Scholarlink

The growing integration of Artificial Intelligence (AI) and Natural Language Processing (NLP) in financial markets is reshaping how asset price prediction is approached. In the context of cryptocurrency—where price movements are heavily influenced by investor sentiment—sentiment analysis has emerged as a critical tool for interpreting market dynamics. This study examines the correlation between sentiment polarity extracted using two domain-specific NLP models, FinBERT and FinancialBERT, and the price fluctuations of three major cryptocurrencies: Bitcoin (BTC), Ethereum (ETH), and Ripple (XRP). The research investigates the extent to which sentiment indicators can act as leading signals for price trends by analyzing correlations across multiple time lags: immediate, 12-h, and 24-h intervals. A hybrid sentiment framework was employed, leveraging FinBERT and FinancialBERT to extract sentiment scores from financial news articles aggregated through the MediaStack API. Concurrently, historical cryptocurrency price data was retrieved from the CoinGecko API. Over a 90-day period, a dataset of 1,300 financial news articles was analyzed. The results indicate a measurable delayed correlation between sentiment and price movements. Ethereum demonstrated the strongest sentiment-price correlation, increasing from 0.3819 to 0.3900 after 24 h. Bitcoin followed, with a correlation rising from 0.2899 to 0.2919, while XRP showed the weakest but still positive correlation (from 0.1005 to 0.1205). These findings suggest that sentiment signals do not immediately influence market prices but may serve as short-term predictors within a 24-h window—particularly for Ethereum. This research underscores the potential of AI-powered sentiment analysis as a complementary indicator to traditional technical and fundamental analysis in the cryptocurrency space. It opens the door for further exploration into real-time predictive systems, multilingual sentiment frameworks, and the integration of hybrid AI models to improve forecasting precision in decentralized finance (DeFi) environments.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mapping Market Mood: A Data-Driven Analysis of Sentiment and Cryptocurrency Price Dynamics

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Franco Farrugia,
  • Cedric Deguara

摘要

The growing integration of Artificial Intelligence (AI) and Natural Language Processing (NLP) in financial markets is reshaping how asset price prediction is approached. In the context of cryptocurrency—where price movements are heavily influenced by investor sentiment—sentiment analysis has emerged as a critical tool for interpreting market dynamics. This study examines the correlation between sentiment polarity extracted using two domain-specific NLP models, FinBERT and FinancialBERT, and the price fluctuations of three major cryptocurrencies: Bitcoin (BTC), Ethereum (ETH), and Ripple (XRP). The research investigates the extent to which sentiment indicators can act as leading signals for price trends by analyzing correlations across multiple time lags: immediate, 12-h, and 24-h intervals. A hybrid sentiment framework was employed, leveraging FinBERT and FinancialBERT to extract sentiment scores from financial news articles aggregated through the MediaStack API. Concurrently, historical cryptocurrency price data was retrieved from the CoinGecko API. Over a 90-day period, a dataset of 1,300 financial news articles was analyzed. The results indicate a measurable delayed correlation between sentiment and price movements. Ethereum demonstrated the strongest sentiment-price correlation, increasing from 0.3819 to 0.3900 after 24 h. Bitcoin followed, with a correlation rising from 0.2899 to 0.2919, while XRP showed the weakest but still positive correlation (from 0.1005 to 0.1205). These findings suggest that sentiment signals do not immediately influence market prices but may serve as short-term predictors within a 24-h window—particularly for Ethereum. This research underscores the potential of AI-powered sentiment analysis as a complementary indicator to traditional technical and fundamental analysis in the cryptocurrency space. It opens the door for further exploration into real-time predictive systems, multilingual sentiment frameworks, and the integration of hybrid AI models to improve forecasting precision in decentralized finance (DeFi) environments.