Research On Investor 's Emotion And Stock Returns Based On Public Opinion

With the development and improvement of internet technology, stock forum become a main platform for investors to communicate with each other. A lot of investors get information from stock forum and use it as reference when they make decisions. At the same time, investors also post in stock forum to express their opinions. So in stock forum, there are quite a few text data which reflect investor sentiment. And in behavior finance theories, in real world, people are limited rational, investors are easily affected by their sentiment, so they will have irrational trade behavior in stock market then influence the stock price. In this paper, we based on behavior finance theories, want to extract investor sentiment from stock forum, and use it to research on stock market.In order to extract investor sentiment from stock market, this paper use WebCrawler to download one year’s post data of one stock in eastmoney.com. Then we use sentiment analysis to classify a post’s emotion tendency automatically. Finally we build three indicate, post volume, bullishness index, disagreement index to reflect stock forum information. Then we use these three indicates and three stock markets variables, trade volume, return, volatility to build simple contemporaneous regression model and two lag vector automation regression model.The result show that, (1) in the same time level, post volume has significant predict ability to trade volume and volatility, and investor bullishness index has significant predict ability to return; (2) In two lag vector automation regression model, there is two-way causality between trade volume and post volume, and the effect between them is positive; Return can cause post volume, and the effect is positive; post volume can cause volatility, and the effect is also positive. Trade volume can cause disagreement index, and the effect is negative.