LAUSR

This paper describes a novel approach to measure individual investor sentiment using text-based analysis of millions of posts extracted from Chinese financial online forums. We describe how we built a database of more than 200 million stock posts from online financial forums, created GubaLex, a sentiment dictionary consisting of 48,878 words to allow sentiment analysis, and how we developed GubaSenti, an individual investor sentiment index for the stock market in China. This allowed (1) the first systemic measurement of individual investor sentiment in China; (2) an approach to text-based analysis that reflects investor sentiment about millions of posts about stocks listed in Guba; (3) a way to flexibly measure investor sentiment of a single stock, a sector or an industry and the whole market; and (4) made this possible for daily, weekly, monthly, quarterly, and yearly time periods. We also examine the relationship of the sentiment proxy and stock returns and compare it with two typical BW metrics in China. Empirical results show that GubaSenti correlates better with market performance than BW metrics in China and can be used to predict market changes in the short term.