An Empirical Study Of Quantitative Stock Selection Strategies Based On Internet Public Opinion

In the internet era,investors access information on financial markets,stock trading and exchange activities are almost carried out through the internet.In particular,online platforms such as stock bars and forums gather a large number of individual investors and institutional investors,who will publish their views on the stock market or the trend of a certain stock on these platforms,and will also comment and exchange views with each other.Behavioral finance holds that people are irrational and vulnerable to the impact of their environment,which also provides a new perspective for the study of stock market investment: researching the network public opinion of stock bars,forums and other online platforms to explor whether it can affect the stock market and investment.In order to research whether the index of network public opinion factor has an impact on the stock selection ability of the quantitative stock selection model,this paper use SVM to construct a quantitative stock selection model based on network public opinion.Firstly,the online stock appraisal texts of HS300 from 2017-08 to 2018-08 are collected by the web crawler,and the network public opinion factor values are calculated after emotional analysis of stock appraisal.Then,adding the network public opinion factor and the seven fundamental and technical factors tested by the validity test to the multi-factor quantitative model,and using the first 3 months factor data to optimize the model parameters and the last 9 months factor data are used to test the results of the quantitative stock selection model with and without the network public opinion factor.The results show that:(1)During the back test period,the quantitative stock selection model based on the network public opinion obtained an excess income of 8.07%(the benchmark is the index of HS300),while the excess income of the quantitative stock selection model that did not contain the network public opinion during the same period was 5.47%(the benchmark is the index of HS300).(2)Compared with the quantitative stock selection model without network public opinion,the maximum drawdown rate is reduced by 1.78%,and a more stable return is obtained.(3)Comparing the quantitative stock selection model without the network public opinion,the model with the network public opinion performed better on the back-test indicators such as Alpha value,daily winning rate and IR value,which effectively improved the stock selecting ability of the model.The study finds that there is a relationship between the stock market and investor sentiment.The network public opinion factor is an important factor affecting the stock selecting ability of the model.The research on network public opinion can not only effectively improve the stock selecting ability of the quantitative stock selection model,but also can provide theoretical support for the selection of new quantitative stock selection factors.