Research on quantitative multi-factor stock selection model based on machine learning

With the popularization of computer technology, the quantitative investment business in China has been developing rapidly, but it started late compared to the developed markets, and its related theories and applications have received very high attention in the investment field in China. In this paper, we focus on the effectiveness of multi-factor stock selection model applied to A-share market and use machine learning algorithms to improve the performance of the model. First, this paper selects 10 relatively independent valid factors from 44 candidate factors by validity testing and redundancy removal; second, builds a multifactor model and tests its validity, and introduces a hedging mechanism to further reduce the maximum retracement of the model; then changes the factor retracement period to select the model with the best effect for backtesting; finally, combines the AdaBoost algorithm with the multi-factor model and and the traditional Finally, the stock selection effect of combining AdaBoost algorithm with multi-factor model and and traditional multi-factor model is compared. The study finds that the machine learning AdaBoost [1] algorithm can enhance the stock selection effect of the traditional multi-factor model, and this study has a significant role in the future development of this field.