hudson-and-thames/mlfinlab

Implements the complete ML pipeline for quantitative finance—from alternative data structures and labeling schemes through feature engineering, cross-validation, and bet sizing to backtest overfitting diagnostics. Built around financial-specific techniques like codependence measures, synthetic data generation, and clustering methods rather than generic ML frameworks. Includes modules for feature importance analysis, hyperparameter tuning, and network analysis tailored to portfolio construction and risk management workflows.

4,590 stars. No commits in the last 6 months.