Exploring Nonlinear Associations Between AI Comment Volume and Retail Investor Sentiment

Yuze Tang

No ms97w_v1, SocArXiv from Center for Open Science

Abstract: Artificial intelligence comment generators have proliferated in online stock forums. However, how AI comment volume associates with retail investor sentiment, and whether this association is conditional on market volatility and annual AI comment density, remain under-explored. Using weekly data from the East Money Stock Forum and the Shanghai Composite Index spanning May 2019–October 2024 in China, this paper proposed a two-detector ensemble for AI comment detection, then investigated nonlinear associations between AI comment volume and retail investor sentiment under three tiers of market volatility and six annual AI comment densities. Our generalized additive model (GAM) results show that AI comment volume exhibits a weak, net-positive nonlinear association with mean retail investor sentiment, and a very weak net-negative nonlinear association with sentiment divergence. We find no statistically significant pairwise differences in these associations across low-, medium- and high-volatility levels. For annual AI comment density, statistically significant cross-density differences are observed for sentiment standard deviation in multiple year pairwise comparisons, whereas only one comparison yields significant differences for mean sentiment. Collectively, our findings indicate that merely considering the number of AI comments provides limited information regarding the sentiment of retail investors.

Date: 2026-10-12
References: Add references at CitEc
Citations:

Downloads: (external link)
https://osf.io/download/6acbf28d3bae2cc8eac5b0a0/

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:osf:socarx:ms97w_v1

DOI: 10.31235/osf.io/ms97w_v1

Access Statistics for this paper

More papers in SocArXiv from Center for Open Science
Bibliographic data for series maintained by OSF ().