An Empirical Study On The Multiple-factor Stock Selection Model In The A Stock Market

In the current period of big data,the importance of data is self-evident.It plays an extremely important role in social life.In data,we can find the connection between many things and find valuable information.For example,through the satisfaction survey data of consumers in market research,can be found in the current consumer consumption tendency,to adjust business strategy;when education researchers want to understand the learning situation of a group of students,learning the students can send through the data of daily performance,test scores in,to further develop appropriate learning plan.In these seemingly useless,messy and massive data,there are extremely valuable information,such as the unearthed.In the same financial research,we can find the law of economic operation in the financial data,and become an important empirical basis in the modern financial research.Quantitative investment,relying on traditional investment theory,improves the efficiency of research by using the efficient computing speed of computers,and is gradually attracting the attention and love of investors.In recent years,financial activities in China's financial market are increasing frequently,and the scale is also growing.How to get high returns in complex and changeable financial market is a concern for every financial investor and researcher.In the daily market activity,the researchers constantly enrich investment theory,effective treatment of financial data with mathematical methods,a number of new index can reflect the change in the market,and provide a very good reference for the majority of investors,established the research upsurge of quantitative research.The new indicators and theoretical innovation need to deal with large amounts of data,and need more mathematical methods.All of which make people pay more and more attention to the research of quantitative investment.In this paper,the multiple-factor selection model as the research basis,the related theory to study quantitative stock,while adding random forest algorithm and fuzzy C means clustering algorithm to verify the new algorithm can adapt to the laws of the market and to improve the multiple-factor model persuasive purpose.Specifically,this paper selects 200 stocks in A stock market capitalization ranking as the research basis,from 2009 to 2017 the transaction and financial data are processed to verify selected 15 common factors of candidate selection,through a combination of various factors of the stock index to evaluate the cumulative excess return,beat the market benchmark probability random forest,the importance of variables selected 9 effective factors,and using the principle of fuzzy C means clustering algorithm for the effective factor clustering,the best performance index in the selection of all kinds of stock as the final basis,other factors as the redundancy factor rejection model;in eliminating the redundant,using the weight method of scoring in test the samples of various effective factors,each factor score in total size,the comprehensive score of samples of size selection The stock of 20%is used as an investment portfolio to test the performance performance of the portfolio.The empirical results show that the classification principle of random forest variable importance description is very suitable for selecting effective factors,through the training of the classification results of each period,the importance of the whole model in probability statistics of higher ranking,if likely its importance is very high,it can be said that the importance of changes in the income in the whole period is very high,most of the time in the empirical classification error rate is low,the effect is significant,the important factor to a great probability to outperform the market,so it can be used as effective factor selection basis;fuzzy C mean clustering algorithm for processing factor fuzzy concept used to explain the complex relationship between the financial variables.The concept of membership will be presented as a basis for poly variables described as variables belonging to the size of this class,the maximum degree of membership class will be divided into a The investment portfolio formed by adding the above two algorithms into multiple-factor stock selection models can win the market benchmark returns in the inspection period,and get a high excess return,making the reference direction of investors more abundant.