Abstract
We construct a monthly measure for online social opinions on stocks in the Chinese market by extracting textual data from the internet. By implementing a “social forecasting” strategy, we find a significantly positive alpha of 0.544% per month based on the model of Fama-French's (2015) five-factor plus Carhart's (1997) momentum, and 0.636% per month based on Liu et al.'s (2018) Chinese four-factor model. This anomaly is driven by the mispricing of underexplored information since online social opinions are a strong predictor of firms' earning surprise and are strengthened when there are information frictions, such as low investor attention, high arbitrage costs, high proportions of sentiment-driven trades, and low percentages of institutional investors. Our results are robust to a series of tests and remain unchanged after taking the “factor zoo” into consideration.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
