work["close"].iloc[min(end_i, len(work) - 1)])
+	price_in_box = (lo - tol * 1.5) <= last_c <= (hi + tol * 1.5)
+	bars = int(end_i - start_i + 1)
+	# 结构置信：归一化 score（启发式）
+	range_conf = float(np.clip(score / 55.0, 0.05, 0.99))
+	best = {
+		"start_idx": int(start_i),
+		"end_idx": int(end_i),
+		"high": float(hi),
+		"low": float(lo),
+		"mid": float(mid),
+		"active": bool(price_in_box),
+		"atr": float(last_atr),
+		"tol": float(tol),
+		"bars": bars,
+		"score": float(score),
+		"quality": float(score),
+		"range_confidence": range_conf,
+	}
+
+	def _ts(row) -> Any:
+		col = _time_col(work)
+		if col and pd.notna(row[col]):
+			return row[col]
+		return None
+
+	best["start_time"] = _ts(work.iloc[best["start_idx"]])
+	best["end_time"] = _ts(work.iloc[best["end_idx"]])
+	# window_offset：slice 相对父 DataFrame 的起点；勿用 len(df)-len(work)
+	offset = int(window_offset)
+	best["abs_start_idx"] = offset + best["start_idx"]
+	best["abs_end_idx"] = offset + best["end_idx"]
+	best["abs_scan_end_idx"] = offset + n - 1
+	return best
+
+
+def _overlap_ratio(a0: int, a1: int, b0: int, b1: int) -> float:
+	"""两闭区间重叠长度 / 较短区间长度。"""
+	lo = max(a0, b0)
+	hi = min(a1, b1)
+	if hi < lo:
+		return 0.0
+	overlap = hi - lo + 1
+	shorter = min(a1 - a0 + 1, b1 - b0 + 1)
+	if shorter <= 0:
+		return 0.0
+	return float(overlap) / float(shorter)
+
+
+def _passes_quality(tr: Dict[str, Any], min_bars: int) -> bool:
+	if tr is None:
+		return False
+	if int(tr.get("bars") or 0) < max(8, min_bars // 2):
+		return False
+	if float(tr.get("score") or 0) < 12.0:
+		return False
+	hi = float(tr["high"])
+	lo = float(tr["low"])
+	atr = float(tr.get("atr") or 0) or 1.0
+	if (hi - lo) / atr > 12.0:
+		return False
+	return True
+
+
+def _passes_trend_filter(work: pd.DataFrame, tr: Dict[str, Any]) -> bool:
+	"""趋势污染：定向位移过大则非震荡箱。"""
+	s = int(tr["start_idx"])
+	e = int(tr["end_idx"])
+	seg = work.iloc[s : e + 1]
+	if len(seg) < 8:
+		return False
+	c0 = float(seg["close"].iloc[0])
+	c1 = float(seg["close"].iloc[-1])
+	atr = float(tr.get("atr") or 0) or 1.0
+	drift = abs(c1 - c0) / atr
+	# 相对箱宽：漂移占箱宽过大 → 趋势
+	width = max(float(tr["high"]) - float(tr["low"]), atr)
+	drift_frac = abs(c1 - c0) / width
+	if drift > 6.0 and drift_frac > 0.55:
+		return False
+	return True
+
+
+def _detect_in_window(
+	df: pd.DataFrame,
+	win_start: int,
+	win_end: int,
+	min_bars: int = 24,
+	atr_mult: float = 1.2,
+	tail_reserve: int = 12,
+	prefer_start_time: Any = None,
+	range_start_time: Any = None,
+) -> Optional[Dict[str, Any]]:
+	"""
+	在 df[win_start:win_end+1] 内检测单个 TradingRange。
+	只返回箱体结构，不含 Phase/Event/VP。
+	"""
+	if df is None or win_end < win_start:
+		return None
+	slice_df = df.iloc[win_start : win_end + 1].reset_index(drop=True)
+	lookback = len(slice_df)
+	if lookback < min_bars + 5:
+		return None
+
+	work = slice_df
+	n = len(work)
+	reserve = min(tail_reserve, max(0, n - min_bars - 2))
+	core_end = n - reserve if reserve > 0 else n
+	core = work.iloc[:core_end]
+	if len(core) < min_bars:
+		core = work
+		core_end = n
+		reserve = 0
+
+	atr = _atr(work)
+	last_atr = float(atr.iloc[core_end - 1]) if atr.notna().iloc[:core_end].any() else float(
+		(core["high"] - core["low"]).mean()
+	)
+	if not np.isfinite(last_atr) or last_atr <= 0:
+		last_atr = float(core["close"].iloc[-1]) * 0.01
+
+	eff_atr_mult = float(atr_mult)
+	if lookback >= 280:
+		eff_atr_mult = atr_mult * 1.7
+	elif lookback >= 160:
+		eff_atr_mult = atr_mult * 1.3
+	width_factor = 3.8 + min(2.2, max(0.0, (lookback - 80) / 100.0))
+	max_width = last_atr * eff_atr_mult * width_factor
+	tol = last_atr * eff_atr_mult * 0.35
+
+	prefer_i = None
+	if prefer_start_time is not None:
+		prefer_i = _bar_index_at_or_after(work, prefer_start_time)
+
+	if range_start_time is not None:
+		start_i = _bar_index_at_or_after(work, range_start_time)
+		if start_i is not None and start_i <= core_end - 8:
+			seg = work.iloc[start_i:core_end]
+			hi = float(seg["high"].max())
+			lo = float(seg["low"].min())
+			rw = _robust_width(seg)
+			if 0 < rw <= max_width * 1.15:
+				near_hi = int((seg["high"] >= hi - tol).sum())
+				near_lo = int((seg["low"] <= lo + tol).sum())
+				inside = float(((seg["close"] >= lo - tol) & (seg["close"] <= hi + tol)).mean())
+				if near_hi >= 2 and near_lo >= 2 and inside >= 0.70:
+					score = _score_segment(len(seg), near_hi, near_lo, inside, rw, last_atr)
+					return _pack_range(
+						work, df, start_i, core_end - 1, hi, lo, tol, last_atr, score, n,
+						window_offset=win_start,
+					)
+
+	eff_min_bars = max(8, int(min_bars))
+	cn = len(core)
+	max_bars = min(cn, max(eff_min_bars * 2, min(96, max(eff_min_bars + 8, int(cn * 0.5)))))
+	cands: List[Tuple[float, int, int, int, float, float, float]] = []
+
+	def _try_seg(start_i: int, end_i: int, prefer_boost: float = 0.0) -> None:
+		if end_i - start_i + 1 < eff_min_bars:
+			return
+		if start_i < 0 or end_i >= cn or start_i > end_i:
+			return
+		seg = work.iloc[start_i : end_i + 1]
+		hi = float(seg["high"].max())
+		lo = float(seg["low"].min())
+		rw = _robust_width(seg)
+		if rw <= 0 or rw > max_width:
+			return
+		raw_w = hi - lo
+		if raw_w > max_width * 1.35:
+			return
+		near_hi = int((seg["high"] >= hi - tol).sum())
+		near_lo = int((seg["low"] <= lo + tol).sum())
+		if near_hi < 2 or near_lo < 2:
+			return
+		inside = float(((seg["close"] >= lo - tol) & (seg["close"] <= hi + tol)).mean())
+		if inside < 0.72:
+			return
+		length = end_i - start_i + 1
+		score = _score_segment(length, near_hi, near_lo, inside, rw, last_atr) + prefer_boost
+		cands.append((score, length, start_i, end_i, hi, lo, rw))
+
+	for length in range(min(cn, max_bars), eff_min_bars - 1, -4):
+		start_i = cn - length
+		boost = 0.0
+		if prefer_i is not None:
+			dist = abs(start_i - int(prefer_i))
+			if dist <= 6:
+				boost = 10.0
+			elif dist <= 14:
+				boost = 4.0
+			elif start_i > int(prefer_i) + 16:
+				boost = -10.0
+		_try_seg(start_i, cn - 1, boost)
+
+	if prefer_i is not None:
+		pi = int(prefer_i)
+		if 0 <= pi < cn:
+			align_max = min(cn, max(max_bars, int(cn * 0.65)))
+			alen = cn - pi
+			if eff_min_bars <= alen <= align_max:
+				_try_seg(pi, cn - 1, prefer_boost=18.0)
+			elif alen > align_max:
+				start_i = max(0, cn - align_max)
+				if start_i > pi:
+					start_i = pi
+					end_i = min(cn - 1, pi + align_max - 1)
+				else:
+					end_i = cn - 1
+				_try_seg(start_i, end_i, prefer_boost=12.0)
+
+	if not cands:
+		return None
+
+	cands.sort(key=lambda x: x[0], reverse=True)
+	best_score = cands[0][0]
+	band = max(4.0, abs(best_score) * 0.10)
+	near = [c for c in cands if c[0] >= best_score - band]
+	chosen = max(near, key=lambda x: (x[1], x[0]))
+	score, _length, start_i, end_i, hi, lo, _rw = chosen
+	return _pack_range(work, df, start_i, end_i, hi, lo, tol, last_atr, score, n, window_offset=win_start)
+
+
+def detect_trading_ranges(
+	df: pd.DataFrame,
+	lookback: Optional[int] = None,
+	min_bars: int = 24,
+	atr_mult: float = 1.2,
+	tail_reserve: int = 12,
+	max_cycles: int = MAX_CYCLES,
+	prefer_start_time: Any = None,
+	range_start_time: Any = None,
+) -> List[Dict[str, Any]]:
+	"""
+	倒序切多段 TradingRange（近→远）。
+	过滤顺序：detect → quality → trend → overlap → accept → mask。
+	返回列表已按时间倒序，调用方将 [0] 标为 ACTIVE。
+	"""
+	if df is None or len(df) < min_bars + 5:
+		return []
+	lb = int(lookback) if lookback is not None else len(df)
+	work = df.tail(lb).reset_index(drop=True)
+	n = len(work)
+	occupied: List[Dict[str, Any]] = []
+	accepted: List[Dict[str, Any]] = []
+
+	# 搜索右端从 n-1 往左收缩；每接受一段后右端移到该段 start 之前
+	search_end = n - 1
+	prefer = prefer_start_time
+	hard_start = range_start_time
+
+	while len(accepted) < max(1, int(max_cycles)) and search_end >= min_bars + 4:
+		# 在剩余历史内从右往左试多个右边界，避免历史箱必须贴住 search_end
+		# （否则中间趋势会挡住更早的真实箱）
+		cand = None
+		step = max(4, min(12, (search_end - min_bars) // 10 or 4))
+		for end_try in range(search_end, min_bars + 4, -step):
+			trial = _detect_in_window(
+				work,
+				0,
+				end_try,
+				min_bars=min_bars,
+				atr_mult=atr_mult,
+				tail_reserve=tail_reserve,
+				prefer_start_time=prefer if len(accepted) == 0 and end_try == search_end else None,
+				range_start_time=hard_start if len(accepted) == 0 and end_try == search_end else None,
+			)
+			# 1) detect
+			if trial is None:
+				continue
+			# 2) quality
+			if not _passes_quality(trial, min_bars):
+				continue
+			# 3) trend contamination
+			if not _passes_trend_filter(work, trial):
+				continue
+			# 4) overlap with accepted
+			a0, a1 = int(trial["abs_start_idx"]), int(trial["abs_end_idx"])
+			overlap_bad = False
+			for occ in occupied:
+				ratio = _overlap_ratio(a0, a1, int(occ["start"]), int(occ["end"]))
+				if ratio >= OVERLAP_RATIO_MAX:
+					overlap_bad = True
+					break
+			if overlap_bad:
+				continue
+			# 取最靠右的合格箱（倒序第一段）
+			cand = trial
+			break
+
+		if cand is None:
+			break
+
+		# 5) accept
+		accepted.append(cand)
+		a0, a1 = int(cand["abs_start_idx"]), int(cand["abs_end_idx"])
+		# 6) mask
+		occupied.append(
+			{
+				"start": a0,
+				"end": max(a1, int(cand.get("abs_scan_end_idx", a1))),
+				"quality": float(cand.get("quality") or 0),
+				"high": float(cand["high"]),
+				"low": float(cand["low"]),
+			}
+		)
+		# 下一轮只在更早窗口搜
+		search_end = int(cand["abs_start_idx"]) - 1
+		hard_start = None
+		prefer = None
+
+	# abs_* 目前相对 work；若 df 比 work 长需加 offset
+	offset = len(df) - len(work)
+	if offset:
+		for tr in accepted:
+			tr["abs_start_idx"] = int(tr["abs_start_idx"]) + offset
+			tr["abs_end_idx"] = int(tr["abs_end_idx"]) + offset
+			tr["abs_scan_end_idx"] = int(tr["abs_scan_end_idx"]) + offset
+
+	return accepted
+
+
+def detect_trading_range(
+	df: pd.DataFrame,
+	lookback: int = 120,
+	min_bars: int = 24,
+	atr_mult: float = 1.2,
+	tail_reserve: int = 12,
+	range_start_time: Any = None,
+	prefer_start_time: Any = None,
+) -> Optional[Dict[str, Any]]:
+	"""兼容旧接口：返回倒序列表中的第一段（ACTIVE 候选）。"""
+	ranges = detect_trading_ranges(
+		df,
+		lookback=lookback,
+		min_bars=min_bars,
+		atr_mult=atr_mult,
+		tail_reserve=tail_reserve,
+		max_cycles=1,
+		prefer_start_time=prefer_start_time,
+		range_start_time=range_start_time,
+	)
+	return ranges[0] if ranges else None

