Bitcoin Sentiment Decoded: How Blockchain Activity Reveals Market Mood

Arthur G. Bubolz, Abreu Quevedo, Giancarlo Lucca, Rafael A. Berri, Eduardo Borges, Bruno L. Dalmazo · July 16, 2026

A new study from researchers at multiple institutions introduces a data-driven approach to classify Bitcoin market sentiment by combining on-chain data, financial metrics, and Twitter posts. Unlike typical price prediction models, this work focuses on explaining sentiment—whether the market is bullish or bearish—using blockchain transactions, historical price data, and daily sentiment from social media. The researchers tested several machine learning models, with XGBoost achieving the best performance, an average F1-score of about 0.84. They also used SHAP (SHapley Additive exPlanations) to interpret which on-chain features most influence sentiment, improving model transparency. Key findings show that combining these diverse data sources yields meaningful signals for understanding market emotion. This approach could help investors and analysts gain deeper insights into market dynamics without relying solely on price forecasts. Future work may incorporate deep learning for even better accuracy.