Font Size: a A A

Empirical Research Of Investor Emotion And Stock Market Return Based On Text Analysis

Posted on:2020-12-24Degree:MasterType:Thesis
Country:ChinaCandidate:T LiFull Text:PDF
GTID:2439330596493438Subject:Applied statistics
Abstract/Summary:PDF Full Text Request
Research on investor sentiment has always been a hot topic in the financial field,and there are still major differences between domestic and foreign research on investor sentiment and financial field.Therefore,it is more accurate to measure these irrational behaviors such as psychological emotions,analyze and clarify the influence effect between them and the stock market,thus injecting fresh blood into the exploration of online public opinion and financial markets.This thesis takes Sina financial stocks and Southwest Securities Golden Ideas Wealth Management Terminal as data sources,collects the texts of SSE 50 community shareholders' comments through web crawling.A series of preprocessing such as denoising,word participle,and stop words deletion are performed on the original data.Based on other dictionaries such as HowNet,Tsinghua University,etc.,the Sogou cell vocabulary is used to construct a special dictionary for stocks.In the emotion index calculation algorithm,we deal with degree words,exclamatory sentences,interrogative sentences,and so on.The final score is subtracted from the positive score by the positive score,directly measuring the positive and negative emotions of the investors,and the results are resonable.Finally,the investor sentiment index value and the Shanghai Composite Index are integrated and analyzed.The main conclusions are as follows:Firstly,there is a significant positive correlation between investor sentiment indicators and the yield of SSE 50 composite stocks,but there is a negative correlation between closing price and stock trading volume.And the higher the closing price,the higher the yield,the more stocks trade,but the investor's sentiment is not necessarily optimistic.Secondly,in the Granger causality test,when the sentiment index lags behind in the third period and later,the p value is 0.04924,indicating that the stock return rate has a great effect on the sentiment index.Thirdly,in the impulse response analysis,the stock return rate has a great influence on the investor's sentiment in the current period,and it is a positive influence.As time goes by,this influence gradually weakens,and disappears in the fourth period basically,indicating that there is a relationship inside the investor's sentiment after the fourth period.The final test is the variance decomposition.In the prediction of the stock return rate,the error of the first forecast value comes from itself.In the following periods,the proportion of the investor's sentiment index increases significantly until the fourth period and later reached a stable state;For the prediction of the investor's sentiment index,1.31% comes from the stock return rate in the first period.In the fourth period and later,the variance of the forecast error of the yield begins to stabilize.It is indicated that in the VAR model prediction,it is necessary to pay attention to the influence of other endogenous variables on the predictor variables.
Keywords/Search Tags:Text mining, Emotion analysis, Behavioral finance, Investor sentiment, VAR model
PDF Full Text Request
Related items