Investor Sentiment, Social Media and Stock Returns: Wisdom of Crowds or Power of Words?

Abstract

Research in behavioural finance has shown the importance of investor sentiment to explain stock market returns. In light of the changes in the media landscape from traditional print media towards social media platforms, this paper analyses and compares the ability of investor sentiment measures obtained from various media sources to predict stock market returns. Using a large data set of daily articles and reader comments from 2006 to 2020, this study shows investor sentiment extracted from social media is better in predicting stock market returns than investor sentiment obtained from traditional print media sources. Additional analyses show that social media is more suitable to extract investor sentiment mainly because of the richer language and timeliness of online media. In contrast, the larger set of articles published on social media platforms as well as the greater variety of contributors cannot explain the relative advantage of social media.

Keywords: Investor sentiment, text mining, financial markets, traditional media, social media, stock returns.

JEL Classification: G11, G40, D53