"""Step 1：全量口径事件研究，给出缠论买卖点信号有效性的乐观上界。

注意这一步是 in-sample 的：买卖点由「看完全部历史」的一次性 pipeline 产出，
中枢与笔的最终形态可能包含事后信息。结论需由 Step 2 的 walk-forward 重放校验。
"""
from __future__ import annotations

import sys
from pathlib import Path

import pandas as pd

sys.path.insert(0, str(Path(__file__).resolve().parent))

from lib.bsp_eval import DEFAULT_HORIZONS, forward_returns, run_pipeline, summarize, to_events
from lib.data import fetch_ohlcv

pd.set_option("display.width", 200)
pd.set_option("display.max_columns", 50)


def main() -> None:
    symbol, tf, limit = "BTC/USDT:USDT", "1h", 50000

    df = fetch_ohlcv(symbol, tf, limit)
    print(f"[data] {symbol} {tf}  rows={len(df)}  {df['date'].iloc[0]} -> {df['date'].iloc[-1]}")

    chan, bsp_list = run_pipeline(df, tf)
    events = to_events(bsp_list, df)
    print(f"[chan] bi={len(chan.bi_list)} seg={len(chan.seg_list)} bsp={len(bsp_list)} events={len(events)}")

    if not events:
        print("没有产出可交易事件，终止。")
        return

    fwd = forward_returns(events, df)

    lag = fwd["lag_bars"]
    print(f"\n[确认滞后] 均值={lag.mean():.1f} 根  中位数={lag.median():.0f} 根  最大={lag.max()} 根")
    print("  各类型滞后中位数:")
    print(fwd.groupby("bsp_type")["lag_bars"].agg(["count", "median", "mean", "max"]).to_string())

    summary = summarize(fwd, df)
    print("\n[事件研究] 方向调整后收益（正=盈利）")
    out = summary.copy()
    for col in ("mean", "median", "excess", "mfe", "mae"):
        out[col] = out[col].map(lambda v: f"{v * 100:+.2f}%")
    out["winrate"] = out["winrate"].map(lambda v: f"{v * 100:.0f}%")
    out["tstat"] = out["tstat"].map(lambda v: f"{v:+.2f}")
    print(out.to_string(index=False))

    Path(__file__).parent.joinpath("out").mkdir(exist_ok=True)
    fwd.to_csv(Path(__file__).parent / "out" / "step1_events.csv", index=False)
    summary.to_csv(Path(__file__).parent / "out" / "step1_summary.csv", index=False)
    print("\n明细已写入 research/out/")


if __name__ == "__main__":
    main()