chanlun/analysis/wyckoff/volume_profile.py

@@ -0,0 +1,72 @@
+"""区间内 Volume Profile。"""
+from __future__ import annotations
+
+from typing import Any, Dict, List
+
+import numpy as np
+import pandas as pd
+
+
+def compute_volume_profile(
+	df: pd.DataFrame,
+	start_idx: int,
+	end_idx: int,
+	bin_count: int = 50,
+	value_area_pct: float = 0.70,
+) -> Dict[str, Any]:
+	seg = df.iloc[start_idx : end_idx + 1]
+	if seg.empty:
+		return {"bins": [], "poc": None, "vah": None, "val": None, "bin_count": bin_count}
+
+	typical = (seg["high"].astype(float) + seg["low"].astype(float) + seg["close"].astype(float)) / 3.0
+	vol = seg["volume"].astype(float).fillna(0.0)
+	lo = float(seg["low"].min())
+	hi = float(seg["high"].max())
+	if not np.isfinite(lo) or not np.isfinite(hi) or hi <= lo:
+		mid = float(seg["close"].iloc[-1])
+		return {
+			"bins": [{"price": mid, "volume": float(vol.sum())}],
+			"poc": mid,
+			"vah": mid,
+			"val": mid,
+			"bin_count": 1,
+		}
+
+	edges = np.linspace(lo, hi, bin_count + 1)
+	# 右开最后一桶闭合
+	idx = np.clip(np.digitize(typical.values, edges) - 1, 0, bin_count - 1)
+	vols = np.zeros(bin_count, dtype=float)
+	for i, v in zip(idx, vol.values):
+		vols[i] += float(v)
+
+	centers = (edges[:-1] + edges[1:]) / 2.0
+	poc_i = int(np.argmax(vols)) if vols.sum() > 0 else bin_count // 2
+	poc = float(centers[poc_i])
+
+	# Value Area：从 POC 向两侧扩展直到累计 >= value_area_pct
+	total = float(vols.sum()) or 1.0
+	target = total * value_area_pct
+	left = right = poc_i
+	acc = float(vols[poc_i])
+	while acc < target and (left > 0 or right < bin_count - 1):
+		left_v = vols[left - 1] if left > 0 else -1.0
+		right_v = vols[right + 1] if right < bin_count - 1 else -1.0
+		if right_v >= left_v and right < bin_count - 1:
+			right += 1
+			acc += float(vols[right])
+		elif left > 0:
+			left -= 1
+			acc += float(vols[left])
+		else:
+			break
+
+	bins: List[Dict[str, float]] = [
+		{"price": float(centers[i]), "volume": float(vols[i])} for i in range(bin_count)
+	]
+	return {
+		"bins": bins,
+		"poc": poc,
+		"vah": float(centers[right]),
+		"val": float(centers[left]),
+		"bin_count": bin_count,
+	}

chanlun/core/ChanBI.py

@@ -15,12 +15,10 @@ class ChanBI():
 		self.sure_time = None
 		self.klc_list = []
 		self.klc_list.append(klc)
-		self._klc_idx = {klc.index}
 		self.end_time = klc.end_time
 		self.start_time = klc.start_time
-		self._macd_hist = 0
+		self.macd_hist = 0
-		self._macd_div = 0
+		self.macd_div = 0
-		self._macd_dirty = False
 		self.seg = None
 		self.height = 0
 		self.width = 0
@@ -35,57 +33,26 @@ class ChanBI():
 	def set_seg(self, seg):
 		self.seg = seg
 		self.seg_index = len(seg.bi_list)-1
-	# macd_hist / macd_div 改为惰性。原来 add_klc 每加一根 KLC 就把整笔的
-	# 所有 KLU 重新累加一遍，是 O(k²)；而这两个值只有背驰判定（bsp.py）在读，
-	# lean 模式下 bsp 根本不算，等于全程白算。这里只标脏，取值时才算。
-	# 语义不变：klc_list 只增不减（set_start_klc 会重置并同时清脏），
-	# dir 在 klc 累加期间固定，所以延后到读取时算与逐次重算结果相同。
-	@property
-	def macd_hist(self):
-		if self._macd_dirty:
-			self.cal_macdhist()
-		return self._macd_hist
-
-	@macd_hist.setter
-	def macd_hist(self, v):
-		self._macd_hist = v
-		self._macd_dirty = False
-
-	@property
-	def macd_div(self):
-		self.cal_macd_div()
-		return self._macd_div
-
-	@macd_div.setter
-	def macd_div(self, v):
-		self._macd_div = v
-
 	def set_macdhist(self, macd_hist):
 		self.macd_hist = macd_hist
 	def set_macd_div(self, macd_div):
 		self.macd_div = macd_div
 	def cal_macd_div(self):
-		self._macd_div = 0.0
+		self.macd_div = 0.0
 		if self.pre and self.pre.pre:
 			if self.pre.pre.macd_hist == 0:
-				self._macd_div = 0.0
+				self.macd_div = 0.0
 			else:
-				self._macd_div = self.macd_hist / self.pre.pre.macd_hist
+				self.macd_div = self.macd_hist / self.pre.pre.macd_hist
 			#print(self.start_time, self.end_time, self.macd_hist, self.pre.pre.macd_hist, self.macd_div)
 	def cal_macdhist(self):
-		self._macd_dirty = False
+		self.macd_hist = 0
-		self._macd_hist = 0
-		up = self.dir == Chan_BI_DIR.UP
-		acc = 0
 		for klc in self.klc_list:
 			for klu in klc.klu_list:
-				h = klu.macdhist
+				if self.dir == Chan_BI_DIR.UP and klu.macdhist > 0: 
-				if up:
+					self.macd_hist += klu.macdhist
-					if h > 0:
+				if self.dir == Chan_BI_DIR.DOWN and klu.macdhist < 0:
-						acc += h
+					self.macd_hist -= klu.macdhist
-				elif h < 0:
-					acc -= h
-		self._macd_hist = acc
 	def check_bi_zs_overlap(self):
 		if self.next and self.next.next:
 			if self.dir == Chan_BI_DIR.UP:
@@ -128,10 +95,6 @@ class ChanBI():
 		self.start_klc = klc
 		self.klc_list = []
 		self.klc_list.append(klc)
-		# klc_list 被整个换掉，去重集合与惰性缓存都要跟着重置，
-		# 否则后续 add_klc 会以为旧下标还在里面而漏加
-		self._klc_idx = {klc.index}
-		self._macd_dirty = True
 		self.high = klc.high
 		self.low = klc.low
 		self.dir = ddir
@@ -140,14 +103,20 @@ class ChanBI():
 	def set_next(self, bi):
 		self.next = bi
 	def add_klc(self, klc):
-		# 去重原本是对 klc_list 线性扫描，配合下面每次全量重算的 macdhist，
+		added = False
-		# 让「往一笔里加 k 根 KLC」变成 O(k²)。改用下标集合，O(1)。
+		if len(self.klc_list) > 0:
-		if klc.index not in self._klc_idx:
+			for index in range(0, len(self.klc_list)):
-			self._klc_idx.add(klc.index)
+				if self.klc_list[index].index == klc.index:
+					added = True
+					break
+		if not added:
 			self.klc_list.append(klc)
+			#print(self.start_time, klc.start_time)
+		#print(klc.end_time, klc.index)
 		self.end_klc = klc
 		self.end_time = klc.klu_list[-1].time
-		self._macd_dirty = True
+		self.cal_macdhist()
+		self.cal_macd_div()
 	def append_klc_list(self, klc_list):
 		self.klc_list.append(klc_list)
 	def get_decimal(self, value):

