Research On Random Forest Multi-factor Stock Selection Strategy Based On Industry Rotation

Quantitative investment,with its extraordinary performance,has gradually entered the vision of investors,and has overturned the traditional investment philosophy in the past 40 years,known as the "revolution in the investment world".With the rapid development of computer technology,this model has made more rapid progress and development.Many asset management companies that use quantitative strategy as their main investment technology have become outstanding among their peers.This also shows that the concept of quantitative investment has become popular.Quantitative investment technology has gradually become one of the main investment methods.Research on new quantitative investment methods and Mining new modeling ideas is of great significance to the development of quantitative investment.This paper constructs a multi-factor stock selection model.Before constructing a multi-factor stock selection model,the industry pooling strategy is used to conduct preliminary screening of the stock pool to enhance the model's stock selection performance.The CSI 300 index component stock data is used for empirical analysis.This article is optimized in the following aspects: First,in the selection of factors,analyst sentiment factors are added.The analyst sentiment factors mainly reflect the analyst's credit evaluation and profit expectations of various listed companies,such as comprehensive ratings,expected earnings per share,etc..Second,when constructing an industry rotation strategy,on the basis of analyzing the theory of investment clock rotation,firstly study the feasibility of investment clocks in the industry configuration through a measurement model,and then examine the investment clocks in China's stock market industry configuration through historical data statistical inspection.Medium effect.Third,before constructing a multi-factor stock selection model,this article combines industry rotation strategies based on the multi-factor stock selection model,effectively connecting the two fundamental models of quantitative investment,and using the industry rotation strategy for preliminary screening.Stock pool,on this basis,research how to use machine learning methods to combine with traditional multi-factor stock selection model,and build a multi-factor stock selection model based on random forest algorithm,simplifying the steps of traditional multi-factor model,using training while filtering The way.In the selection of machine learning algorithms,the advantages and disadvantages of support vector machines are compared,and it is confirmed that random forest is the most suitable for performance and stability.According to the above design ideas,the CSI 300 index stocks from 2012 to December 2018 were selected to construct a multi-factor stock selection model based on industry rotation.After optimization and modification,the backtesting was performed,and the extremely high yield,Obtained a higher rate of return than the market and related indexes and strategies,which has certain feasibility and practical significance,and at the same time the design and development of existing stock selection strategies and future stock selection strategies of fund companiesProvides new ideas.