Twitter's daily happiness sentiment and the predictability of stock returns

Abstract

Using a novel investor sentiment proxy extracted from Twitter, this paper investigates whether investor sentiment as expressed in daily happiness has predictive power for stock returns in 10 international stock markets. To account for complex relationships between sentiment and stock returns, a Granger non-causality test in quantiles is used. Our empirical results indicate that the causal relations vary across different quantiles. We observe that the causal relationship from happiness sentiment to stock returns exist only in high quantiles interval. The causal relationship from stock returns to happiness sentiment exists only in the tail area.