guideshiny / quantitative-trading-strategy-based-on-machine-learning Goto Github PK

This project forked from majiajue/quantitative-trading-strategy-based-on-machine-learning

Firstly, multiple effective factors are discovered through IC value, IR value, and correlation analysis and back-testing. Then, XGBoost classification model is adopted to predict whether the stock is profitable in the next month, and the positions are adjusted monthly. The idea of mean-variance analysis is adopted for risk control, and the volatility of the statistical benchmark index (HS300 Index) is used as a threshold for risk control. Back-testing results: the annual return rate is 11.54%, and the maximum drawdown is 17.91%.