Walk-Forward Analysis and Out-of-Sample Testing: How to Actually Validate a Strategy
Everyone says to forward-test a strategy to catch overfitting. Almost no one explains how to structure that testing rigorously. Out-of-sample validation and walk-forward analysis are the methods: optimize on data the strategy is allowed to see, evaluate only on data it is not. This explains how they work, the anchored-versus-rolling choice, the data-leakage traps, and the honest limit that even these methods can be gamed.