Alice — an open-source AI trading bot covering equities, crypto, commodities, forex, and macro
| 1 | """Backtesting engine.""" |
| 2 | |
| 3 | from __future__ import annotations |
| 4 | |
| 5 | from dataclasses import dataclass |
| 6 | |
| 7 | import pandas as pd |
| 8 | |
| 9 | |
| 10 | @dataclass |
| 11 | class BacktestResult: |
| 12 | """Results from a backtest run.""" |
| 13 | |
| 14 | equity_curve: pd.Series |
| 15 | returns: pd.Series |
| 16 | total_return: float |
| 17 | sharpe_ratio: float |
| 18 | max_drawdown: float |
| 19 | num_trades: int |
| 20 | win_rate: float |
| 21 | |
| 22 | |
| 23 | def run_backtest( |
| 24 | df: pd.DataFrame, |
| 25 | initial_capital: float = 100_000.0, |
| 26 | commission: float = 0.001, |
| 27 | ) -> BacktestResult: |
| 28 | """Run a simple backtest on a DataFrame with Signal and Position columns. |
| 29 | |
| 30 | Args: |
| 31 | df: DataFrame with Close, Signal, and Position columns. |
| 32 | initial_capital: Starting capital. |
| 33 | commission: Per-trade commission as a fraction (0.001 = 0.1%). |
| 34 | |
| 35 | Returns: |
| 36 | BacktestResult with performance metrics. |
| 37 | """ |
| 38 | out = df.copy() |
| 39 | |
| 40 | # Strategy returns: position * daily return |
| 41 | out["Strategy_Return"] = out["Position"].shift(1).fillna(0) * out["Return"] |
| 42 | |
| 43 | # Apply commission on position changes |
| 44 | out["Position_Change"] = out["Position"].diff().abs().fillna(0) |
| 45 | out["Strategy_Return"] = out["Strategy_Return"] - out["Position_Change"] * commission |
| 46 | |
| 47 | # Equity curve |
| 48 | out["Equity"] = initial_capital * (1 + out["Strategy_Return"].fillna(0)).cumprod() |
| 49 | |
| 50 | equity = out["Equity"] |
| 51 | returns = out["Strategy_Return"].fillna(0) |
| 52 | |
| 53 | # Metrics |
| 54 | total_return = float(equity.iloc[-1] / initial_capital - 1) |
| 55 | |
| 56 | # Sharpe ratio (annualized, assuming 252 trading days) |
| 57 | mean_return = returns.mean() |
| 58 | std_return = returns.std() |
| 59 | sharpe = float((mean_return / std_return) * (252**0.5)) if std_return > 0 else 0.0 |
| 60 | |
| 61 | # Max drawdown |
| 62 | rolling_max = equity.cummax() |
| 63 | drawdown = (equity - rolling_max) / rolling_max |
| 64 | max_drawdown = float(drawdown.min()) |
| 65 | |
| 66 | # Trade count and win rate |
| 67 | num_trades = int((out["Position_Change"] > 0).sum()) |
| 68 | winning_trades = int((returns > 0).sum()) |
| 69 | total_trades_with_return = int((returns != 0).sum()) |
| 70 | win_rate = winning_trades / total_trades_with_return if total_trades_with_return > 0 else 0.0 |
| 71 | |
| 72 | return BacktestResult( |
| 73 | equity_curve=equity, |
| 74 | returns=returns, |
| 75 | total_return=total_return, |
| 76 | sharpe_ratio=sharpe, |
| 77 | max_drawdown=max_drawdown, |
| 78 | num_trades=num_trades, |
| 79 | win_rate=win_rate, |
| 80 | ) |