Every strategy runs through the same validation pipeline before it's considered ready for a live account — built to catch curve-fitting and overfitting, not just produce a nice-looking equity curve.
Logic is built on one slice of historical data and validated on data it's never seen. A strategy that only works in-sample gets discarded.
Trade sequences and price paths are randomly shuffled thousands of times to check whether the edge is statistically real, not just a product of how history happened to unfold.
Historical trades are resampled with replacement to build a distribution of likely outcomes, so we report a range of expected performance instead of a single curve.
The in-sample/out-of-sample window is rolled forward through time, re-validating the strategy across multiple market regimes rather than one historical stretch.
Inputs are nudged away from their optimal values to confirm performance doesn't collapse — a sign a strategy was curve-fit rather than genuinely robust.
Every backtest bakes in commissions and a conservative slippage assumption per fill, rather than assuming perfect execution at the stated price.
Backtests appear on this site clearly labeled as such, for transparency into methodology — not as a substitute for live results.