持仓中放量不是出场信号，恰恰是最好那批单子的标记
用户问：既然 step50 证明入场根放量是接盘，那持仓中出现放量根是不是也说明 这一波走完了、该直接平掉？测下来方向是反的，且比入场那条还干净。 规则：持仓期间任一根 vr60 ≥ 阈值就收盘市价平掉（taker）。同根内优先级 止损（盘中）> 目标（盘中）> 放量平仓（收盘）。10 币 3941 笔： 基线（不看量） 毛R 1.068 R夏普 0.627 PF 3.80 余量 18.45bp 均持仓 30.0 vr60≥3 就平 毛R 0.617 R夏普 0.459 PF 2.81 余量 7.81bp 均持仓 11.1 vr60≥5 就平 毛R 0.858 R夏普 0.568 PF 3.37 余量 12.85bp 均持仓 19.8 vr60≥8 就平 毛R 1.009 R夏普 0.617 PF 3.69 余量 16.68bp 均持仓 26.3 阈值越高、触发越少就越接近基线——这条曲线的最优点是「永不触发」，规则纯扣分。 「浮盈才平」的变体把胜率抬到 74.4%（基线 71.7%）而毛R 掉到 0.655，是过早 止盈的教科书特征：胜率上升、期望下降。 根因：vr60≥5 触发的 1769 笔若不平，止盈率 45.7%、止损率 16.6%，而全体基线 是 30.1% / 33.1%。持仓中的放量根标记的是最好的那批单子，平掉每笔让出 +0.484R。 幸存者偏差已控——按基线持仓 ≥K 根分层后 5/5 档同向，放量组止盈率约为无量组 两倍（K=5 时 41.2% vs 19.8%）。 同一个事件入场为负、持仓为正，区别只在站在它的哪一边：入场那根放量你是买方， 持仓中那根是资金来接你的货。用户「有资金的趋势才是好趋势」的直觉成立， 作用点在持仓期而非入场点。 自带逐根模拟器不走 walk_exits，所以先与它对拍基线（毛收益差 <1e-12、出场 原因零分歧），10/10 币通过才往下算。 Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
jackyu66git
@@ -0,0 +1,262 @@
|
||||
"""Step 52：持仓中放量就平仓 —— 把成交量当出场信号。
|
||||
|
||||
用户提出：既然 step50 证明「入场根放量 = 给走完这一冲的人接盘」，那持仓过程中
|
||||
出现放量根，是不是也说明这一波被走完了，该直接平掉？
|
||||
|
||||
这和 step50 是同一个机制的延伸，但**方向未知**：放量既可能是衰竭（该走），
|
||||
也可能是突破续势的起点（走了就砍在起涨点）。step50 只证明了「入场时撞上放量
|
||||
不好」，推不出「持仓时撞上放量该跑」——入场是你在接别人的盘，持仓时那根量
|
||||
可能正是把你送上去的那批资金。
|
||||
|
||||
出场腿是 taker（收盘市价），成本按 TIME 计。触发优先级：同一根内止损（盘中）
|
||||
> 目标（盘中）> 放量平仓（收盘）。止损优先是保守侧。
|
||||
|
||||
⚠️ 本步自己写了逐根模拟器，不走 walk_exits。**基线必须与 walk_exits 逐笔
|
||||
相等才继续**——否则后面所有对比都是在和一个错的基线比。
|
||||
"""
|
||||
from __future__ import annotations
|
||||
|
||||
import argparse
|
||||
import os
|
||||
import sys
|
||||
import warnings
|
||||
from concurrent.futures import ProcessPoolExecutor, as_completed
|
||||
from pathlib import Path
|
||||
|
||||
import numpy as np
|
||||
import pandas as pd
|
||||
|
||||
warnings.filterwarnings("ignore")
|
||||
for v in ("OMP_NUM_THREADS", "OPENBLAS_NUM_THREADS", "MKL_NUM_THREADS"):
|
||||
os.environ.setdefault(v, "1")
|
||||
|
||||
HERE = Path(__file__).resolve().parent
|
||||
sys.path.insert(0, str(HERE))
|
||||
sys.path.insert(0, str(HERE.parent))
|
||||
pd.set_option("display.width", 340)
|
||||
|
||||
SL, SCALE_AT, RUNNER, MAXB = 2.0, 3.0, 8.0, 48
|
||||
GATE_BP = 8.0
|
||||
THRESHOLDS = (3.0, 5.0, 8.0)
|
||||
OUT = HERE / "out" / "step52_volume_exit.feather"
|
||||
IS_START = pd.Timestamp("2026-01-30", tz="Asia/Shanghai")
|
||||
|
||||
TP, SL_, TIME = 0, 1, 2
|
||||
|
||||
|
||||
def simulate(cdf, sig, vr):
|
||||
"""逐根前推。返回每笔在基线与各放量出场变体下的 (毛收益, 原因, 是否分批)。
|
||||
|
||||
runner 止损位与初始止损同为 2 ATR（HANDOFF §3.5 定的 rstop=SL），
|
||||
所以全程止损线不动，不需要分段处理。
|
||||
"""
|
