Optimization

Tools for finding good parameters and testing their robustness.

Tool	Purpose
Parameter Sweep	Parallel grid search across parameter combinations
Walk-Forward	Rolling train/test optimization
Monte Carlo	Randomized robustness analysis

Philosophy

Optimization is dangerous. The best parameters on historical data are often the worst parameters going forward. replaybt provides optimization tools alongside validation tools to help you distinguish robust parameters from overfit ones.

Recommended workflow:

1. Sweep — find parameter regions that work
2. Walk-forward — test if the best parameters are stable across time windows
3. Monte Carlo — estimate the range of outcomes and probability of ruin
4. OOS split — final blind validation on held-out data