When opinions collide：Investor sentiment divergence and stock liquidity

We build a micro-level model that combines contagion dynamics with evolutionary game theory. Guided by this framework, we use large language models to analyze more than 80 million Chinese social media posts and examine the effect of investor sentiment divergence on stock liquidity. We find that investor sentiment divergence enhances stock liquidity by stimulating noise trading, attracting market attention, and reducing information efficiency, but at the cost of heightened crash risk. Extending the analysis to a three-dimensional disclosure perspective, we show that faster disclosure amplifies the liquidity effect only during periods of high divergence, higher-quality disclosure consistently mitigates it, while disclosure quantity has little influence. Additional evidence reveals a pronounced “Monday effect” and a weakening impact under tighter short-selling constraints. Overall, our study uncovers the mechanisms through which sentiment divergence shapes liquidity in an emerging market and underscores the implications for financial stability, highlighting how different dimensions of disclosure moderate these effects.