Stock price prediction using sentic API

• Main Author: Phoa, Justyn Zairen
• Other Authors: Erik Cambria
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; NLP; Sentiment analysis; Fama-Macbeth regression
• Online Access: https://hdl.handle.net/10356/175063
• Institution: Nanyang Technological University

This study investigates the potential of sentiment analysis derived from textual data across platforms like Reddit, StockTwits, Benzinga, and Twitter to enhance stock price prediction and develop trading strategies. Leveraging SenticNet for sentiment analysis, we explore the relationship between investor sentiments and stock price movements. While some trading strategies show abnormal excess returns over 8 years, outperforming the market with higher Sharpe and CAGR ratios, Fama-Macbeth regressions reveal a lack of systemic alpha. We acknowledge limitations in using news headlines as sentiment proxies and suggest further research into the interplay between sentiment analysis and established financial factors to refine predictive models and understand stock price movements better.