Moods on the Move: Overnight Sentiment and the Intraday Return Dynamics

Abstract

Employing high-frequency sentiment metrics from texts in social and news media, we find that moods accrued during nontrading hours explain subsequent price behaviour of stocks at the market opening. We find stronger effects from social media compared with the news. We show that highly positive (negative) overnight tonality in social media results, on average, in 4.75% (-4.99%) excess returns annually. Equivalently, for the news, these figures are only 2.5% and -3.05%. The effect is not subsumed by previous day price performance, nor is it driven by fundamental news such as earnings announcements. Our findings, consistent across a number of model and parameter specifications, establish an unexplored link between overnight sentiment and the opening return.

Keywords: investor sentiment, Social Media, Overnight Return, High-frequency data, Thomson Reuters MarketPsych Indices (TRMI)