Add initial implementation of backtesting framework with CLI interface. Introduce core modules for data loading, trade management, performance metrics, and logging. Include Supertrend indicator calculations and slippage estimation. Update .gitignore to exclude logs and CSV files.

This commit is contained in:
2026-01-09 19:53:01 +08:00
parent a25499e016
commit c4aa965a98
15 changed files with 424 additions and 568 deletions

54
metrics.py Normal file
View File

@@ -0,0 +1,54 @@
from __future__ import annotations
from dataclasses import dataclass
import numpy as np
import pandas as pd
@dataclass
class Perf:
total_return: float
max_drawdown: float
sharpe_ratio: float
win_rate: float
num_trades: int
final_equity: float
initial_equity: float
num_stop_losses: int
total_fees: float
total_slippage_usd: float
avg_slippage_bps: float
def compute_metrics(equity_curve: pd.Series, trades: list[dict]) -> Perf:
ret = equity_curve.pct_change().fillna(0.0)
total_return = equity_curve.iat[-1] / equity_curve.iat[0] - 1.0
cummax = equity_curve.cummax()
dd = (equity_curve / cummax - 1.0).min()
max_drawdown = dd
if ret.std(ddof=0) > 0:
sharpe = (ret.mean() / ret.std(ddof=0)) * np.sqrt(252 * 24 * 60) # minute bars -> annualized
else:
sharpe = 0.0
closes = [t for t in trades if t.get("side") == "SELL"]
wins = [t for t in closes if t.get("pnl", 0.0) > 0]
win_rate = (len(wins) / len(closes)) if closes else 0.0
fees = sum(t.get("fee", 0.0) for t in trades)
slip = sum(t.get("slippage", 0.0) for t in trades)
slippage_bps = [t.get("slippage_bps", 0.0) for t in trades if "slippage_bps" in t]
return Perf(
total_return=total_return,
max_drawdown=max_drawdown,
sharpe_ratio=sharpe,
win_rate=win_rate,
num_trades=len(closes),
final_equity=float(equity_curve.iat[-1]),
initial_equity=float(equity_curve.iat[0]),
num_stop_losses=sum(1 for t in closes if t.get("reason") == "stop"),
total_fees=fees,
total_slippage_usd=slip,
avg_slippage_bps=float(np.mean(slippage_bps)) if slippage_bps else 0.0,
)