||||
high = cdf["high"].to_numpy(float)
|
||||
low = cdf["low"].to_numpy(float)
|
||||
open_ = cdf["open"].to_numpy(float)
|
||||
close = cdf["close"].to_numpy(float)
|
||||
atr = cdf["atr"].to_numpy(float)
|
||||
n = len(cdf)
|
||||
|
||||
variants = ["base"]
|
||||
for t in THRESHOLDS:
|
||||
variants += [f"v{t:g}", f"v{t:g}w"]
|
||||
|
||||
rows = []
|
||||
for s, d in zip(sig["entry_idx"].astype(int), sig["direction"].astype(int)):
|
||||
e = s + 1
|
||||
if e >= n - 1:
|
||||
continue
|
||||
a = atr[s]
|
||||
if not np.isfinite(a) or a <= 0:
|
||||
continue
|
||||
entry = open_[e]
|
||||
end = min(e + MAXB, n - 1)
|
||||
row = {"sig_idx": s, "atr_pct": a / entry}
|
||||
|
||||
for var in variants:
|
||||
thr = None if var == "base" else float(var[1:].rstrip("w"))
|
||||
only_win = var.endswith("w")
|
||||
scaled = False
|
||||
g = r = None
|
||||
xb = end
|
||||
for j in range(e, end + 1):
|
||||
adv = (high[j] - entry) / a if d == 1 else (entry - low[j]) / a
|
||||
ret = (entry - low[j]) / a if d == 1 else (high[j] - entry) / a
|
||||
# ① 止损（盘中）。同根内优先于目标，保守侧
|
||||
if ret >= SL:
|
||||
hit = -SL * a / entry
|
||||
g = 0.5 * (SCALE_AT * a / entry) + 0.5 * hit if scaled else hit
|
||||
r, xb = SL_, j
|
||||
break
|
||||
# ② 目标（盘中限价）
|
||||
if not scaled and adv >= SCALE_AT:
|
||||
# 同根内可能既到 3 ATR 又到 8 ATR，按先减仓后续跑处理
|
||||
scaled = True
|
||||
if adv >= RUNNER:
|
||||
g = 0.5 * (SCALE_AT * a / entry) + 0.5 * (RUNNER * a / entry)
|
||||
r, xb = TP, j
|
||||
break
|
||||
elif scaled and adv >= RUNNER:
|
||||
g = 0.5 * (SCALE_AT * a / entry) + 0.5 * (RUNNER * a / entry)
|
||||
r, xb = TP, j
|
||||
break
|
||||
# ③ 放量平仓（收盘市价）
|
||||
if thr is not None and np.isfinite(vr[j]) and vr[j] >= thr:
|
||||
px = d * (close[j] - entry) / entry
|
||||
if not only_win or px > 0:
|
||||
g = 0.5 * (SCALE_AT * a / entry) + 0.5 * px if scaled else px
|
||||
r, xb = TIME, j
|
||||
break
|
||||
if g is None: # 超时：末根收盘市价
|
||||
px = d * (close[end] - entry) / entry
|
||||
g = 0.5 * (SCALE_AT * a / entry) + 0.5 * px if scaled else px
|
||||
r, xb = TIME, end
|
||||
row[f"{var}_g"], row[f"{var}_r"] = g, r
|
||||
row[f"{var}_c"] = int(scaled)
|
||||
row[f"{var}_b"] = xb - e + 1
|
||||
rows.append(row)
|
||||
return pd.DataFrame(rows)
|
||||
|
||||
|
||||
def collect(sym: str, rows: int):
|
||||
import warnings as _w
|
||||
_w.filterwarnings("ignore")
|
||||
sys.path.insert(0, str(HERE))
|
||||
sys.path.insert(0, str(HERE.parent))
|
||||
|
||||
from chanlun import TF_DF
|
||||
from chanlun.analysis.fast_bsp import (
|
||||
