This study demonstrates the positive impact of tweet sentiment features on stock price prediction by integrating sentiment analysis, technical indicators, and Generative Adversarial Networks. We collected stock-related tweets from Kaggle and used the VADER sentiment analysis model to quantify market sentiment into scores that reflect investor emotions. These sentiment scores were combined with historical stock data and technical indicators, creating a comprehensive dataset for our prediction model. Our structured approach encompasses data collection, feature engineering, model training, and ablation testing, highlighting the important role of tweet sentiment features in stock prediction. We expect that the integration of sentiment analysis, technical indicators, and deep learning techniques will improve prediction accuracy and effectively capture market dynamics.
Liu et al. (Tue,) studied this question.