Walk-Forward Analysis
Out-of-sample testing by rolling a training window forward and evaluating on the next period repeatedly.
Definition
Walk-forward analysis splits history into rolling windows: train (or optimize) on window ( t ), then evaluate on the next out-of-sample window ( t+1 ); roll forward and repeat.
Why it matters
- Reduces overfitting by forcing repeated out-of-sample tests.
- Mimics how we would have used the strategy in real time (no look-ahead).
Common mistakes
- Too short OOS window (noise).
- Peeking at full sample to choose parameters, then “walk-forward” with same params.
- Ignoring transaction costs and regime changes in OOS period.