image: Fig. 7 visualizes the relationship between the emotional vulnerability of CSI 300 constituent stocks and network structural characteristics. In this study, emotional vulnerability is divided into five ascending levels (Group 0 to Group 4) corresponding to low vulnerability (low curvature) to high vulnerability (high curvature), marked by points of different colors.
Credit: Chen, J., Chen, Z., Zhang, N., & Zheng, Y.
Financial markets are increasingly shaped by investor emotions, yet most sentiment indicators focus only on whether investors feel optimistic or pessimistic. A new study published in Risk Sciences introduces a different perspective: instead of sentiment direction, it measures how fragile or unstable investor emotions are.
Researchers from the Central University of Finance and Economics in Beijing propose a novel concept called emotional vulnerability, which captures fluctuations and instability in market sentiment. The study combines large language models (LLMs) with network structure analysis to quantify this previously unexplored dimension of investor behavior.
The team analyzed millions of stock forum posts from China’s A-share market. Using the Moka Massive Mixed Embedding (M3E) model, an advanced large language model for Chinese text, they converted investor comments into numerical vectors. These vectors were then used to build sentiment networks for individual stocks, where connections reflect similarities in investor opinions.
“The vulnerability of each sentiment network was measured using Ricci curvature, a concept from geometry that has been widely applied in financial system risk analysis,” shares corresponding author Ning Zhang, a professor at the university. “In this context, higher curvature indicates tighter connections and more homogeneous investor sentiment, making markets more fragile when disrupted.”
The results showed that when emotional vulnerability is high, stock prices tend to fall or become unstable, whereas when emotional vulnerability is low, markets exhibit stronger resilience and generate higher excess returns.
“We also found that portfolios consisting of emotionally stable stocks consistently outperform those with high emotional vulnerability. This relationship remains significant even after controlling for traditional asset pricing factors, including the Fama–French five-factor model, as well as for sentiment direction (optimistic or pessimistic),” adds Zhang.
By shifting attention from sentiment direction to sentiment structure, the research expands the toolkit of empirical asset pricing. The authors suggest that emotional vulnerability can improve risk assessment, enhance market trend prediction, and support more informed investment decision-making.
###
Contact the author: Ning Zhang, Central University of Finance and Economics
CUFE Chinese Fintech Research Center Beijing 102206, China, zhang-ning@vip.163.com
The publisher KeAi was established by Elsevier and China Science Publishing & Media Ltd to unfold quality research globally. In 2013, our focus shifted to open access publishing. We now proudly publish more than 200 world-class, open access, English language journals, spanning all scientific disciplines. Many of these are titles we publish in partnership with prestigious societies and academic institutions, such as the National Natural Science Foundation of China (NSFC).
Journal
Risk Sciences
Method of Research
Data/statistical analysis
Subject of Research
Not applicable
Article Title
Construction of the emotional vulnerability factor based on network and large language models
COI Statement
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.