Сравнение торговых стратегий через бэктест

★ 7.1 · general

strategy-compare is a Claude Code skill that compares multiple trading strategies — or directions (long / short / both) — on the same symbol and produces a side-by-side metrics table. It generates a Python script inside `backtesting/strategy_comparison/`, fetching data via OpenAlgo or directly from a DuckDB file, computing all indicators with `openalgo.ta` (EMA-crossover, RSI, Donchian, Supertrend, etc.), cleaning signals with `ta.exrem()`, and running backtests in vectorbt with Indian delivery fees (`fees=0.00111, fixed_fees=20`). The comparison table includes a NIFTY benchmark alongside each strategy's Total Return, Sharpe, Sortino, Max Drawdown, Win Rate, and Profit Factor, followed by a plain-language summary of which strategy performed best and why. Overlaid equity curves are rendered in Plotly dark theme and the full comparison is saved to CSV — ideal for systematically selecting strategies on Indian equities.