Research on Bitcoin Price Prediction Based on Text Analysis and Deep Learning

With the increasing popularity of Bitcoin, many investors are attracted to it. However, Bitcoin’s decentralized and anonymous transaction features facilitate certain illegal activities, leading to significant price fluctuations and potential risks for investors. Therefore, predicting Bitcoin prices is crucial. This study analyzes factors influencing Bitcoin prices and trends by examining four years of closing prices. Data categories include trading, public, technical, macroeconomic factors, and global currency market indicators. The study involves LDA topic clustering, NLP sentiment analysis, data preprocessing, and the application of LSTM and CNN-LSTM deep learning models for price prediction. Machine learning methods are used to assess factors impacting Bitcoin prices, aiding investors and decision-makers in making more accurate predictions and improving investment decisions. The study suggests that integrating sentiment features enhances prediction accuracy, emphasizing the importance of public sentiment in Bitcoin price forecasting. By incorporating news headlines and public sentiment indicators, this research enriches cryptocurrency price prediction methods, offering new perspectives and theoretical support for digital currency research, potentially helping investors mitigate risks and supporting regulatory efforts in financial markets.