Investor Sentiment and Stock Market Investment Amid Public Health Crises: A Study Based on Double DQN

In recent years, public health crises have impeded economic development and exerted significant shocks on capital markets, particularly affecting investor confidence. Although numerous scholars have examined economic stability during public health crises from various perspectives, few have investigated the stability and recovery of capital markets from the standpoint of investor sentiment. In light of this gap, this study employs the Double Deep Q-Network (Double DQN) model within a multifactor pricing framework to explore how investor sentiment influences stock return predictions and portfolio optimization during public health crises. Using data from China’s A-share market during the COVID-19 pandemic, we construct and incorporate several sentiment indices as key indicators of investor sentiment, including the Baidu Sentiment Index (BD), Douyin Sentiment Index (DY), Toutiao Sentiment Index (TT), Stock Market Investor Sentiment Index (CICS), and the Investor Confidence Index (ICI). The experimental results reveal that incorporating investor sentiment indices significantly enhances the predictive performance of the Double DQN model for stock returns and effectively optimizes the Sharpe ratio of investment portfolios. Among these sentiment indices, the BD index exhibits the highest importance, whereas the ICI index shows the lowest. Moreover, the sentiment indices demonstrate a more pronounced effect in optimizing long-short portfolios compared to long-only portfolios, suggesting that market sentiment plays a crucial role in amplifying irrational market fluctuations during public health crises. These findings underscore the need for governments, investment institutions, and individual investors to recognize the impact of investor sentiment on market volatility to prevent domino effects that could escalate into systemic financial risks. This study provides both theoretical insights and practical implications for investment return forecasting and risk management under such conditions.