The question
Would you have found it in time?
Walk-forward analysis answers the only question that matters: if you had re-optimized on the past and traded the result on the unseen future — window after window, for decades — would it have worked? AniQuant re-optimizes each window for real and aggregates it into a weighted AQ WF Score.
The trap
Overfitting has a probability. Measure it.
When you test thousands of variants, the best one is partly best by luck. The PBO/DSR module (combinatorially symmetric cross-validation) estimates how likely it is that your winner is a mirage — before you find out the expensive way.