Feature Engineering for Traders

Many traders who use machine learning keep swapping models while feeding them the same mediocre features. This book argues that better inputs often matter more than better algorithms, and shows you how to design them.

Feature Engineering for Traders takes you from foundational price and volatility features to an advanced feature stack. You will learn how to select, evaluate, and prune features, and how to do so without sliding into leakage, redundancy, or feature bloat. Several chapters go deep on order flow, including advanced order flow features and ways of detecting institutional activity. You will build cross-asset and temporal features that give a model wider context.

The final chapter shows how to shape features for different model families.