<p>This paper proposes a new agent-based model grounded in the minority‑game framework to reveal the underlying mechanism of excess comovement. We model two key information diffusion behaviors of investors on social media: common attention to different stocks and information interaction about a single stock. The simulation results show that both behaviors significantly influence excess comovement, but their roles differ contextually. For stock pairs with historically positive return correlations, the impact of common attention dominates excess comovement when information interactions are infrequent, and a higher ratio of co-investors amplifies this effect. In contrast, for pairs with historically negative correlations, information interaction becomes the dominant driver of excess comovement when the ratio of co-investors is low, especially during periods of high market herding. Furthermore, the model provides accurate forecasts of excess comovement for both the next day and week.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Modeling excess comovement with information diffusion on social media

  • Zhang-Hangjian Chen,
  • Xiang Gao,
  • Fei Ren,
  • Xiong Xiong,
  • Wei Zhang

摘要

This paper proposes a new agent-based model grounded in the minority‑game framework to reveal the underlying mechanism of excess comovement. We model two key information diffusion behaviors of investors on social media: common attention to different stocks and information interaction about a single stock. The simulation results show that both behaviors significantly influence excess comovement, but their roles differ contextually. For stock pairs with historically positive return correlations, the impact of common attention dominates excess comovement when information interactions are infrequent, and a higher ratio of co-investors amplifies this effect. In contrast, for pairs with historically negative correlations, information interaction becomes the dominant driver of excess comovement when the ratio of co-investors is low, especially during periods of high market herding. Furthermore, the model provides accurate forecasts of excess comovement for both the next day and week.