Backtest a strategy, after real costs
Rank a signal panel into a long/short book, hold it for a horizon, and pay each market's actual transaction tax. You get a costed equity curve, Sharpe / IR / turnover, a walk-forward out-of-sample split, and a Deflated Sharpe that tells you whether the best parameters survive the multiple-testing you found them through. Everything runs in your browser.
Columns: date, ticker, market, score, forward_return (header required). One row per name per date; score is your signal, forward_return the realised return as a fraction (0.01 = +1%). Computed in your browser; nothing is uploaded.
Strategy
Factual statistical analysis of historical, resolved data — not investment advice and not a forward recommendation. Backtested results are hypothetical, ignore slippage beyond the modelled spread, and do not predict future returns. Your own pasted data never leaves the browser. About the dataset.