chanlun/core/ChanEnum.py

@@ -281,10 +281,6 @@ class Chan_BSP_TYPE(Enum):
     S1 = auto()
     S2 = auto()
     S3 = auto()
-    # 第四类：三类买卖点的低滞后变体，几何位置同 B3/S3，但不等笔确认。
-    # 见 chanlun/analysis/fast_bsp.py
-    B4 = auto()
-    S4 = auto()
     NONE = auto()
 """
 class Chan_BSP_TYPE(Enum):

chanlun/core/ChanFastBSP.py

@@ -1,40 +0,0 @@
-from chanlun.core.ChanEnum import Chan_BSP_DIR, Chan_BSP_TYPE
-
-
-class ChanFastBSP():
-	"""第四类买卖点（B4/S4）。
-
-	与 ChanBSP 的区别在于它不挂在笔上：fast_bsp 刻意不等笔确认，入场点是一根具体的
-	K线而非一笔的端点，所以时间与价格直接取自 K 线，没有 bi / klc 可依附。
-
-	htf_agree 与 ladder_ok 是两个独立的过滤标志，不在这里合成——上层（图表或策略）
-	自己决定要不要用、怎么组合。
-	"""
-
-	def __init__(self, time, price, ddir, entry_idx, bo_time=None, pb_time=None,
-	             lag=0, depth=0.0, zg=None, zd=None, occ=1,
-	             htf_dir=None, htf_agree=None, ladder_ok=None):
-		self.time = time
-		self.price = float(price)
-		self.dir = ddir
-		self.type = Chan_BSP_TYPE.B4 if ddir == Chan_BSP_DIR.BUY else Chan_BSP_TYPE.S4
-		self.entry_idx = int(entry_idx)
-		self.bo_time = bo_time
-		self.pb_time = pb_time
-		self.lag = int(lag)
-		self.depth = float(depth)
-		self.zg = float(zg) if zg is not None else None
-		self.zd = float(zd) if zd is not None else None
-		self.occ = int(occ)
-		self.htf_dir = htf_dir
-		self.htf_agree = htf_agree
-		self.ladder_ok = ladder_ok
-		# 入场即成立，没有「等待确认」这个状态；留此字段是为了与 ChanBSP 的序列化对齐
-		self.is_sure = True
-		self.start_time = time
-		self.end_time = time
-		self.sure_time = time
-
-	def __repr__(self):
-		name = str(self.type).replace('Chan_BSP_TYPE.', '')
-		return f"<ChanFastBSP {name} {self.time} {self.price} lag={self.lag}>"

chanlun/core/ChanKLC.py

@@ -63,11 +63,10 @@ class ChanKLC():
 		self.bsp = False
 		self.bsp_type = Chan_BSP_TYPE.NONE
 		# EMA状态字典：key为EMA名称，value为 {'pos': Chan_EMA_POS, 'semantic': Chan_EMA_SEMANTIC}
-		self._ema_status = {}
+		self.ema_status = {}
-		self._ema_status_dirty = False
 		# 向后兼容：保留 ema52_status 和 ema52_pos
-		self._ema52_status = 0
+		self.ema52_status = 0
-		self._ema52_pos = Chan_EMA_POS.UNKNOWN
+		self.ema52_pos = Chan_EMA_POS.UNKNOWN
 		self.bb2633upper = klu.bb2633upper
 		self.bb2633lower = klu.bb2633lower
 		self.bb2633middle = klu.bb2633middle
@@ -284,44 +283,21 @@ class ChanKLC():
 			'ema156': self.ema156,
 			'ema208': self.ema208,
 		}
-		self._ema_status_dirty = False
+		self.ema_status = {}
-		self._ema_status = {}
 		for name, value in ema_configs.items():
 			# 按 EMA 值的百分比自动计算阈值
 			threshold = abs(value) * self.threshold_pct if value and self.threshold_pct > 0 else 0
 			pos = ChanKLC.cal_ema_pos(self.high, self.low, self.close, value, threshold)
 			semantic = ChanKLC.cal_ema_semantic(pos, self.dir, self.ema_dir)
-			self._ema_status[name] = {
+			self.ema_status[name] = {
 				'pos': pos,
 				'semantic': semantic,
 				'value': value,
 				'threshold': threshold,
 			}
 		# 向后兼容
-		self._ema52_pos = self._ema_status['ema52']['pos']
+		self.ema52_pos = self.ema_status['ema52']['pos']
-		self._ema52_status = ChanKLC.semantic_to_int(self._ema_status['ema52']['semantic'])
+		self.ema52_status = ChanKLC.semantic_to_int(self.ema_status['ema52']['semantic'])
-
-	# 以下三个改成惰性求值。原来 set_end_klu 每次合并 KLU 都会立刻重算一遍，
-	# 实测占 TF_DF 构建的约 25%，而全仓（含前端）没有任何地方读取它的产出。
-	# 保留属性形式是为了任何外部读取仍拿到正确值，只是推迟到真被读时才算。
-	@property
-	def ema_status(self):
-		if self._ema_status_dirty:
-			self.cal_all_ema_status()
-		return self._ema_status
-
-	@property
-	def ema52_pos(self):
-		if self._ema_status_dirty:
-			self.cal_all_ema_status()
-		return self._ema52_pos
-
-	@property
-	def ema52_status(self):
-		if self._ema_status_dirty:
-			self.cal_all_ema_status()
-		return self._ema52_status
-
 	def get_ema_pos(self, ema_name):
 		"""获取指定EMA的客观位置，如 klc.get_ema_pos('ema24')"""
 		if ema_name in self.ema_status:
@@ -472,7 +448,7 @@ class ChanKLC():
 			klu.set_klc(self)
 		self.klc_dir = Chan_KLINE_DIR.UP if self.close > self.open else Chan_KLINE_DIR.DOWN
 		self.cal_indicators()
-		self._ema_status_dirty = True
+		self.cal_all_ema_status()
 		if self.open > self.high:
 			self.open = self.high
 		if self.close > self.high:

