Glossary
Walk-Forward Analysis
A validation method that repeatedly optimizes a strategy on in-sample data and tests it on subsequent out-of-sample data.
Walk-forward analysis simulates real-world trading by rolling an optimization window forward through history. It helps detect curve-fitting because the strategy must perform on data that was not used during parameter selection.
Key Points
- More robust than a single train/test split.
- Shows whether a strategy remains effective across different market regimes.
- Use it before live deployment and after major strategy updates.