An Empirical Study On Investor Sentiment And Stock Market Based On Online Public Opinion

At present,with the rapid development of Internet social media,ordinary investors can express their emotions and opinions through the Internet.In this context,the amount of investor sentiment information based on online public opinions is growing at an unimaginable rate,and the research on investor sentiment based on online public opinions emerges.At the same time,traditional financial theory has long been questioned,behavioral finance has gradually established its position in theory and practice,and stock market research based on investor sentiment has become the mainstream.First of all,this paper selects the stock bar of Eastmoney as social platform of Chinese stock investors,and constructs the index of investor sentiment through the text data from Eastmoney.The data was obtained with the help of web crawler tools and the sentiment analysis with the help of the Bert-LSTM model,and finally the investor sentiment index was established.Secondly,this paper analyzes the correlation between investor sentiment index and stock market returns.First,the descriptive statistical results of investor sentiment and stock market returns show that Chinese mass investors are emotional and have a general tendency to complain.The overall performance of the monthly investor sentiment index is negative.The monthly investor sentiment index is correlated with the monthly stock yield,and the intuitive trend is relatively consistent.Second,the daily investor sentiment index has the strongest relationship with the stock yield in the current period,and the daily investor sentiment is positively correlated with the stock yield.Thirdly,the relationship between daily investor sentiment index and stock yield is slightly different in different periods.In the rising stage of the market,the correlation between investor sentiment and stock yield is greater than that in the falling stage.Fourth,the conclusion of granger causality test shows that investor sentiment is the granger cause of stock returns.Finally,this paper makes an empirical study of the FF3 model of investor sentiment and stock returns.The results show that the classical FF3 model has a good empirical effect in this paper,and the goodness of model fit of each stock combination reaches 79.70% on average.From the point of view of each influence factor,the performance of excess return factor and scale factor is better,and the performance of book-to-market ratio factor is general.After the investor sentiment factor added,the effect of the FF3 model was improved.The mean goodness of fit of each stock combination reached 81.73% on average,and the investor sentiment factor was significant.It was believed that investor sentiment was beneficial to the improvement of the model and had a promising effect on the prediction of stock returns.In summary,investor sentiment based on online public opinions has a significant relationship with the stock market,and the Chinese stock market represented by the SSE 50 index has an obvious emotional bias.From the conclusion,this paper puts forward countermeasures and Suggestions for investors,listed companies and market regulators.For investors,they should fully understand their own emotions,make their own investment plans and track their records,and finally find and adhere to the investment strategy that suits them.For listed companies,they must ensure the accuracy and efficiency of information release and improve the communication mechanism with investors.For market regulators,they should strengthen investor education and guidance,establish early warning of investor sentiment in the market,strengthen supervision of information disclosure of listed companies,and strictly control market manipulation,insider trading and other illegal behaviors.