chanlun/core/ChanKLU.py

@@ -160,40 +160,27 @@ class ChanKLU:
 			return False
 		else:
 			return True
-	# 属性名 ← dataframe 列名。两条赋值路径（逐行 Series / 预取 ndarray）共用这张表，
-	# 免得将来加指标时只改一处、另一处静默漏掉。
-	INDICATOR_FIELDS = (
-		('macd', 'macd'), ('signal', 'macdsignal'), ('macdhist', 'macdhist'),
-		('ema26', 'ema26'), ('ema52', 'ema52'), ('ema24', 'ema24'),
-		('ema104', 'ema104'), ('ema156', 'ema156'), ('ema208', 'ema208'),
-		('ema13', 'ema13'), ('ema7', 'ema7'), ('ema5', 'ema5'),
-		('rsi', 'rsi'), ('volume_ratio', 'volume_ratio'),
-		('bb52upper', 'bb52upper'), ('bb52lower', 'bb52lower'),
-		('bb2633upper', 'bb2633upper'), ('bb2633lower', 'bb2633lower'),
-		('bb2633middle', 'bb2633middle'), ('ma5', 'ma5'),
-	)
-
 	def set_indicators(self, item):
-		"""单根赋值。增量追加时每次只有一根，走这条即可。
+		self.macd = float(item['macd']) if 'macd' in item and item['macd'] else 0
-
+		self.signal = float(item['macdsignal']) if 'macdsignal' in item and item['macdsignal'] else 0
-		原写法是 `float(item[c]) if c in item and item[c] else 0`。其中的真值判断
+		self.macdhist = float(item['macdhist']) if 'macdhist' in item and item['macdhist'] else 0
-		是空转：值为 0.0 时 float(0.0) 仍是 0，值为 NaN 时 NaN 为真值、照样透传。
+		self.ema26 = float(item['ema26']) if 'ema26' in item and item['ema26'] else 0
-		唯一起作用的是「列不存在则填 0」，所以这里只保留那一层。
+		self.ema52 = float(item['ema52']) if 'ema52' in item and item['ema52'] else 0
-		"""
+		self.ema24 = float(item['ema24']) if 'ema24' in item and item['ema24'] else 0
-		for attr, col in self.INDICATOR_FIELDS:
+		self.ema104 = float(item['ema104']) if 'ema104' in item and item['ema104'] else 0
-			v = item[col] if col in item else 0
+		self.ema156 = float(item['ema156']) if 'ema156' in item and item['ema156'] else 0
-			setattr(self, attr, float(v) if v else 0)
+		self.ema208 = float(item['ema208']) if 'ema208' in item and item['ema208'] else 0
-
+		self.ema13 = float(item['ema13']) if 'ema13' in item and item['ema13'] else 0
-	def set_indicators_from(self, cols, i):
+		self.ema7 = float(item['ema7']) if 'ema7' in item and item['ema7'] else 0
-		"""从预取的 {列名: ndarray} 按下标赋值，语义与 set_indicators 相同。
+		self.rsi = float(item['rsi']) if 'rsi' in item and item['rsi'] else 0
-
+		self.volume_ratio = float(item['volume_ratio']) if 'volume_ratio' in item and item['volume_ratio'] else 0
-		全量构建时用这条：避免每根 `df.iloc[i]` 构造一个 Series，再在其上做
+		self.bb52upper = float(item['bb52upper']) if 'bb52upper' in item and item['bb52upper'] else 0
-		几十次逐键查找——那是 TF_DF 构建 96% 的耗时所在。
+		self.bb52lower = float(item['bb52lower']) if 'bb52lower' in item and item['bb52lower'] else 0
-		"""
+		self.bb2633upper = float(item['bb2633upper']) if 'bb2633upper' in item and item['bb2633upper'] else 0
-		for attr, col in self.INDICATOR_FIELDS:
+		self.bb2633lower = float(item['bb2633lower']) if 'bb2633lower' in item and item['bb2633lower'] else 0
-			arr = cols.get(col)
+		self.bb2633middle = float(item['bb2633middle']) if 'bb2633middle' in item and item['bb2633middle'] else 0
-			v = arr[i] if arr is not None else 0
+		self.ma5 = float(item['ma5']) if 'ma5' in item and item['ma5'] else 0
-			setattr(self, attr, float(v) if v else 0)
+		self.ema5 = float(item['ema5']) if 'ema5' in item and item['ema5'] else 0
 	def cal_macd_state(self):
 		# 按定义精简实现：优先级 CROSS0 > 位置(HIGH/HE/RETURN_ZERO) > NEAR0 > UNKNOWN
 		# 首条或缺前一根

chanlun/indicators/ta.py

@@ -1,196 +0,0 @@
-"""Drop-in replacement for the `talib.abstract` calls this project makes.
-
-Same call signatures, same column names, same NaN warm-up lengths, so call
-sites only change their import line.
-
-Only what the codebase actually uses is implemented: SMA, MA, EMA, RSI, ATR,
-MACD, BBANDS. Numerical agreement with TA-Lib is enforced by
-`chanlun/tests/test_ta_compat.py`, which skips when talib is absent.
-
-The warm-up conventions below are TA-Lib's, not the textbook ones, and they
-differ between functions — getting them wrong shifts every downstream Chan
-structure by a bar:
-
-  SMA/BBANDS  first value at index period-1
-  EMA         seeded with the SMA of the first `period` values, at index period-1
-  RSI/ATR     Wilder smoothing (alpha = 1/period), first value at index period
-"""
-
-from __future__ import annotations
-
-import numpy as np
-import pandas as pd
-
-__all__ = ["SMA", "MA", "EMA", "RSI", "ATR", "MACD", "BBANDS"]
-
-
-def _series(data, price: str = "close") -> pd.Series:
-    """Accept the abstract-API shapes: DataFrame, Series, or ndarray."""
-    if isinstance(data, pd.DataFrame):
-        return data[price].astype(float)
-    if isinstance(data, pd.Series):
-        return data.astype(float)
-    return pd.Series(np.asarray(data, dtype=float))
-
-
-def _recursive(values: np.ndarray, seed: float, start: int, alpha: float, n: int) -> np.ndarray:
-    """out[start] = seed; out[i] = alpha*values[i] + (1-alpha)*out[i-1].
-
-    Delegates the recursion to pandas' C implementation rather than a Python
-    loop — `research/` runs this over long histories.
-    """
-    out = np.full(n, np.nan)
-    if start >= n:
-        return out
-    tail = values[start:].astype(float).copy()
-    tail[0] = seed
-    out[start:] = pd.Series(tail).ewm(alpha=alpha, adjust=False).mean().to_numpy()
-    return out
-
-
-def SMA(data, timeperiod: int = 30, price: str = "close") -> pd.Series:
-    s = _series(data, price)
-    return s.rolling(window=timeperiod, min_periods=timeperiod).mean()
-
-
-def MA(data, timeperiod: int = 30, matype: int = 0, price: str = "close") -> pd.Series:
-    if matype != 0:
-        raise NotImplementedError(f"MA matype={matype} is not used by this codebase")
-    return SMA(data, timeperiod, price=price)
-
-
-def _ema(x: np.ndarray, period: int, start: int) -> np.ndarray:
-    """EMA whose first output lands on `start`, seeded by the SMA of the
-    `period` values ending there.
-
-    `start` is a parameter because MACD needs the fast EMA to begin later than
-    it naturally would; see the note in MACD().
-    """
-    n = x.size
-    if n <= start or start < period - 1:
-        return np.full(n, np.nan)
-    seed = x[start - period + 1: start + 1].mean()
-    return _recursive(x, seed, start, 2.0 / (period + 1.0), n)
-
-
-def EMA(data, timeperiod: int = 30, price: str = "close") -> pd.Series:
-    s = _series(data, price)
-    x = s.to_numpy(dtype=float)
-    return pd.Series(_ema(x, timeperiod, timeperiod - 1), index=s.index)
-
-
-def RSI(data, timeperiod: int = 14, price: str = "close") -> pd.Series:
-    s = _series(data, price)
-    x = s.to_numpy(dtype=float)
-    n = x.size
-    out = np.full(n, np.nan)
-    if n <= timeperiod:
-        return pd.Series(out, index=s.index)
-
-    delta = np.diff(x)
-    gain = np.where(delta > 0.0, delta, 0.0)
-    loss = np.where(delta < 0.0, -delta, 0.0)
-
-    # delta[k] corresponds to bar k+1, so the first `timeperiod` deltas seed bar `timeperiod`.
-    alpha = 1.0 / timeperiod
-    avg_gain = _recursive(gain, gain[:timeperiod].mean(), timeperiod - 1, alpha, n - 1)
-    avg_loss = _recursive(loss, loss[:timeperiod].mean(), timeperiod - 1, alpha, n - 1)
-
-    ag = avg_gain[timeperiod - 1:]
-    al = avg_loss[timeperiod - 1:]
-    with np.errstate(divide="ignore", invalid="ignore"):
-        rsi = np.where(al == 0.0, 100.0, 100.0 - 100.0 / (1.0 + ag / al))
-    out[timeperiod:] = rsi
-    return pd.Series(out, index=s.index)
-
-
-def ATR(data, timeperiod: int = 14) -> pd.Series:
-    if not isinstance(data, pd.DataFrame):
-        raise TypeError("ATR needs a DataFrame with high/low/close")
-    high = data["high"].to_numpy(dtype=float)
-    low = data["low"].to_numpy(dtype=float)
-    close = data["close"].to_numpy(dtype=float)
-    n = high.size
-    out = np.full(n, np.nan)
-    if n <= timeperiod:
-        return pd.Series(out, index=data.index)
-
-    prev_close = close[:-1]
-    tr = np.maximum.reduce([
-        high[1:] - low[1:],
-        np.abs(high[1:] - prev_close),
-        np.abs(low[1:] - prev_close),
-    ])
-
-    # tr[k] is bar k+1; the first `timeperiod` true ranges seed bar `timeperiod`.
-    smoothed = _recursive(tr, tr[:timeperiod].mean(), timeperiod - 1, 1.0 / timeperiod, n - 1)
-    out[timeperiod:] = smoothed[timeperiod - 1:]
-    return pd.Series(out, index=data.index)
-
-
-def MACD(
-    data,
-    fastperiod: int = 12,
-    slowperiod: int = 26,
-    signalperiod: int = 9,
-    price: str = "close",
-) -> pd.DataFrame:
-    if slowperiod < fastperiod:
-        fastperiod, slowperiod = slowperiod, fastperiod
-
-    s = _series(data, price)
-    x = s.to_numpy(dtype=float)
-    n = x.size
-
-    macd = np.full(n, np.nan)
-    signal = np.full(n, np.nan)
-    hist = np.full(n, np.nan)
-    empty = pd.DataFrame({"macd": macd, "macdsignal": signal, "macdhist": hist}, index=s.index)
-
-    # Both EMAs emit their first value on the same bar. That makes the slow one
-    # ordinary, but re-seeds the fast one from the SMA of the `fastperiod`
-    # values ending there instead of carrying the recursion forward from bar
-    # fastperiod-1 — the two disagree by ~0.2 on a 100-priced series.
-    macd_start = slowperiod - 1
-    if n <= macd_start:
-        return empty
-    line = _ema(x, fastperiod, macd_start) - _ema(x, slowperiod, macd_start)
-
-    # The signal EMA runs over the MACD line, so everything shifts by another
-    # signalperiod-1 bars, and TA-Lib trims the MACD line to match.
-    valid = line[macd_start:]
-    if valid.size < signalperiod:
-        return empty
-    sig = _recursive(
-        valid, valid[:signalperiod].mean(), signalperiod - 1, 2.0 / (signalperiod + 1.0), valid.size
-    )
-    start = macd_start + signalperiod - 1
-    macd[start:] = line[start:]
-    signal[macd_start:] = sig
-    hist = macd - signal
-
-    return pd.DataFrame({"macd": macd, "macdsignal": signal, "macdhist": hist}, index=s.index)
-
-
-def BBANDS(
-    data,
-    timeperiod: int = 5,
-    nbdevup: float = 2.0,
-    nbdevdn: float = 2.0,
-    matype: int = 0,
-    price: str = "close",
-) -> pd.DataFrame:
-    if matype != 0:
-        raise NotImplementedError(f"BBANDS matype={matype} is not used by this codebase")
-    s = _series(data, price)
-    middle = s.rolling(window=timeperiod, min_periods=timeperiod).mean()
-    # TA-Lib uses the population standard deviation.
-    std = s.rolling(window=timeperiod, min_periods=timeperiod).std(ddof=0)
-    return pd.DataFrame(
-        {
-            "upperband": middle + nbdevup * std,
-            "middleband": middle,
-            "lowerband": middle - nbdevdn * std,
-        },
-        index=s.index,
-    )

