The Heterogeneous Impact of Social Media Sentiment on Cryptocurrency Valuation

Abstract

In this study, we investigate how Twitter sentiment is associated with cryptocurrency prices, emphasizing the differences observed across various news categories and coin types. First, we develop a machine-learning framework that (1) categorizes tweets into market/business, technology, or policy news and (2) assigns a sentiment category to each tweet. Next, we analyze how the sentiment of tweets, moderated by their news categories, predicts the prices of various cryptocurrencies. Our analysis of a large-scale Twitter dataset reveals that traditional cryptocurrencies (e.g. Bitcoin and Ethereum) present strong positive correlations with market- and technology-related sentiments, whereas meme coins are less predictable by such news signals. Notably, major events—such as the 2022 Terra–Luna crash—appear to weaken these sentiment effects, suggesting that risk aversion and market disruptions can overshadow social media signals. By highlighting the role of news types in shaping sentiment-driven price movements, this research highlights the diverse ways in which digital assets respond to online discourse. The findings offer actionable insights for investors and researchers, supporting the use of targeted sentiment analytics to navigate the evolving cryptocurrency market and make more informed investment decisions.