An Empirical Study On An Improved Multi-factor Stock Selection Model For A Shares

Along with the rapid development of finance theory and computer science in today's world,the use of computer programming technology to achieve quantitative investment transactions came into being.Since the introduction of quantitative investment into the domestic market,it has received attention from investors and is also controversial.Especially since 2017,the market style has undergone a big change.The biggest feature is that the whole market has obviously changed into a performance-driven market.In addition,value investment returns to the more mainstream investment model,and the short-term investment style is further weakened.Due to the transformation of the style,the quantitative investment strategy of judging the investment target through historical data has undergone a very large adjustment.Among them,there is a different degree of retracement in the quantitative investment strategy exposed by the stock market value.Therefore,in the current domestic market environment,quantitative investment institutions need to dynamically adjust and improve the quantitative investment strategy according to different market styles,so that the improved quantitative investment strategy can adapt to the continuous rotation style of the domestic A-share market.Based on the multi-factor stock selection model and the common factors,this paper explores the market value factor,and in the process of single-factor to multi-factor integration,the original multi-factor model is improved through symmetric orthogonal of factors and ICIR weighting(the IC average of the factor over the past period divided by the value of its standard deviation as its weight vector).Improvements are made and tested for effectiveness in the domestic Ashare market.Firstly,this paper selects 28 factors that are recognized by the market as long-term effective,as well as the market value factor,and tests their effectiveness by IC(information coefficient)and IR(information ratio).Secondly,in the single factor to multi-factor integration process,by comparing the combined yield and retracement,the symmetric orthogonal mode and the ICIR weighting mode are selected among different orthogonal modes and factor weighting modes.Then the improved multi-factor stock selection model is constructed.Finally,the model is tested in the domestic A-share market,and it is verified that the improved model can obtain a relatively stable excess return.This article provides ideas for other researchers and investors to explore new stock selection factors,at the same time,it also has important implications for factor integration processing.