Quantitative Stock Selection Strategy Based On Method Of Variable Selection:An Empirical Analysis Of Chinese Stock Market

Since the establishment of modern financial theories such as the capital asset pricing model and the efficient market hypothesis,market disparities beyond the theoretical scope have also emerged.These cross-sectional return are unexplained by traditional financial theory and are anomalous phenomena in financial markets.The research on market vision is a hot spot in the field of asset pricing.Scholars at home and abroad have done a lot of research in order to explore the characteristics and formation mechanism of cross-sectional return phenomenon in the market.The industry is also paying close attention to this,with a view to studying the cross-sectional return phenomenon and formulating a stock picking strategy to obtain excess returns.The purpose of this paper is to investigate whether the new characteristic variables constructed by Partial Least Square(PLS)have a predictive effect on stock cross-sectional return,and to guide investors in quantitative stock selection.In addition to the commonly used methods such as portfolio grouping and Fama-Mac Beth regression,this paper also uses the PLS method to construct new characteristic variables for stock cross-sectional return information.We consider cross-sectional return as explanatory variables.Assuming that company characteristics are related to stock cross-sectional return through one or several common underlying factors,the PLS method can filter out noise information in company characteristics and extract company characteristics.Information about the cross-sectional return of stocks provides a significant predictive effect on the future return of stock cross-sections.This paper uses 2439 stock sample data from January 2003 to May 2016,and uses PLS method to extract new characteristic variables from 11 company characteristics.The portfolio analysis and Fama-Mac Beth regression analysis method are used to empirically analyze the correlation between new characteristic variables and stock cross-sectional return.The conclusions of this paper are mainly as follows: First,the new characteristic variables constructed by the PLS method have a significant effect on predicting stock cross-sectional returns,and are tested by different set robustness.It is proved that the PLS method can be used to construct new variables in the China's stock market,so as to formulate a quantitative stock selection strategy.Second,through the analysis of the correlation between company characteristics and cross-sectional return phenomenon,it also provides evidence for the abnormal phenomenon of asset pricing in China's stock market.Company characteristics such as market capitalization,volatility,and book-to-market ratio are factors that affect cross-sectional returns.