Research On Quantitative Investment Strategy Based On Public Opinion Data

So far,China’s securities market has developed for more than 30 years.During this period,the investment enthusiasm of the general public has gradually increased.China’s stock market has obvious "retail characteristics".The proportion of individual investors in the market is high,and the phenomenon of irrational investment is significant,which is very different from the foreign mature markets that institutional investors account for the majority.Therefore,we should not completely copy other mature foreign markets.We need to build a quantitative trading strategy suitable for China’s financial market.Quantitative investment is based on the theory of mathematical statistics,with the help of computer technology to analyze the massive data in the stock market,trying to explore the change law of stock price,so as to obtain a certain excess return.At present,it has become one of the main methods in the field of financial investment.Compared with foreign countries,China’s quantitative investment started late,but developed rapidly in the later stage.Colleges and universities and major fund companies began to study quantitative investment technology.Many quantitative trading platforms such as Datayes,RiceQuant,JoinQuant and TradeBlazer also appeared.Combined with the high proportion of individual investors in the domestic stock market,and different from previous studies using proxy investor sentiment variables,this paper takes GubaSenti and BaiduIndex into account,which directly represent the sentiment of investors,and selects stocks and designs timing strategies based on sample data,and obtains a relatively feasible back-test return.This paper uses CSI 300 constituent shares from 2012 to 2019 as the research sample.Firstly,for the samples from 2015 to 2018,using the data of 19 factors in the seven categories of growth factors,valuation factors,quality factors,capital structure factors,operating capacity factors,momentum factors and technology factors,a multi factor stock selection model is constructed through the principal component regression method.The difference between the expected rate of return and the actual rate of return of the current year is used as the index to measure the growth ability of stocks,and then select the stocks with high growth ability as the stock pool to be selected in the next trading year.Secondly,for the samples from 2012 to 2019,public opinion factors indirectly reflecting investor sentiment are constructed through daily proxy sentiment data,and comprehensive public opinion factors are constructed through daily proxy sentiment data and Baidu Index and GubaSenti index which directly reflecting investor sentiment.Based on investor sentiment factors,the signals of stocks contained in the stock pool are predicted by SVM model,and the simulated trading back test is carried out.Then,the cumulative net value of the two SVM quantitative timing strategies based on indirect investor sentiment and comprehensive investor sentiment are compared with the buy and hold stock selection strategy and CSI 300 index respectively.The empirical results show that the two SVM quantitative timing strategies based on investor sentiment can obtain higher excess returns than the other two strategies.Finally,compared the back test results of the two SVM quantitative timing strategies based on different emotional characteristics,the results show that the quantitative timing strategy constructed according to the comprehensive investor emotional characteristics that considered the direct investor emotion has a better performance,that is,the direct investor emotion can improve the effect of the quantitative investment model to a certain extent.