chanlun/pipeline/builders/bi.py

@@ -6,7 +6,9 @@ from decimal import Decimal
 
 import numpy as np
 import pandas as pd
+import talib.abstract as ta
 from pandas import DataFrame
+from technical.util import resample_to_interval
 
 from chanlun.core.ChanBI import ChanBI
 from chanlun.core.ChanBIZS import ChanBIZS
@@ -465,7 +467,7 @@ class BiBuilderMixin:
 											pre_last_bi = bi_list[-2]
 											last_bi = bi_list[-1]
 											if pre_last_bi.is_sure and not last_bi.is_sure and pre_last_bi.dir == Chan_BI_DIR.UP and False:
-												#pre_last_bi.update_bi(klc)
+												pre_last_bi.update_bi(klc)
 												bi_list.remove(last_bi)
 												pre_last_bi.set_next(None)
 												#last_top.set_fx(Chan_FX_TYPE.PTOP)
@@ -583,7 +585,7 @@ class BiBuilderMixin:
 											pre_last_bi = bi_list[-2]
 											last_bi = bi_list[-1]
 											if pre_last_bi.is_sure and not last_bi.is_sure and pre_last_bi.dir == Chan_BI_DIR.DOWN and False: 
-												#pre_last_bi.update_bi(klc)
+												pre_last_bi.update_bi(klc)
 												bi_list.remove(last_bi)
 												pre_last_bi.set_next(None)
 												last_bottom = klc

chanlun/pipeline/builders/bsp.py

@@ -6,7 +6,9 @@ from decimal import Decimal
 
 import numpy as np
 import pandas as pd
+import talib.abstract as ta
 from pandas import DataFrame
+from technical.util import resample_to_interval
 
 from chanlun.core.ChanBI import ChanBI
 from chanlun.core.ChanBIZS import ChanBIZS

chanlun/pipeline/builders/fast_bsp.py

@@ -1,148 +0,0 @@
-"""第四类买卖点（B4/S4）接入 TF_DF。
-
-判定逻辑全在 chanlun/analysis/fast_bsp.py，这里只负责把引擎的中枢/K线喂进去，
-再把结果包成 ChanFastBSP。
-
-刻意不在 init_TF_DF 里默认计算：现有构造路径的开销保持不变，由调用方按需触发。
-"""
-from __future__ import annotations
-
-import re
-
-import pandas as pd
-
-from chanlun.analysis.fast_bsp import (
-	add_zone_ladder,
-	attach_htf_agree,
-	attach_zone_ladder,
-	ensure_timestamp,
-	find_fast_bsp3,
-	htf_fx_timeline,
-	zones_from_zs_list,
-)
-from chanlun.core.ChanEnum import Chan_BSP_DIR
-from chanlun.core.ChanFastBSP import ChanFastBSP
-
-# 区间套配对：小级别出中枢与买卖点，大级别只出分型定方向。
-# 取值来自 research/HANDOFF.md §1.5，是回测里实际用过的组合。
-FAST_BSP_HTF_PAIR = {
-	'1m': '5m',
-	'5m': '30m',
-	'15m': '1h',
-	'30m': '2h',
-}
-
-# 未列入配对表的周期回落到这个倍数
-FAST_BSP_HTF_FALLBACK_RATIO = 4
-
-_TF_UNIT_MINUTES = {'m': 1, 'h': 60, 'd': 1440, 'w': 10080}
-
-
-def timeframe_minutes(tf: str) -> int | None:
-	"""'30m' -> 30，'2h' -> 120。无法解析时返回 None。"""
-	if not tf:
-		return None
-	m = re.fullmatch(r'(\d+)\s*([mhdw])', str(tf).strip().lower())
-	if not m:
-		return None
-	return int(m.group(1)) * _TF_UNIT_MINUTES[m.group(2)]
-
-
-def resolve_htf(tf: str) -> tuple[str, int] | None:
-	"""给小级别找配套的大级别，返回 (标签, 分钟数)。"""
-	minutes = timeframe_minutes(tf)
-	if minutes is None:
-		return None
-	paired = FAST_BSP_HTF_PAIR.get(str(tf).strip().lower())
-	if paired:
-		return paired, timeframe_minutes(paired)
-	return f'{minutes * FAST_BSP_HTF_FALLBACK_RATIO}m', minutes * FAST_BSP_HTF_FALLBACK_RATIO
-
-
-class FastBspBuilderMixin:
-	def build_fast_bsp_htf(self, df, timeframe=None):
-		"""对同一份 df 重采样得到大级别，不额外拉数据。
-
-		大级别只用来取分型方向，样本太少就没有过滤意义，故重采样后不足 60 根时放弃。
-		"""
-		tf = timeframe or getattr(self, 'timeframe', None)
-		htf = resolve_htf(tf)
-		ltf_minutes = timeframe_minutes(tf)
-		if htf is None or not ltf_minutes:
-			return None
-		label, minutes = htf
-		if not minutes or len(df) * ltf_minutes < minutes * 60:
-			return None
-		try:
-			from chanlun.pipeline.timeframe import TF_DF
-
-			return TF_DF(df, minutes, label)
-		except Exception:
-			return None
-
-	def cal_fast_bsp(self, df=None, bi_zs_list=None, htf_chan=None, with_htf=True,
-	                 timeframe=None, **kw):
-		"""算第四类买卖点，返回 ChanFastBSP 列表。
-
-		bi_zs_list 传入已算好的 pure 笔中枢可免去重复计算。
-		with_htf=False 时跳过大级别构建，只留 ladder_ok 这一个过滤标志。
-		kw 透传给 find_fast_bsp3（scan / pullback_win / tol / require_touch 等）。
-		"""
-		src = df if df is not None else getattr(self, 'dataframe', None)
-		if src is None or len(src) == 0:
-			self.fast_bsp_list = []
-			return self.fast_bsp_list
-
-		src = ensure_timestamp(src)
-		if bi_zs_list is None:
-			bi_zs_list = getattr(self, 'bi_zs_list', None)
-			if not bi_zs_list:
-				bi_zs_list = self.cal_bi_zs_list_pure(self.cal_bi_list(self.get_klc_list(self.cal_kl_data(src))))
-
-		zones = zones_from_zs_list(bi_zs_list, src)
-		if zones.empty:
-			self.fast_bsp_list = []
-			return self.fast_bsp_list
-
-		zones = add_zone_ladder(zones)
-		sig = find_fast_bsp3(src, zones, **kw)
-		if sig.empty:
-			self.fast_bsp_list = []
-			return self.fast_bsp_list
-
-		sig = attach_zone_ladder(sig, zones)
-
-		if with_htf:
-			if htf_chan is None:
-				htf_chan = self.build_fast_bsp_htf(src, timeframe)
-			sig = attach_htf_agree(sig, src, htf_fx_timeline(htf_chan) if htf_chan is not None else pd.DataFrame())
-		else:
-			sig['htf_dir'] = None
-			sig['htf_agree'] = None
-
-		times = src['date'].dt.strftime('%Y-%m-%d %H:%M:%S').to_numpy()
-		close = src['close'].to_numpy(dtype=float)
-
-		out = []
-		for r in sig.itertuples(index=False):
-			entry_idx = int(r.entry_idx)
-			agree = getattr(r, 'htf_agree', None)
-			htf_dir = getattr(r, 'htf_dir', None)
-			out.append(ChanFastBSP(
-				time=times[entry_idx],
-				price=close[entry_idx],
-				ddir=Chan_BSP_DIR.BUY if r.direction == 1 else Chan_BSP_DIR.SELL,
-				entry_idx=entry_idx,
-				bo_time=times[int(r.bo_idx)],
-				pb_time=times[int(r.pb_idx)] if r.pb_idx == r.pb_idx else None,
-				lag=r.lag,
-				depth=r.depth,
-				zg=r.zg,
-				zd=r.zd,
-				occ=r.occ,
-				htf_dir=None if htf_dir is None or htf_dir != htf_dir else int(htf_dir),
-				htf_agree=None if agree is None or agree != agree else bool(agree),
-				ladder_ok=bool(r.ladder_ok),
-			))
-		self.fast_bsp_list = out
-		return out

