backtester.py

import numpy as np
import pandas as pd

class Backtester:
    def __init__(self, data, strategy, initial_cash=100000, transaction_cost=0.001):
        self.data = data.copy()
        self.strategy = strategy
        self.initial_cash = initial_cash
        self.transaction_cost = transaction_cost
        self.results = None

    def run(self):
        signals = self.strategy.generate_signals(self.data)
        self.data['Signal'] = signals
        self.data['Position'] = self.data['Signal'].shift().fillna(0)
        self.data['Returns'] = self.data['Close'].pct_change().fillna(0)
        self.data['Strategy_Returns'] = self.data['Returns'] * self.data['Position']
        # Transaction costs
        trades = self.data['Position'].diff().abs()
        self.data['Strategy_Returns'] -= trades * self.transaction_cost
        # Portfolio value
        self.data['Equity'] = self.initial_cash * (1 + self.data['Strategy_Returns']).cumprod()
        self.results = self.data
        return self.data

    def metrics(self, risk_free_rate=0.0):
        df = self.results if self.results is not None else self.run()
        total_return = df['Equity'].iloc[-1] / self.initial_cash - 1
        returns = df['Strategy_Returns']
        sharpe = (returns.mean() - risk_free_rate/252) / (returns.std() + 1e-8) * np.sqrt(252)
        equity = df['Equity']
        roll_max = equity.cummax()
        drawdown = (equity - roll_max) / roll_max
        max_drawdown = drawdown.min()
        trades = df['Position'].diff().abs()
        wins = ((returns > 0) & (trades > 0)).sum()
        total_trades = (trades > 0).sum()
        win_rate = wins / total_trades if total_trades > 0 else np.nan
        return {
            'Total Return': total_return,
            'Sharpe Ratio': sharpe,
            'Max Drawdown': max_drawdown,
            'Win Rate': win_rate
        }