Optimization Of MLP Neural Network Quantitative Stock Selection Based On Bagging Algorithm

Before the rise of quantitative investment,the market is popular with Buffett's "value investment" as the representative of the qualitative investment.Qualitative investment on the personal qualities of investors and market sensitivity demanding,has a certain subjectivity,risk,earnings uncertain.With the rapid development of information technology and computer technology,a quantitative investment analysis method quietly arises.In 1971,Barclays Capital Management released the world's first passive quantitative fund,which opened the mystery of quantitative funds,but also opened the door to quantitative investment.Quantitative investment management integrates the traditional investment portfolio theory and quantitative analysis technology,uses mathematical knowledge and computer software to establish a quantitative model,and through the analysis and processing of large amounts of data,objectively selects the highest-yielding investment portfolio,thereby avoiding the problems of human-induced Cognitive bias and subjective assumptions,which greatly improved the accuracy of the investment.The stock market is a complex non-linear system with a large amount of data.However,machine learning has proved to be a powerful tool for modeling fuzzy non-linear data in many fields.Therefore,the machine learning method is used to quantify the construction of stock selection strategy Has the natural advantage.Based on this,this paper combines Bagging algorithm and MLP neural network,which are relatively mature in machine learning,to construct a quantitative stock selection strategy.This paper takes more than 500 stocks of Shanghai-Hong Kong Stock Connect as the research object,selects 28 technical indexes and financial indexes as input variables,and uses stock classification as output variables.It uses Bagro-MLP neural network modeling with R language software.The empirical results show that the forecasting accuracy of Shanghai-Hong Kong stock data is up to 70% based on the Bagging-MLP neural network model and the average yield of over 500 stocks in the selected Shanghai-Hong Kong Stock Connect is obtained.