add_zone_ladder, attach_htf_agree, attach_zone_ladder,
|
||||
build_htf_zones, find_fast_bsp3, htf_fx_timeline,
|
||||
)
|
||||
from lib.data import fetch_ohlcv
|
||||
from lib.exit_model import cfg_name, walk_exits
|
||||
|
||||
try:
|
||||
df = fetch_ohlcv(f"{sym}/USDT:USDT", "1m", rows)
|
||||
if df is None or len(df) < 50_000:
|
||||
return None
|
||||
chan = TF_DF(df, 1, "1m", lean=True)
|
||||
cdf = chan.dataframe
|
||||
zones = build_htf_zones(cdf, "1m", chan=chan)
|
||||
if zones.empty:
|
||||
return None
|
||||
zl = add_zone_ladder(zones.reset_index(drop=True))
|
||||
sig = find_fast_bsp3(cdf, zl)
|
||||
if sig.empty:
|
||||
return None
|
||||
dh = fetch_ohlcv(f"{sym}/USDT:USDT", "5m", 10 ** 9)
|
||||
ch = TF_DF(dh, 1, "5m", lean=True)
|
||||
sig = attach_zone_ladder(
|
||||
attach_htf_agree(sig, cdf, htf_fx_timeline(ch, ch.dataframe)), zl)
|
||||
|
||||
vr = (cdf["volume"] / cdf["volume"].rolling(60, min_periods=10).mean()
|
||||
).to_numpy(float)
|
||||
res = simulate(cdf, sig, vr)
|
||||
|
||||
# 对拍：基线必须与已验证的 walk_exits 逐笔相等
|
||||
ref = walk_exits(cdf, sig, [SL], [RUNNER], [MAXB], scale_at=SCALE_AT,
|
||||
runners=(RUNNER,), runner_stops=(SL,))
|
||||
cfg = cfg_name(SL, RUNNER, MAXB, SL)
|
||||
if len(ref) != len(res):
|
||||
print(f" {sym} 对拍失败：笔数 {len(ref)} vs {len(res)}", flush=True)
|
||||
return None
|
||||
dg = np.abs(ref[f"{cfg}_g"].to_numpy() - res["base_g"].to_numpy())
|
||||
dr = (ref[f"{cfg}_r"].to_numpy() != res["base_r"].to_numpy()).sum()
|
||||
if dg.max() > 1e-12 or dr:
|
||||
print(f" {sym} 对拍失败：毛收益最大差 {dg.max():.3e}，原因分歧 {dr} 笔",
|
||||
flush=True)
|
||||
return None
|
||||
|
||||
idx = sig["entry_idx"].to_numpy().astype(int)
|
||||
keep = np.isin(idx, res["sig_idx"].to_numpy())
|
||||
idx = idx[keep]
|
||||
atr = cdf["atr"].to_numpy(float)[idx]
|
||||
close = cdf["close"].to_numpy(float)[idx]
|
||||
out = pd.DataFrame({
|
||||
"sym": sym,
|
||||
"date": cdf["date"].to_numpy()[idx],
|
||||
"atr_bp": atr / close * 1e4,
|
||||
"htf": sig["htf_agree"].to_numpy()[keep],
|
||||
"lad": sig["ladder_ok"].to_numpy()[keep],
|
||||
"vr60_entry": vr[idx],
|
||||
})
|
||||
for c in res.columns:
|
||||
if c not in ("sig_idx",):
|
||||
out[c] = res[c].to_numpy()
|
||||
return out
|
||||
except Exception as e:
|
||||
print(f" {sym} 失败: {e!r}", flush=True)
|
||||
return None
|
||||
|
||||
|
||||
def stat(d: pd.DataFrame, var: str, label: str) -> dict:
|
||||
from lib.exit_model import fee_of, taker_notional
|
||||
|
||||
g = d[f"{var}_g"].to_numpy()
|
||||
r, c = d[f"{var}_r"].to_numpy(), d[f"{var}_c"].to_numpy()
|
