Research on the relationship between online public opinion and investor sentiment: from the perspectives of correlation and spillover effects

This study investigates the relationship between online public opinion and investor sentiment. Utilizing textual analysis and professional indicators, we construct an online public opinion index and an investor sentiment index. We then delineate their correlation and spillover effects under both “static and dynamic” and “normal and extreme” conditions. Findings reveal a consistently positive correlation and spillover effects across all scenarios. Investor sentiment frequently exhibits stronger outward spillovers in the estimated system. Furthermore, these interactions exhibit heterogeneity during bull and bear markets. Notably, the study identifies a structural breakpoint in these effects during the COVID-19 pandemic. These results provide evidence of the dynamic linkage between public opinion and market sentiment. This study constructs a text-based online public opinion index from stock-forum posts using a fine-tuned BERT model and a professional investor sentiment index via PLS, then examines their correlation and spillover effects under static/dynamic and normal/extreme conditions using TVP-VAR-DY and Copula-CoVaR models. It finds persistent positive correlation and bidirectional spillovers, with investor sentiment generally dominant, stronger linkages in bull markets, and structural breaks during COVID-19. The work matters because it moves beyond one-way public-opinion-to-sentiment analysis and provides a time-varying, tail-sensitive account of sentiment linkages. Its evidence supports differentiated monitoring of online discourse and investor sentiment, helps regulators adjust surveillance intensity across market phases and crises, and encourages researchers and investors to distinguish genuine investor views from broader netizen sentiment, thereby improving sentiment-risk assessment and information governance in digital financial markets.