This study examines how collective public opinion expressed on social media is connected to stock market behavior, focusing on whether sentiment precedes, follows, or interacts dynamically with price movements. Using a large-scale dataset covering more than 200,000 social media messages and 23 globally significant publicly traded companies over a six-year period, the research moves beyond simple correlations to explore how information actually flows between public sentiment and financial markets. The objective is to understand whether social discourse merely reacts to markets or actively contributes to price formation.

To address this question, the study combines modern natural language processing techniques with an information-theoretic framework capable of capturing nonlinear and time-dependent relationships. Social media content is transformed into sentiment indexes that reflect aggregated public opinion, while stock prices are analyzed through their daily variations. Instead of relying on traditional linear models, the analysis applies Transfer Entropy, a methodology designed to detect directionality and intensity of information flow, allowing the study to distinguish influence from coincidence. This approach provides a more realistic representation of how information propagates in complex financial systems.

The results reveal a consistent and statistically significant flow of information between social sentiment and stock prices across multiple companies and sectors. In many cases, public opinion is shown to lead market movements with measurable delays, while in others, feedback loops between markets and sentiment emerge. These findings challenge the notion that markets instantly and fully absorb information and highlight the strategic relevance of monitoring collective sentiment. The study has direct implications for trading strategies, risk monitoring, and corporate communication policies in an environment where public discourse and financial markets are increasingly intertwined.

Download the full report