Mining the relationship between COVID-19 sentiment and market performance

Area of Science:

• Financial markets
• Computational social science
• Epidemiology

Background:

• The COVID-19 pandemic caused unprecedented stock market volatility starting in March 2020.
• Understanding the drivers of market fluctuations is crucial for investors and policymakers.
• Public sentiment, amplified by social media, may influence financial markets.

Purpose of the Study:

• To investigate the relationship between public sentiment regarding COVID-19 and stock market fluctuations.
• To determine if social media sentiment can predict stock market trends across pandemic phases.
• To validate a novel Sentiment(S)-LSTM model for analyzing sentiment-market dynamics.

Main Methods:

• Natural language processing and sentiment analysis of Twitter data using pandemic-related keywords.
• Integration of expert-annotated financial sentiment data.
• Analysis of long-term social media sentiment from pandemic to endemic phases.
• Application of a Sentiment(S)-LSTM model for time-series analysis.

Main Results:

• A significant correlation was found between social media sentiment and stock market volatility.
• Sentiments directly related to stocks showed a particularly strong predictive relationship.
• The Sentiment(S)-LSTM model demonstrated effectiveness in capturing evolving sentiment-market dynamics.
• Distinct sentiment patterns emerged across pandemic, endemic, and new normalcy phases.

Conclusions:

• Public sentiment, particularly on social media, is a significant factor influencing stock market volatility.
• The Sentiment(S)-LSTM model provides valuable insights into the complex interplay between public mood and financial markets.
• Monitoring social media sentiment can offer predictive capabilities for understanding market behavior during health crises and beyond.