Bitcoin Price Forecasting Leveraging X Data and Sentiment Indicators Via an LSTM-Enhanced Deep Learning Architecture

Abstract

Role of sentiment analysis in forecasting cryptocurrency market trends. The research employs a Long Short-Term Memory (LSTM) deep learning model integrated with sentimentdata to enhance predictive accuracy. Tweets related to Bitcoin were collected during 2022–2023 and analyzed using the VADER tool, which classified them into positive, neutral, and negative categories. Sentiment scores were combined with historical Bitcoin prices to investigate their correlation before developing predictive LSTM models. The findings reveal that social media sentiment significantly influences Bitcoin price fluctuations, particularly during periods of high volatility. Incorporating sentiment in-formation improved the model’s performance, as indicated by lower error metrics (MAE, RMSE, and MSE) compared to price-only models. This demonstrates that market emotions expressed on social media can serve as valuable predictive indicators in cryptocurrency forecasting. The study concludes that integrating sentiment data enhances the accuracy of Bitcoin price prediction and highlights the importance of emotional factors in market dynamics. Future research should extend this approach to other cryptocurrencies, multilingualsentiment contexts, and real-time analysis to further advance forecasting capabilitiesin digital financial markets.