Decoding Bitcoin Market Sentiment from Blockchain Activity and Twitter: XGBoost with SHAP (arXiv 2607.15258)

Forum topic · 小凯 · 2026-07-18

Summary

A new arXiv paper (2607.15258) by Arthur G. Bubolz and colleagues presents a data-driven approach to explain Bitcoin market sentiment rather than predict prices. The study fuses on-chain blockchain transaction metrics, historical Bitcoin price data, and daily Twitter sentiment classifications into a normalized dataset for market analysis. Multiple machine learning models were evaluated with cross-validation, and Gradient Boosting (XGBoost) proved the most reliable sentiment classifier, achieving an average F1 score of approximately 0.84. To improve transparency, the authors applied SHAP (SHapley Additive exPlanations), a game-theoretic explainability method, to quantify the contribution of on-chain features to model predictions. The results show that combining blockchain activity, financial indicators, and social media sentiment yields meaningful predictive signals and insights, supporting data-driven cryptocurrency analysis and future deep learning improvements. The paper bridges on-chain analytics, NLP-based sentiment classification, and explainable ML for cryptocurrency market research.

Overview

A new machine learning paper, "Decoding Market Emotion from Blockchain Activity: A Data-Driven Sentiment Analysis of Bitcoin" (arXiv: 2607.15258), proposes a novel approach to analyze Bitcoin market sentiment by combining on-chain and financial data with social media posts.

Authors: Arthur G. Bubolz, Abreu Quevedo, Giancarlo Lucca, Rafael A. Berri, Eduardo Borges, Bruno L. Dalmazo Published: 2026-07-16

Key Points

Focus on explanation, not price prediction: Unlike models that aim to forecast prices, this work focuses on explaining market sentiment using blockchain transactions, historical Bitcoin price data, and daily Twitter sentiment classifications.
Data fusion: Sentiment trends are merged with on-chain and financial metrics, normalized into a unified dataset for detailed market analysis.
Model results: Multiple machine learning models were tested using cross-validation, with Gradient Boosting (XGBoost) emerging as the most reliable model for classifying sentiment, achieving an average F1 score of ~0.84.
Explainability: SHAP (SHapley Additive exPlanations), a game-theoretic model interpretability method, was used to quantify the contribution of on-chain features to model predictions, improving transparency.
Conclusion: The combination of blockchain activity, financial data, and social sentiment produces meaningful predictive signals and insights, supporting data-driven cryptocurrency analysis and future deep learning improvements.

Original Abstract (excerpt)

> The growing use of Bitcoin as a decentralized digital asset and investment tool has sparked strong interest in understanding its market behavior. This study presents a new approach to analyze Bitcoin market sentiment by combining on-chain and financial data with social media posts. Unlike models that aim to predict prices, this work focuses on explaining market sentiment using blockchain transactions, historical price data of Bitcoin, and daily Twitter sentiment classifications. The method merges sentiment trends with on-chain and financial metrics, normalized into a dataset for detailed market analysis. Multiple machine learning models were tested using cross-validation, with Gradient Boosting (XGBoost) emerging as the most reliable model for classifying sentiment, achieving an average F1...

Paper: arxiv.org/abs/2607.15258