Summary

ai_quant_trade is a comprehensive open-source AI quantitative trading learning repository on GitHub with over 5,100 stars and nearly 1,000 forks. Written natively in Chinese and tailored to the A-share market environment, it covers the full quant workflow: data processing, strategy research, backtesting, and live deployment. Recent 2025 updates add large language model (LLM) capabilities, including a reasoning-based stock price prediction model trained with Unsloth (reporting 20% accuracy improvement with interpretability), automated research report generation, and financial RAG applications. The project includes modules for traditional strategies (dual moving average, small-cap) using backtrader, reinforcement learning (PPO, DDPG) via FinRL and Stable-Baselines3, deep learning models (LSTM, GNN) with PyTorch and TensorFlow, data integrations (Wind, Baostock, Tushare), and a factor library featuring Alpha101 and tsfresh with 5,000+ factors. It targets students, career changers, experienced quants, and retail investors, offering Jupyter Notebook tutorials and Excel-based market monitoring tools. Repository is available on GitHub and Gitee for educational purposes only.