chanlun/pipeline/builders/incremental.py

@@ -1,171 +0,0 @@
-"""增量更新：新K只追加 KLU/KLC，笔与笔中枢在当前列表上重算。
-
-不改 init_TF_DF 的整段语义。笔必须整表重扫：最后一笔 is_sure 允许收回
-（OWN_CHAN_ZS_001 上 60 天出现 7 次）。笔中枢用 cal_bi_zs_list_pure。
-"""
-from __future__ import annotations
-
-from datetime import datetime
-
-import pandas as pd
-from pandas import DataFrame
-from chanlun.pipeline.resample import resample_to_interval
-
-from chanlun.core.ChanEnum import Chan_FX_TYPE, Chan_KLC_FX, Chan_KLC_STATE
-from chanlun.core.ChanKLU import ChanKLU
-
-
-class IncrementalBuilderMixin:
-	def init_stream(self, df, interval=1, timeframe=None):
-		"""用历史K线初始化流式状态，之后用 append_bar / replace_last_bar。"""
-		if df is None or df.empty:
-			raise ValueError("DataFrame for stream is empty.")
-		if "date" not in df.columns:
-			raise ValueError(f"DataFrame missing 'date' column. Columns: {df.columns.tolist()}")
-		self.timeframe = timeframe
-		self.interval = interval
-		if interval == 1:
-			self.dataframe = df.copy()
-		else:
-			self.dataframe = resample_to_interval(df, interval)
-		self.dataframe = self.add_indicators(self.dataframe)
-		self.klu_list = []
-		self.klc_list = []
-		self.bi_list = []
-		self.bi_zs_list = []
-		self.seg_list = []
-		self.zs_list = []
-		self.bsp_list = []
-		self.klc_fx_list = []
-		self.big_zs_list = []
-		self._klc_feed_last_klu = None
-		for i in range(len(self.dataframe)):
-			self._append_row_at(i, rebuild=False)
-		self.rebuild_bi_zs()
-		return self
-
-	def append_bar(self, row):
-		"""追加一根已收盘K线。同一时间戳则改为替换最后一根。"""
-		self._ensure_stream_state()
-		item = self._normalize_row(row)
-		if self.klu_list and self.klu_list[-1].time == self._row_time_str(item):
-			return self.replace_last_bar(item)
-		self._append_item_to_dataframe(item)
-		self.dataframe = self.add_indicators(self.dataframe)
-		self._append_row_at(len(self.dataframe) - 1, rebuild=True)
-		return self
-
-	def replace_last_bar(self, row):
-		"""更新最后一根K（未完成K线走新OHLC）。包含关系从 KLU 列表重放。"""
-		self._ensure_stream_state()
-		if not self.klu_list:
-			return self.append_bar(row)
-		item = self._normalize_row(row)
-		idx = self.dataframe.index[-1]
-		for key, val in item.items():
-			self.dataframe.at[idx, key] = val
-		self.dataframe = self.add_indicators(self.dataframe)
-		self._apply_item_to_klu(self.klu_list[-1], self.dataframe.iloc[-1])
-		self._rebuild_klc_from_klu()
-		self.rebuild_bi_zs()
-		return self
-
-	def rebuild_bi_zs(self):
-		"""在当前 KLC 上重算笔 + cal_bi_zs_list_pure。会先清分型标记。"""
-		self._reset_klc_bi_marks(self.klc_list)
-		self.bi_list = self.cal_bi_list(self.klc_list) if self.klc_list else []
-		self.bi_zs_list = self.cal_bi_zs_list_pure(self.bi_list) if self.bi_list else []
-		return self.bi_zs_list
-
-	def _ensure_stream_state(self):
-		if not hasattr(self, "klu_list") or self.klu_list is None:
-			self.klu_list = []
-		if not hasattr(self, "klc_list") or self.klc_list is None:
-			self.klc_list = []
-		if not hasattr(self, "dataframe") or self.dataframe is None:
-			self.dataframe = DataFrame(
-				columns=["date", "open", "high", "low", "close", "volume"]
-			)
-		if not hasattr(self, "_klc_feed_last_klu"):
-			self._klc_feed_last_klu = self.klu_list[-1] if self.klu_list else None
-		if not hasattr(self, "bi_zs_list"):
-			self.bi_zs_list = []
-
-	def _rebuild_klc_from_klu(self):
-		self.klc_list = []
-		last_klu = None
-		for klu in self.klu_list:
-			self._push_klu_into_klc_list(self.klc_list, klu, last_klu)
-			last_klu = klu
-		self._klc_feed_last_klu = last_klu
-
-	def _append_row_at(self, idx, rebuild=True):
-		item = self.dataframe.iloc[idx]
-		klu = self._klu_from_item(item, idx)
-		if self.klu_list:
-			self.klu_list[-1].set_next(klu)
-			klu.set_pre(self.klu_list[-1])
-		self._push_klu_into_klc_list(self.klc_list, klu, self._klc_feed_last_klu)
-		self._klc_feed_last_klu = klu
-		self.klu_list.append(klu)
-		if rebuild:
-			self.rebuild_bi_zs()
-
-	def _klu_from_item(self, item, idx):
-		klu = ChanKLU(
-			self._item_time_str(item),
-			item["open"],
-			item["high"],
-			item["low"],
-			item["close"],
-			item["volume"],
-		)
-		klu.set_idx(idx)
-		if not hasattr(klu, "ema13"):
-			klu.ema13 = 0
-		if "macd" in item:
-			klu.set_indicators(item)
-		return klu
-
-	def _apply_item_to_klu(self, klu, item):
-		klu.time = self._item_time_str(item)
-		klu.open = item["open"]
-		klu.high = item["high"]
-		klu.low = item["low"]
-		klu.close = item["close"]
-		klu.volume = item["volume"]
-		klu.range = klu.high - klu.low
-		klu.body = abs(klu.close - klu.open)
-		if "macd" in item:
-			klu.set_indicators(item)
-
-	def _reset_klc_bi_marks(self, klc_list):
-		for klc in klc_list:
-			klc.fx = Chan_FX_TYPE.UNKNOWN
-			klc.klc_fx_type = Chan_KLC_FX.UNKNOWN
-			klc.klc_state = Chan_KLC_STATE.UNKNOWN
-			klc.bi = None
-			klc.fx_confirmed = False
-
-	def _item_time_str(self, item):
-		date = item["date"]
-		if hasattr(date, "to_pydatetime"):
-			date = date.to_pydatetime()
-		if isinstance(date, datetime):
-			return date.strftime("%Y-%m-%d %H:%M:%S")
-		return str(date)
-
-	def _row_time_str(self, item):
-		return self._item_time_str(item)
-
-	def _normalize_row(self, row):
-		if isinstance(row, pd.Series):
-			return row
-		return pd.Series(row)
-
-	def _append_item_to_dataframe(self, item):
-		row_df = DataFrame([item])
-		if self.dataframe is None or self.dataframe.empty:
-			self.dataframe = row_df
-		else:
-			self.dataframe = pd.concat([self.dataframe, row_df], ignore_index=True)

chanlun/pipeline/builders/indicators.py

@@ -6,8 +6,9 @@ from decimal import Decimal
 
 import numpy as np
 import pandas as pd
-from chanlun.indicators import ta
+import talib.abstract as ta
 from pandas import DataFrame
+from technical.util import resample_to_interval
 
 from chanlun.core.ChanBI import ChanBI
 from chanlun.core.ChanBIZS import ChanBIZS
@@ -54,14 +55,8 @@ class IndicatorsBuilderMixin:
 		return None
 
