The ML4T Software Ecosystem
Five production libraries covering the full workflow, from data acquisition to live deployment.
End-to-End Workflow
Each library covers one stage of the ML4T pipeline.
ML4T Data
Unified market data acquisition from 19+ providers
High-performance market data management library with unified multi-provider interface. Handles data ingestion, normalization, storage, and retrieval across multiple asset classes and data frequencies. Supports 19 data providers with efficient Parquet …
ML4T Engineer
Features, labels, alternative bars, and leakage-safe dataset preparation
High-performance quantitative finance feature engineering library. Provides efficient computation of 120+ technical indicators, fundamental factors, triple-barrier labeling, and alternative data features for ML model training.
ML4T Diagnostic
Feature validation, strategy diagnostics, and Deflated Sharpe Ratio
Diagnostic and evaluation framework spanning both feature research and backtest analysis. Validates signals with Information Coefficient (IC) analysis and feature importance, then evaluates strategies with Deflated Sharpe Ratio (DSR), overfitting …
ML4T Backtest
Event-driven backtesting with realistic execution
State-of-the-art event-driven backtesting engine for quantitative trading. Supports complex order types, realistic execution simulation, multi-asset portfolios, point-in-time correctness, and seamless integration with ml4t-engineer features.
ML4T Live
Production trading with broker integrations
Live trading platform for ML4T strategies. Connects backtested strategies to live markets with broker integrations (IBKR, Alpaca), real-time data feeds, safety controls, and production-grade execution management.