基于机器学习的优化股票多因子模型

Optimized Stock Multi-factor Model Based on Machine Learning

笔者旨在构建机器学习优化股票多因子模型,用以处理A股市场风格切换和选股问题来最终获得超额收益,首先构建因子分析模型来筛选出7个最优因子,进而构建基于机器学习的随机森林模型,通过随机森林回测某段时间的股票波动情况。该模型分别从因子表达、机器学习算法两个角度对A股市场股票的波动规律进行研究,获取最大回撤的超额收益。笔者使用公开的2016年1月1日至2018年9月30日我国A股市场的数据对算法性能进行评估。实验结果显示回测的正确率为83%,收益的平均利率约为1.57%。

The purpose of this paper is to build a multi factor model of machine learning to optimize the stock market,which is used to deal with the style switching and stock selection of A-share market in order to ultimately obtain excess returns.First,we build a factor analysis model to screen out seven optimal factors,and then build a random forest model based on machine learning to test the stock volatility in a certain period of time through random forest.In this model,we study the volatility of A-share market from two aspects of factor expression and machine learning algorithm,and obtain the maximum return.The performance of the algorithm is evaluated by using the data of China’s A-share market from January 1,2016 to September 30,2018.The experimental results show that the correct rate of back testing is 83%,and the average interest rate of return is about 1.57%.