Quantitative Stock Selection Strategy Based On Machine Learning

The economic foundation determines the superstructure of the society,and the sustainable development of a society cannot be separated from the prosperity of the economy.Investment,as one of the "three carriages" driving the social economy,plays an irreplaceable important role.Since the financial crisis in 2008,the global economic situation is constantly changing,and China's economic development structure is also in the process of transformation,which leads to the increasingly complex investment environment in China's capital market.Compared with foreign countries,China's capital market started relatively late,with a history of only more than 30 years,but its development speed should not be underestimated.No matter from the investment amount,the number of investors or the total number of listed companies,China's capital market has a considerable scale,and the development of the capital market has effectively promoted the financing of various large and small enterprises in China and the investment of residents and institutions.And the complicated financial environment and the rapid development of computer technology make the modern investment theory innovation,and also the disadvantages of traditional fundamental analysis and technical analysis are emerging:on the one hand,due to the continuous expansion of the capital market,the traditional statistical model based on small samples has been unable to deal with the huge amounts of data in today's capital markets;On the other hand,a large number of investment practices show that investment strategies based solely on fundamental analysis and technical analysis are subject to the subjective experience of investors and a variety of greed and fear emotions,leading to unnecessary failure of investment strategies.Under the comprehensive promotion of various factors,quantitative investment theory began to rise.Due to its strict discipline,systematicness,probability and diversification characteristics,quantitative investment strategy has a good performance in the domestic and foreign investment markets.The purpose of this paper is to establish a machine learning model to study the applicability of quantitative stock selection strategy in China's securities market.First of all,through reading a large number of domestic and foreign literature about quantitative stock selection,understand the research status of quantitative stock selection and mainstream research methods;Then,based on the history of China's capital market,this paper selects the suitable machine learning algorithm to predict the stock returns of all companies in Shanghai and shenzhen a-share markets.The machine learning algorithms used in the paper is support vector machine(SVM)and random forest algorithm(RF),based on from 2013 to 2017,the annual financial data and the corresponding annual yield of each stock in China's Shanghai and Shenzhen a-share markets(excluding ST stock).Support vector machine(SVM)and random forest classification model are set up respectively,with the method of back to test,respectively for 2015,2016 and 2017 the stock yield forecast classification.According to the classification results,the top 10 stocks,20 stocks and 30 stocks with the largest rise probability are selected each year to establish portfolio 1,portfolio 2 and portfolio 3.The average return rate of each portfolio is calculated and compared with the benchmark return rate of hushen 300 index in the same period.On the surface of the results,the quantitative investment portfolio established by the support vector algorithm and the random forest algorithm both beat the market,with the annual return rate exceeding the benchmark return rate of the hushen 300 index in the same period,and the performance of the portfolio market constructed by the random forest algorithm was slightly better than that of the portfolio constructed by the support vector mechanism.It can be seen that the study of this paper has a good guiding significance for the actual investment.