Empirical Research On Investor Sentiment And Stock Return,Volatility Based On Net-mediated Public Sentiment

The study of effect of investor sentiment on the stock market is one of the main topic of behavioral finance.With the development of related research,the problem is no longer whether investor sentiment influences stock return or not,which has become how to measure investor sentiment accurately.This paper uses online text data as direct measure of investor sentiment.To finish this job,we construct a sentiment dictionary for stock market specially,then we use an unsupervised method for sentiment classification.Finally,we construct several indexes which reflect investor sentiment,which includes Investor's Attention,Optimistic Index,Pessimistic Index etc.Compared with previous research,our indexes are much more accurate and direct.Based on the above work,this paper construct a GARCH-M model to analyze the influence of investor sentiment on stock return and volatility.A classification model is also built to predict the fluctuation of stock market.The main conclusions are as follows:First,Optimistic Index influences market return significantly and positively.On the contrary,Pessimistic Index influence market return negatively and significantly.However,the influences of optimistic index and pessimistic index on market return are not symmetry in period one.Pessimistic Index has a larger influence than Optimistic Index,which means there is herd effect on pessimistic investors.Second,volatility cluster is shown in period one,a higher volatility is followed by another one,while this effect is not obvious in period two.Finally,Investor Sentiment has different predictive capability for stock market in different period.The prediction accuracy is higher when market has a clear trend.