Sentiment dispersion of individual investors in stock market

This paper aims to study the role of sentiment dispersion in stock market. We extract the investor sentiment from tweets that are specifically about opinions on stocks. Naïve Bayes is then used to assign each tweet a conditional probability representing how positive each tweet is. We did not discretize the probability so as to reduce the information loss. Sentiment dispersion is then measured by standard deviation. The resulting sentiment dispersion is then correlate with future stock returns and realized volatility. This research is able to show whether sentiment dispersion contains information about future return and volatility, which are helpful in formulating investment strategy.