 	def add_indicators(self, df):
-		"""算指标并一次性挂到 df 上。
+		fast = 12
-
+		slow = 26
-		这里不逐列 `df['x'] = ...`：那样每一列都触发一次 BlockManager 插入，
-		30 多列的开销比全部 TA 计算本身还大（2001 行实测 TA 合计 2.5ms，
-		逐列赋值 3.6ms）。增量路径每根都要走一遍，这笔开销是白付的。
-		"""
-		fast = 26
-		slow = 52
 		period = 9
 		macd = ta.MACD(df, fastperiod=fast, slowperiod=slow, signalperiod=period)
 		bb365 = ta.BBANDS(df, timeperiod=365, nbdevup=3.0, nbdevdn=3.0, matype=0)
@@ -81,43 +76,40 @@ class IndicatorsBuilderMixin:
 		bbp30 = (df['close'] - bb30['lowerband']) / (bb30['upperband'] - bb30['lowerband'])
 		bbp302 = (df['close'] - bb302['lowerband']) / (bb302['upperband'] - bb302['lowerband'])
 		bbp2633 = (df['close'] - bb2633['lowerband']) / (bb2633['upperband'] - bb2633['lowerband'])
-		cols = {
+		df['bb2633upper'] = bb2633['upperband']
-			'bb2633upper': bb2633['upperband'],
+		df['bb2633lower'] = bb2633['lowerband']
-			'bb2633lower': bb2633['lowerband'],
+		df['bbp2633'] = bbp2633
-			'bbp2633': bbp2633,
+		df['bb2633middle'] = bb2633['middleband']
-			'bb2633middle': bb2633['middleband'],
+		df['atr'] = ta.ATR(df, timeperiod=14)
-			'atr': ta.ATR(df, timeperiod=14),
+		df['bbup365'] = bb365['upperband']
-			'bbup365': bb365['upperband'],
+		df['bblow365'] = bb365['lowerband']
-			'bblow365': bb365['lowerband'],
+		df['bbp365'] = bbp365
-			'bbp365': bbp365,
+		df['bbup120'] = bb120['upperband']
-			'bbup120': bb120['upperband'],
+		df['bblow120'] = bb120['lowerband']
-			'bblow120': bb120['lowerband'],
+		df['bbp120'] = bbp120
-			'bbp120': bbp120,
+		df['bbup30'] = bb30['upperband']
-			'bbup30': bb30['upperband'],
+		df['bblow30'] = bb30['lowerband']
-			'bblow30': bb30['lowerband'],
+		df['bbmiddle30'] = bb30_middle  # 添加bb30中轨
-			'bbmiddle30': bb30_middle,
+		df['bbp30'] = bbp30
-			'bbp30': bbp30,
+		df['bbup302'] = bb302['upperband']
-			'bbup302': bb302['upperband'],
+		df['bblow302'] = bb302['lowerband']
-			'bblow302': bb302['lowerband'],
+		df['bbp302'] = bbp302
-			'bbp302': bbp302,
+		df['macd'] = macd['macd']
-			'macd': macd['macd'],
+		df['macdsignal'] = macd['macdsignal']
-			'macdsignal': macd['macdsignal'],
+		df['macdhist'] = macd['macdhist']
-			'macdhist': macd['macdhist'],
+		df['ema5'] = ta.EMA(df, timeperiod=5)
-		}
+		df['ema10'] = ta.EMA(df, timeperiod=10)
-		for _p, _n in ((5, 'ema5'), (10, 'ema10'), (24, 'ema24'), (52, 'ema52'),
+		df['ema24'] = ta.EMA(df, timeperiod=24)
-		               (104, 'ema104'), (156, 'ema156'), (208, 'ema208'),
+		df['ema52'] = ta.EMA(df, timeperiod=52)
-		               (26, 'ema26'), (13, 'ema13'), (7, 'ema7')):
+		df['ema104'] = ta.EMA(df, timeperiod=104)
-			cols[_n] = ta.EMA(df, timeperiod=_p)
+		df['ema156'] = ta.EMA(df, timeperiod=156)
-		cols['rsi'] = ta.RSI(df, timeperiod=14)
+		df['ema208'] = ta.EMA(df, timeperiod=208)
-		cols['volume_ratio'] = self.cal_volume_ratio(df)
+		df['ema26'] = ta.EMA(df, timeperiod=26)
-
+		df['ema13'] = ta.EMA(df, timeperiod=13)
-		# 重复调用（增量路径每根都会）时先摘掉旧列，否则 concat 出重名列。
+		df['ema7'] = ta.EMA(df, timeperiod=7)
-		# 摘掉再接回末尾，列序与逐列覆盖的结果一致。
+		df['rsi'] = ta.RSI(df, timeperiod=14)
-		new = pd.DataFrame(cols, index=df.index)
+		df['volume_ratio'] = self.cal_volume_ratio(df)
-		dup = [c for c in new.columns if c in df.columns]
+		return df
-		if dup:
-			df = df.drop(columns=dup)
-		return pd.concat([df, new], axis=1)
 
 	def get_ema_state(self, dataframe):
 		klu_list = self.get_klu_list(dataframe)

chanlun/pipeline/builders/kline.py

@@ -6,7 +6,9 @@ from decimal import Decimal
 
 import numpy as np
 import pandas as pd
+import talib.abstract as ta
 from pandas import DataFrame
+from technical.util import resample_to_interval
 
 from chanlun.core.ChanBI import ChanBI
 from chanlun.core.ChanBIZS import ChanBIZS
@@ -84,8 +86,7 @@ class KlineBuilderMixin:
 		return Chan_FX_TYPE.UNKNOWN
 
 	def check_fx(self, klc):
-		# 右K未完成（仍在包含合并）时不分型：否则确认笔会随 next 扩区间被 check_*_fx 收回
+		if klc.pre and klc.next:
-		if klc.pre and klc.next and klc.next.end_klu is not None:
 			if klc.high > klc.pre.high and klc.high > klc.next.high and klc.low > klc.pre.low and klc.low > klc.next.low:
 				#if (klc.close > klc.ema52 or klc.next.close > klc.next.ema52) and klc.macd > 0:
 				klc.set_fx(Chan_FX_TYPE.TOP)
@@ -98,21 +99,6 @@ class KlineBuilderMixin:
 				return Chan_FX_TYPE.BOTTOM
 		return Chan_FX_TYPE.UNKNOWN
 
-	def check_fx3(self, klc):
-		# 右K未完成（仍在包含合并）时不分型：否则确认笔会随 next 扩区间被 check_*_fx 收回
-		if klc.pre and klc.next and klc.next.end_klu is not None:
-			next_klu = klc.next.end_klu.next
-			if klc.high > klc.pre.high and klc.high > klc.next.high and klc.low > klc.pre.low and klc.low > klc.next.low and klc.high > next_klu.high:
-				#if (klc.close > klc.ema52 or klc.next.close > klc.next.ema52) and klc.macd > 0:
-				klc.set_fx(Chan_FX_TYPE.TOP)
-				#print(klc.start_time, klc.end_time,klc.next.start_time, klc.next.end_time, klc.macd, klc.state, klc.fx, "TOP")
-				return Chan_FX_TYPE.TOP
-			elif klc.low < klc.pre.low and klc.low < klc.next.low and klc.high < klc.pre.high and klc.high < klc.next.high and klc.low < next_klu.low:
-				#if (klc.close < klc.ema52 or klc.next.close < klc.next.ema52) and klc.macd < 0:
-				klc.set_fx(Chan_FX_TYPE.BOTTOM)
-				#print(klc.start_time, klc.end_time,klc.next.start_time, klc.next.end_time, klc.macd, klc.state, klc.fx, "BOTTOM")
-				return Chan_FX_TYPE.BOTTOM
-		return Chan_FX_TYPE.UNKNOWN
 	def check_fx2(self, klc):
 		if klc.pre and klc.next:
 			if klc.high > klc.pre.close and klc.close > klc.next.close and klc.close > klc.pre.close and klc.close > klc.next.close:
@@ -128,138 +114,128 @@ class KlineBuilderMixin:
 		return Chan_FX_TYPE.UNKNOWN
 
 	def check_fx_pattern(self, klc):
-		"""给分型前后三根 KLC 的裸 K 打 pattern 标记。
-
-		原本还会把 `klu.to_string()` 拼成一个字符串——那是给下面那行注释掉的
-		print 用的，拼完就丢。它在 cal_bi_list 的内层，2000 根上要跑近三万次
-		f-string + 六万次 enum 格式化，是纯废动作，已删。
-
-		`klu.pattern` 只被 cal_klu_pattern 自己的双 K / 三 K 判定读，
-		不出这个模块，也不进 web 序列化。所以 lean 下整个调用可跳。
-		"""
-		if getattr(self, 'lean', False):
-			return
 		klu_list = klc.pre.klu_list + klc.klu_list + klc.next.klu_list
+		
 		self.cal_klu_pattern(klu_list)
+		p = ""
+		for klu in klu_list:
+			p += klu.to_string()
+		#print(p)
 
 	def cal_volume_ratio(self, dataframe, window=10):
-		"""当根量 / 前 window 根均量。前 window-1 根无基准，填 1.0。
+		df = dataframe.copy()
-
+		# 计算过去N根K线的平均成交量
-		原写法先 `dataframe.copy()` 再挂两列——为算一列 rolling 复制了整张
+		df['avg_volume'] = df['volume'].rolling(window=window).mean()
-		四十列的表。直接在 Series 上算，结果逐值相同。
+		# 计算量比
-		"""
+		df['volume_ratio'] = df['volume'] / df['avg_volume']
-		vol = dataframe['volume']
+		# 填充缺失值（前N根K线）
-		return (vol / vol.rolling(window=window).mean()).fillna(1.0).rename('volume_ratio')
+		df['volume_ratio'] = df['volume_ratio'].fillna(1.0)
+		return df['volume_ratio']
 
 	def cal_kl_data(self, dataframe:DataFrame):
-		"""按行构造 KLU 链。
+		fields = "time,open,high,low,close,volume"
-
-		这里刻意不用 `dataframe.iloc[i]`：那会为每一根新建一个几十列的 Series，
-		随后 set_indicators 再在其上做几十次逐键查找。实测这两件事合计占 TF_DF
-		构建耗时的 96%。改为先把用到的列取成 ndarray，循环里只做整数下标访问。
-		"""
-		n = len(dataframe)
-		if n == 0:
-			return []
-
-		times = self._format_times(dataframe['date'])
-		o_a = dataframe['open'].to_numpy(dtype=float)
-		h_a = dataframe['high'].to_numpy(dtype=float)
-		l_a = dataframe['low'].to_numpy(dtype=float)
-		c_a = dataframe['close'].to_numpy(dtype=float)
-		v_a = dataframe['volume'].to_numpy(dtype=float)
-
-		has_ind = 'macd' in dataframe.columns
-		ind_cols = {}
-		if has_ind:
-			for _attr, col in ChanKLU.INDICATOR_FIELDS:
-				if col in dataframe.columns:
-					ind_cols[col] = dataframe[col].to_numpy(dtype=float)
-
 		klu_list = []
 		last_klu = None
-		for i in range(n):
+		for i in range(0, len(dataframe)):
-			klu = ChanKLU(times[i], o_a[i], h_a[i], l_a[i], c_a[i], v_a[i])
+			item = dataframe.iloc[i]
+			date = item['date']
+			o = item['open']
+			h = item['high']
+			l = item['low']
+			c = item['close']
+			v = item['volume']
+			# time_obj = date.fromtimestamp(date)
+			# date = date + timedelta(hours=8)
+			time_str = date.strftime('%Y-%m-%d %H:%M:%S')
+			item_data = [
+				time_str,
+				o,
+				h,
+				l,
+				c,
+				v
+			]
+			# klu =  KLU(self.create_item_dict(item_data, GetColumnNameFromFieldList(fields)))
+			klu = ChanKLU(time_str, o, h, l, c, v)
+			# print(klu.time, klu.open, klu.high, klu.low, klu.close, klu.volume)	
 			klu.set_idx(i)
 			klu_list.append(klu)
 			if last_klu:
 				last_klu.set_next(klu)
 				klu.set_pre(last_klu)
 			last_klu = klu
-			if has_ind:
+			if 'macd' in item:
-				klu.set_indicators_from(ind_cols, i)
+				klu.set_indicators(item)
 		return klu_list
 
