INTEGRATING SOCIAL MEDIA-BASED EXPERT SENTIMENT INTO LSTM MODELS FOR STOCK INDEX PRICE PREDICTION

Due to the dynamic nature of stock prices, price prediction in the stock market has been considered a difficult task and one of the topical concerns of many investors for a long time. Existing studies addressing stock price prediction often lack the distinction of sentiment subjects when introducing sentiment factors, and it is unclear whether sentiment indices of expert groups in social media can help predict stock prices. Therefore, this study calculates the sentiment index of investment expert groups based on expert opinions in social media and text sentiment analysis algorithms, and then constructs a deep neural network-based LSTM stock index price prediction model and uses the model to predict the closing price of the SSE Composite Index. The results reveal that there is a correlation between the social media-based expert sentiment index and SSE stock index prices, and that the method of using expert sentiment information on social media combined with other information can predict stock index prices more accurately, indicating that social media expert sentiment is one of the factors that affect stock index prices. This study contributes to investors' deeper understanding of the stock market, and the collective sentiment of expert groups on social media can be used as an effective variable to provide valuable support for portfolio decisions.