Research On The Relationship Between Investor Sentiment And Stock Market Volatility Based On Text Mining

Since the establishment of the Shanghai stock exchange in 1990,China’s stock market has expanded rapidly and developed rapidly.By the end of 2018,1,449 companies and 12,373 securities had been listed.In China’s stock market to mature in the course of development,investors is an important part of in the market,but due to the immature market,such as affected by the policy,information asymmetry in severe cases,easy to produce surge phenomena such as stock market slump,combined with investor’s lack of financial knowledge and psychological cognitive biases,cause and there are many irrational behavior in stock market,make more severe stock market fluctuations.At the same time,with the continuous development of network and the advent of the era of Internet,more and more investors to use the Internet for stock investment,and real time on the network platform of leaving their views or express their emotions,positive views bullish on the stock market will leave buying on the Internet,the information such as warehouse and hold shares,on on the stock market is bearish negative attitude will leave to sell on the Internet,underweight and launch information.In this context,it is possible to study the interaction between investor sentiment and the stock market.Based on the theory of behavioral finance and investors’ psychological cognitive bias,this paper describes the irrational behavior of investors in the stock market by constructing appropriate investor sentiment indicators.Python statements based on Oriental wealth after post within the site stocks are done and using text mining technology based on data analysis after cleaning,make the text data quantization for investor sentiment index,thus to reflect the characteristics of the stock market volatility four indicators combined with building SVAR model to study the interaction between investor sentiment and the stock market fluctuations affect the relationship.The main conclusions of this paper are as follows: firstly,in text elimination processing,the selection of classification model is made by using randomly extracted training sets,and the naive bayesian classification algorithm is found to be more effective;Secondly,through the analysis of investor sentiment index volatility chart,it is found that the calculated investor sentiment index is roughly consistent with the trend of the stock market,indicating that the constructed investor sentiment index is reasonable and feasible.Finally,through the establishment of five yuan third-order SVAR model research investor sentiment index and the reflection of the trading volume and volatility of the stock market fluctuations,turnover rate and yieldrelationship between the influence of these four characteristics index,found that investor sentiment have short-term effects on these indexes,and the four indicators affect investor sentiment existence lag effect.