-	@staticmethod
-	def _format_times(col):
-		"""向量化 strftime。非 datetime 列（少见）退回逐个格式化。"""
-		fmt = '%Y-%m-%d %H:%M:%S'
-		try:
-			return col.dt.strftime(fmt).to_numpy()
-		except AttributeError:
-			return np.array([d.strftime(fmt) for d in col], dtype=object)
-
 	def get_kl_data(self, dataframe:DataFrame):
 		return self.cal_kl_data(dataframe)
 
-	def _push_klu_into_klc_list(self, klc_list, klu, last_klu):
+	def get_klc_list(self, klu_list):
-		"""把一根 KLU 并入包含K线列表。与 get_klc_list 的几何规则相同。"""
+		klc_list = []
-		if len(klc_list) > 0:
+		last_klu = None
-			last_klc = klc_list[-1]
+		# ChanMACD.__init__ 已调用 cal_macd_state，切勿再调一次（会重复堆积 seg/unittf）
-			if klu.exception:
+		macd = ChanMACD(klu_list)
-				ddir = Chan_KLINE_DIR.DOWN
+		klu_list = macd.klu_list
-				if last_klc.high < klu.high:
+		self._last_chan_macd = macd
-					ddir = Chan_KLINE_DIR.UP
+		ema_up_list = []
-				klc = ChanKLC(klu, index=len(klc_list), ddir=ddir)
+		ema_down_list = []
-				klc.high = klu.close if klu.close > klu.open else klu.open
+		ema_up_count = 0
-				klc.low = klu.open if klu.close > klu.open else klu.close
+		ema_down_count = 0
-				klc_list.append(klc)
+		last_klu = None
-				last_klc.set_next(klc)
+		for klu in klu_list:
-				klc.set_pre(last_klc)
+			ema = klu.ema52
-				last_klc.set_end_klu(last_klu)
+			last_ema = last_klu.ema52 if last_klu else 0
-				klc.set_pre_fx()
+			if klu.close >= ema:
-			else:
+				ema_up_count += 1
-				included = last_klc.check_klu_included(klu)
+			elif klu.close < ema:
-				if not included:
+				ema_down_count += 1
+			if last_klu and last_klu.close >= last_ema and klu.close < ema:
+				ema_up_list.append(ema_up_count)
+				#print(last_klu.time, ema_up_count, "UP END")
+				ema_up_count = 0
+			elif last_klu and last_klu.close < last_ema and klu.close >= ema:
+				ema_down_list.append(ema_down_count)
+				#print(last_klu.time, ema_down_count, "DOWN END")
+				ema_down_count = 0
+			if len(klc_list) > 0:
+				last_klc = klc_list[-1]
+				if klu.exception:
 					ddir = Chan_KLINE_DIR.DOWN
 					if last_klc.high < klu.high:
 						ddir = Chan_KLINE_DIR.UP
 					klc = ChanKLC(klu, index=len(klc_list), ddir=ddir)
+					klc.high = klu.close if klu.close > klu.open else klu.open
+					klc.low = klu.open if klu.close > klu.open else klu.close
 					klc_list.append(klc)
 					last_klc.set_next(klc)
 					klc.set_pre(last_klc)
 					last_klc.set_end_klu(last_klu)
 					klc.set_pre_fx()
+					#print(klu.time, klu.high, klu.low, klu.close, klu.open, klu.exception)
 				else:
-					last_klc.add_klu(klu)
+					included = last_klc.check_klu_included(klu)
-		else:
+					if not included:
-			ddir = Chan_KLINE_DIR.UP
+						ddir = Chan_KLINE_DIR.DOWN
-			if klu.open > klu.close:
+						if last_klc.high < klu.high:
-				ddir = Chan_KLINE_DIR.DOWN
+							ddir = Chan_KLINE_DIR.UP
-			klc = ChanKLC(klu, 0, ddir)
+						klc = ChanKLC(klu, index=len(klc_list), ddir=ddir)
-			klc_list.append(klc)
+						klc_list.append(klc)
-
+						last_klc.set_next(klc)
-	def get_klc_list(self, klu_list):
+						klc.set_pre(last_klc)
-		klc_list = []
+						last_klc.set_end_klu(last_klu)
-		# ChanMACD.__init__ 已调用 cal_macd_state，切勿再调一次（会重复堆积 seg/unittf）。
+						klc.set_pre_fx()
-		# lean 模式跳过整套 MACD 状态机：它只服务于 bsp/背驰/web 展示，笔与中枢不依赖它。
+					else:
-		if getattr(self, 'lean', False):
+						last_klc.add_klu(klu)
-			self._last_chan_macd = None
+			else:
-		else:
+				ddir = Chan_KLINE_DIR.UP
-			macd = ChanMACD(klu_list)
+				if klu.open > klu.close:
-			klu_list = macd.klu_list
+					ddir = Chan_KLINE_DIR.DOWN     
-			self._last_chan_macd = macd
+				klc = ChanKLC(klu, 0, ddir)
-
+				klc_list.append(klc)
-		last_klu = None
-		for klu in klu_list:
-			self._push_klu_into_klc_list(klc_list, klu, last_klu)
 			last_klu = klu
-		# cal_trend 只产出 klc.trend，而全仓只有它自己（经 prev_klcs 读自身序列
+		klc_list = self.cal_trend(klc_list)
-		# 状态）、一个 __str__ 和 web 的 klc_trend 图层消费它——笔与中枢不读。
+		#print(ema52_up_list, ema52_down_list)
-		# 增量路径的 klc 其 trend 恒为 UNKNOWN 却与批量构建逐字段相同，是实证。
-		# 所以 lean 下可跳；web 走非 lean，图层不受影响。
-		if not getattr(self, 'lean', False):
-			klc_list = self.cal_trend(klc_list)
 		return klc_list
 
 

chanlun/pipeline/builders/seg.py

@@ -6,7 +6,9 @@ from decimal import Decimal
 
 import numpy as np
 import pandas as pd
+import talib.abstract as ta
 from pandas import DataFrame
+from technical.util import resample_to_interval
 
 from chanlun.core.ChanBI import ChanBI
 from chanlun.core.ChanBIZS import ChanBIZS

chanlun/pipeline/builders/zs.py

@@ -6,7 +6,9 @@ from decimal import Decimal
 
 import numpy as np
 import pandas as pd
+import talib.abstract as ta
 from pandas import DataFrame
+from technical.util import resample_to_interval
 
 from chanlun.core.ChanBI import ChanBI
 from chanlun.core.ChanBIZS import ChanBIZS
@@ -353,12 +355,9 @@ class ZsBuilderMixin:
 			return bi_zs_list
 
 		def get_zs_range(bis):
-			bis_list = bis[0:3]
+			zg = min(bi.high for bi in bis)
-			zg = min(bi.high for bi in bis_list)
+			zd = max(bi.low for bi in bis)
-			zd = max(bi.low for bi in bis_list)
+			return zg, zd
-			dd = min(bi.low for bi in bis_list)
-			gg = max(bi.high for bi in bis_list)
-			return zg, zd, dd, gg
 
 		def is_bi_overlap_range(bi, zg, zd):
 			return bi.high >= zd and bi.low <= zg
@@ -376,8 +375,8 @@ class ZsBuilderMixin:
 			zs.bi_list = list(bis)
 			for bi in zs.bi_list:
 				bi.set_bi_zs(zs)
-			#zs.set_gg(max(bi.high for bi in zs.bi_list))
+			zs.set_gg(max(bi.high for bi in zs.bi_list))
-			#zs.set_dd(min(bi.low for bi in zs.bi_list))
+			zs.set_dd(min(bi.low for bi in zs.bi_list))
 			zs.classify_zs()
 
 		last_zs = None
@@ -395,7 +394,7 @@ class ZsBuilderMixin:
 				start_idx += 1
 				continue
 
-			zg, zd, dd, gg = get_zs_range([bi1, bi2, bi3])
+			zg, zd = get_zs_range([bi1, bi2, bi3])
 			if zg <= zd:
 				start_idx += 1
 				continue
@@ -421,8 +420,6 @@ class ZsBuilderMixin:
 			zs = ChanBIZS(bi1, len(bi_zs_list), zs_dir)
 			zs.set_zg(zg)
 			zs.set_zd(zd)
-			zs.set_dd(dd)
-			zs.set_gg(gg)
 
 			set_zs_bi_list(zs, bis_for_zs)
 			zs.set_end_bi(bis_for_zs[-1], bis_for_zs[-1].sure_time)