||||
tn = taker_notional(r, c)
|
||||
net = g - fee_of(r, c)
|
||||
risk = SL * d.atr_pct.to_numpy()
|
||||
R = net / risk
|
||||
w, o = net[net > 0].sum(), -net[net <= 0].sum()
|
||||
return {"方案": label, "笔数": len(d),
|
||||
"胜率": f"{(net > 0).mean()*100:.1f}%",
|
||||
"毛R": round((g / risk).mean(), 3), "净均R": round(R.mean(), 3),
|
||||
"R夏普": round(R.mean() / R.std(ddof=1), 3),
|
||||
"PF": round(w / o, 2) if o > 0 else np.inf,
|
||||
"余量bp": round(net.mean() / tn.mean() * 1e4, 2),
|
||||
"均持仓": round(d[f"{var}_b"].mean(), 1),
|
||||
"触发率": f"{(r == TIME).mean()*100:.0f}%"}
|
||||
|
||||
|
||||
def report(d: pd.DataFrame, label: str) -> None:
|
||||
rows = [stat(d, "base", "基线（不看量）")]
|
||||
for t in THRESHOLDS:
|
||||
rows.append(stat(d, f"v{t:g}", f"vr60≥{t:g} 就平"))
|
||||
rows.append(stat(d, f"v{t:g}w", f"vr60≥{t:g} 且浮盈才平"))
|
||||
print(f"\n--- {label}（{len(d)} 笔）---")
|
||||
print(pd.DataFrame(rows).to_string(index=False))
|
||||
|
||||
|
||||
def main() -> None:
|
||||
ap = argparse.ArgumentParser()
|
||||
ap.add_argument("--symbols", default="BTC,BNB,ETH,SOL,LINK,LTC,AVAX,XRP,DOGE,ADA")
|
||||
ap.add_argument("--rows", type=int, default=800_000)
|
||||
ap.add_argument("--workers", type=int, default=3)
|
||||
ap.add_argument("--reuse", action="store_true")
|
||||
args = ap.parse_args()
|
||||
|
||||
if args.reuse and OUT.exists():
|
||||
d = pd.read_feather(OUT)
|
||||
else:
|
||||
syms = [s.strip() for s in args.symbols.split(",")]
|
||||
print(f"[放量出场] {len(syms)} 币 × {args.rows} 根 1m\n", flush=True)
|
||||
parts = []
|
||||
with ProcessPoolExecutor(max_workers=args.workers) as ex:
|
||||
fut = {ex.submit(collect, s, args.rows): s for s in syms}
|
||||
for i, f in enumerate(as_completed(fut), 1):
|
||||
r = f.result()
|
||||
print(f" [{i}/{len(syms)}] {fut[f]} {0 if r is None else len(r)}"
|
||||
f"{' ⚠对拍未过' if r is None else ' 对拍通过'}", flush=True)
|
||||
if r is not None:
|
||||
parts.append(r)
|
||||
if not parts:
|
||||
print("无结果")
|
||||
return
|
||||
d = pd.concat(parts, ignore_index=True)
|
||||
d.to_feather(OUT)
|
||||
|
||||
d["date"] = pd.to_datetime(d["date"])
|
||||
d = d[(d.htf == 1.0) & d.lad & (d.atr_bp >= GATE_BP)].copy()
|
||||
print(f"\n实盘口径 {len(d)} 笔 {d.date.min():%Y-%m-%d} ~ {d.date.max():%Y-%m-%d}")
|
||||
|
||||
print("\n" + "=" * 118)
|
||||
print("########## 放量出场 vs 基线 ##########")
|
||||
report(d[d.date < IS_START], "样本外")
|
||||
report(d[d.date >= IS_START], "发现期")
|
||||
report(d, "全样本")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
Reference in New Issue
Block a user