The Effect of Social Media Sentiments on Daily Stock Returns

Abstract

In today’s economic world, the influences of sentiment are increasing. Since investor’s have adopted social media as source to spread information, thoughts and feelings, the general level of sentiment becomes measurable. This study investigates whether sentiment contains pricing power towards daily stock returns. A sample of Tweets, specifically related to Apple, Amazon, Google, Microsoft and Tesla was used to conduct and compare five different methods of sentiment analysis. It is found that social media sentiment had a significant impact on the same day’s stock returns. Accordingly, the inclusion of sentiment variables in a traditional CAPM model significantly increased its explanatory power to determine stock returns. Since there is no consensus on how to appropriately measure sentiment on social media, this study compared five lexicon-based approaches in terms of performance. It is found that field-specific lexicons outperform general lexicons. This finding contributes to the process of creating consensus, and improvement of transparency and replicability in the field of sentiment analysis.