Twitter Sentiment Analysis and Influence on Stock Performance Using Transfer Entropy and EGARCH Methods.

Area of Science:

• Financial Economics
• Computational Social Science
• Artificial Intelligence

Background:

• Extensive financial economic research indicates that negative news disproportionately affects stock prices compared to positive news.
• The 'asymmetric response effect' is a well-documented phenomenon in financial markets.
• Traditional methods may not fully capture the nuances of real-time sentiment impact on stock prices.

Purpose of the Study:

• To investigate the asymmetric response effect of news sentiment on stock prices using artificial intelligence.
• To quantify the differential impact of negative versus positive news on stock market behavior.
• To validate existing financial theories with novel computational approaches.

Main Methods:

• Web-scraping Twitter data for top tweets related to 24 large-cap companies over a decade using an AI algorithm.
• Analyzing tweet content with a second AI algorithm to generate social sentiment indexes and time series.
• Employing transfer entropy to compare social sentiment indexes with daily stock prices.
• Utilizing the Exponential GJR GARCH (EGARCH) model for further validation.

Main Results:

• Confirmed a statistically significant difference in the impact intensity of negative and positive news on daily stock prices.
• Demonstrated that negative news has a greater impact on stock prices than positive news.
• Validated the asymmetric effect of sentiment, where negative sentiment exerts a stronger influence.

Conclusions:

• The study provides empirical evidence supporting the asymmetric effect of news sentiment on stock prices.
• Artificial intelligence methods effectively capture and quantify real-time market reactions to sentiment.
• Findings reinforce the importance of monitoring social media for understanding market dynamics.