EFFECT OF NEWS SENTIMENT ON STOCK RETURNS: A TEXTUAL ANALYSIS.

Abstract

This paper examines whether sentiment embedded in news stories is able to predict daily stock returns. After controlling for the existence of news, I find that news sentiment could predict stock prices in the short-term. Specifically, positive news predict a contemporaneous price increase, while negative news places downward pressure on prices for up to 7 days. Portfolio returns from a trading strategy that longs positive stocks and shorts negative stocks are positive and statistically significant. This study also finds that this news-based trading strategy shows the greatest returns for smaller stocks; this might carry practical implications for investors as they can focus on the subset of stocks with highest returns. Finally, I find that portfolio returns experience continuation for up to 5 days before reversing. I also find that trading volume increases after the release of negative news. These findings are aligned to sentiment rather than informational theories.