Stock Market Analysis and Social Media Index Based on Deep Learning

In this paper, we introduce a hybrid sentiment analysis model, FinBERT-BiLSTM-TextCNN, which integrates deep learning and machine learning techniques to accurately analyze the sentiment strength of comments in financial forums. We conduct a regression analysis of stock prices and apply economic analysis methods to evaluate the impact of social media sentiment on market dynamics. By processing user comments, we construct a comprehensive investor sentiment measurement indicator known as the Social Media Sentiment Index (SMI). This index innovatively combines two key metrics: sentiment strength and user engagement, offering a novel analytical tool for quantifying and characterizing investor sentiment. Our experimental results indicate two main findings: (1) Compared to current sentiment classification approaches, the hybrid FinBERT-BiLSTM-TextCNN sentiment analysis model achieves over a 3% improvement in sentiment classification accuracy on stock market forum datasets; (2) Based on real trading data from the Chinese stock market over two years and an analysis of 20 million comments from the East Money Forum, the proposed Social Media Sentiment Index more accurately reflects changes in individual stock characteristics and tracks market sentiment fluctuations more efficiently, with a correlation to stock price movements improving by more than 4% compared to traditional methods. This study demonstrates the significant role of integrating economic analysis with sentiment analysis in predicting stock market trends, highlighting the value of the Social Media Sentiment Index as a robust indicator of investor behavior and market sentiment.