Research On Quantitative Value Investmengt Of A Share Market Based On Machine Learning

Value investment theory was put forward by Graham.After decades of continuous improvement and development,it has become a complete investment system.In today’s mainstream investment market,value investment has been widely used.China’s stock market started late,and the initial stage is full of irrational investment and insider trading,which makes value investment useless.However,in recent years,with the strict supervision of the CSRC,the market environment has been improving,the securities market has gradually become standardized,and the investment behavior has become increasingly rational.In this case,value investment in China’s securities market will have a broad space for play.According to the actual situation of China’s stock market,the empirical study of value investment theory is of great practical significance.On this basis,this paper will try to expand from the theoretical and empirical aspects.This paper focuses on three issues.(1)how to identify companies with excellent performance and good growth potential and investment value from more than 3,000 listed companies in the a-share market?(2)how to identify companies with investment value that are undervalued by the market so as to establish a stock pool.(3)how to configure the stock weights in the stock pool to meet the needs of different investment styles.This paper combines theoretical analysis and empirical research to study the above problems.Firstly,the traditional value investment theory and other three investment theories(growth investment theory,index investment theory and portfolio investment theory)are elaborated in the theoretical analysis,and then the advantages of other investment theories are integrated on the basis of traditional investment and reflected in the empirical part.In the empirical analysis,for the first question,this paper adopts the BP neural network method to screen 2630 listed companies with complete data.The screening criteria are based on the fundamental information of listed companies from 2014 to 2017,and 656 listed companies with investment value are finally selected.In view of the second problem,this paper USES support vector and BSM model to evaluate the selected listed companies,find out those undervalued stocks,and finally find out 91 undervalued listed companies,and establish the stock pool.For the third question,this article USES five different styles of weights allocation strategy(price weighted method,market value weighted method,such as weight method,mean variance model,as well as the BL model)to weight ratio of shares in the pool,and compares the different styles of weight allocation strategy,finally obtained the weight method such as configuration has the highest return on equity portfolio,and finally the weights of the five different style allocation strategy of yield compared with the yield of the csi 300 index,obtained by A share value investment strategy based on machine learning can get excess returns.The feature of this paper is that it combines the advantages of other investment theories on the basis of value investment theory,and applies the method of machine learning to the practical operation of value investment theory,and finally finds that the value investment strategy based on machine learning can obtain excess returns in A stock market.