Sentimental showdown: News media vs. social media in stock markets

Abstract

Motivated by the growing convergence between news media and social media as dominant sources of information dissemination, this study examines the connection between textual sentiment and stock returns. Previous studies have examined the effect of sentiment extracted from these two sources on stock returns independently, without modelling how one source can confound the relationship between stock returns and the other source. We investigate this using data from four markets (USA, UK, South Africa and Brazil) and a sample period stretching from January 2016 to April 2023. Employing a suite of methods that encompass both simple parametric techniques and complex models designed to address nonlinearity, chaos and deviations from normality, the analysis uncovers a pronounced impact of social media sentiment on stock returns in the United States. This influence overshadows the effect of news media sentiment across the employed methods. Interestingly, in other markets, news media exhibits a greater effect on stock returns compared to social media sentiment. By emphasising the convergence of news media and social media, the study highlights the important interplay between these sources, offering valuable insights into understanding the complex dynamics of modern financial markets.