An Empirical Study On Stock Market Return And Investor Sentiment In China

In the past 20 years,China's financial market developed rapidly,and great changes have taken place in transaction scale and the number of investors.However,compared with the financial markets of western developed countries,China's financial market has some characteristics,such as short developing time,imperfect system,unexperienced investors and excessive government intervention.Financial market is a barometer reflecting the trend of national economy.With the acceleration of economic globalization,the degree of opening to the outside world is getting higher and higher in China,and the real economy is more and more affected by the fluctuation of financial market.Understanding the relationship between investor sentiment and China's stock market helps to protect the interests of small and medium investors,improve the supervision level of the whole market,and stabilize the development of the stock market.The "rational man hypothesis" and "efficient market hypothesis" in traditional finance have various puzzles,which can't explain some financial anomalies well.Behavioral finance,as a new discipline which integrates psychology,behavioral science,finance and sociology,makes up for the shortcomings of traditional finance.Scholars at home and abroad have developed many theories based on behavioral finance research,among which investor sentiment research has been popular in recent years.So far.there is no unified definition of investor sentiment.When measuring investor sentiment,they often use a single emotional indicator,or when building comprehensive indicators,they do not use the combination of subjective and objective indicators.Many researches are limited to the relationship between investor sentiment and stock market,or only based on the traditional linear model to use investor sentiment to predict future stock market returns.This study enriches the behavioral finance theory,constructs a comprehensive and effective index of investor sentiment,and uses non-traditional methods to predict future stock market returns.Firstly,this paper constructs the investor sentiment index CISI based on principal component analysis by using three subjective sentiment indicators and five objective sentiment indicators.Secondly,based on VAR model,this paper conducts Granger causality test,impulse response analysis and variance decomposition analysis on investor sentiment indicator and stock market returns,and tests the causality and dynamic relationship between them.Finally,this paper uses GARCH model and the Gradient Boosting Decision Tree GBDT algorithm in supervisory learning to predict the closing price of the stock market,and compares their effects.The results show that(1)CISI,a comprehensive index of investor sentiment,fully reflects the information of eight single sentiment indicators,the correlation coefficient between CISI and the closing price of Shanghai Composite Index is 0.708,and the validity is achieved;(2)There is a one-way causal relationship between investor sentiment and the change of Shanghai Composite Index stock price,and the historical fluctuation of Shanghai Composite Index stock price has no significant impact on the current change of investor sentiment.The change of investor sentiment is the Granger cause for the change of Shanghai Composite Index's stock price.Investor sentiment changes contributed 42%to the variance of closing price changes.The variance contribution rate of Shanghai Composite Index closing price changes to investor sentiment changes finally stabilized at 10%.There is a long-term dynamic relationship between them;(3)After comparing the prediction errors,it is found that the GBDT algorithm is better than GARCH model in predicting the closing price of stock market,and GBDT algorithm has a certain reference value in predicting the short-term stock price.