A Study On The Relationship Of The Investor's Sentiment Mined From Wechat Text Information And Stock Market Performance

Mining and selecting investor sentiment indicators is a hot topic in behavioral finance research.Behavioral finance theory,points out that the current stock market does not conform to the efficient market hypothesis,investors cannot be totally rational.Therefore,the key to study the performance of the stock market is to explore indicators that can reasonably and effectively reflect investor sentiment.Compared with the traditional investor sentiment proxies and investor sentiment directly mined from weibo,BBS and other public social media,the investor sentiment mined from WeChat text has a higher credibility.Because WeChat relies on interpersonal network to build and the spread of information in WeChat is mainly dependent on the "circle" of acquaintances,this natural trust from interpersonal network makes the information in the circle of friends easy to be forwarded and intersect between circles.Therefore,this paper makes an innovative use of the WeChat text to mine investors' sentiment,and obtain investors' sentimental data that is more closely related to investors' core and stronger credibility.In this paper,we mined investors' sentimental data from WeChat public articles about the stock market from August 3,2015 to March 24,2017,and collect the corresponding period of the Shanghai composite index.Using correlation analysis and Granger causality test to verify whether investor sentiment is the cause of the change of the stock market closing price,trading volume and the rate of return.Then based on the relationship between the investors' sentiment mined form WeChat text and the performance of the stock market,we established a stock price forecasting model based on support vector machine.The results show that:(1)The investor sentiment contained in WeChat data is closely related to the stock market performance,the inf luence of investors' sentiment and investors' attention mined from WeChat to the stock market performance significantly different;(2)the negative sentiment tendency of investors based on WeChat text has a stable ability to predict stock price;(3)the change of stock market closing price and trading volume can predict the change of investors' attention based on WeChat text mined;(4)the SVM stock-price prediction model based on the investors' negative sentiment indicators mined from WeChat text is more accurate.