Walk-Forward Analysis

Walk-forward analysis is a robustness test for rule-based and automated strategies. Historical data is split into consecutive segments: parameters are optimized on an in-sample window, then applied unchanged to the next out-of-sample window, and the pair of windows rolls forward until the data is used up. Stringing the out-of-sample results together produces an equity curve built only from data the optimizer never saw, which is a far more honest picture than one backtest run over the whole period. Walk-forward optimization is used mainly to expose curve fitting: if performance collapses out of sample, the parameters were describing past noise rather than a repeatable edge. It cannot prove a strategy will work in future, only that it survived a stricter test.