stefan-jansen/machine-learning-for-trading

Contains 150+ executable notebooks demonstrating ML signal extraction from market, fundamental, and alternative data (SEC filings, satellite imagery), alongside model training and strategy backtesting workflows. Covers supervised and unsupervised learning through deep reinforcement learning, with implementations of recent research including CNNs for time-series-to-image conversion, autoencoders for asset pricing, and GANs for synthetic data generation. Integrates a customized Zipline backtesting engine for evaluating ML-driven trading strategies end-to-end, using modern Python libraries (pandas, TensorFlow 2.2+) across multiple asset classes and data frequencies.

16,745 stars. No commits in the last 6 months.