Research On The Relationship Between Investor Emotion And Volatility In China's Stock Market

Different from traditional financial research methods,behavioral finance studies the abnormality of financial market by observing the non-consistent expectation and irrational behavior of market participants and predicting the impact of irrational behavior on asset pricing.fluctuation.This irrationality in the Internet age is often reflected in network information.Therefore,if we can more accurately identify the network public opinion information and judge its impact on the stock price in the short term according to the behavioral financial model,it will be able to effectively improve the network public opinion on the stock.The price impact mechanism and provide a basis for the implementation of relevant policies.This paper first uses network technology to collect and acquire network information related to investor sentiment,and builds a sentiment index based on statistical methods and deep learning methods to characterize investors' emotions.Specifically,it includes two aspects of the sentiment index: the investor sentiment index based on the comments of the Oriental Wealth Stocks and the investor search index based on the Baidu search index.Based on the sentiment index of the comment text,the latest model of the current NLP domain,the Google-BERT network,is applied to obtain the emotional status score of the investor by constructing the emotional classification problem of the text.The sentiment index based on the Baidu search index is applied based on The elastic network and the dimensionality reduction method of principal component analysis construct an evaluation index that reflects investors' attention to different aspects of the stock market.Based on the previous work on investor sentiment index,this paper applies the vector autoregressive(VAR)model to study the investor sentiment index based on the data from January 4,2011 to June 29,2018.The interaction between stock market yield volatility.The results of Granger causality test show that there is a mutual causal relationship between investor sentiment index and logarithmic rate of return,and there is also a mutual causal relationship between the two sentiment indices.The results of the impulse response analysis show that the commentary sentiment index or the search index will have a significant positive or negative impact on the fluctuation of the Shanghai Stock Exchange's rate of return,and gradually decrease over time and eventually converge to zero.The results of analysis of variance show that the fluctuation of the return rate of the Shanghai Composite Index is mostly from itself,and the contribution of the variance index will be larger in the short term.In the long run,the contribution of the search index will be greater,that is,the emotion in the comment is often It's a temporary one,and search behavior tends to have a more lasting impact.Finally,based on the multi-information LSTM-CNNs network,the fluctuation of the Shanghai Composite Index was predicted based on the existing sentiment index,and was carried out on the data from December 7,2016 to June 29,2018.test.Preliminary predictions show that the model can achieve an accuracy of 79.3% on the three-category problem.Further based on the actual trading rules,the backtesting found that the modelbased trading strategy can obtain an excess return of 20.15% compared with the holding,but with a maximum retreat of 5.64%,it has certain predictive ability.