News-driven peer co-movement in crypto markets
利用自然语言处理分析金融新闻，识别加密货币间的同伴关系，发现同伴资产在单个币种受冲击时出现反向异常收益，这种错误定价持续数周并可获利，表明投资者对新闻过度反应驱动了联动。
This paper develops a novel methodology to identify peer linkages among cryptocurrencies using natural language processing applied to financial news. We document a distinct pattern of conditional co-movement among peer assets: when a cryptocurrency experiences a large idiosyncratic shock, its peers — identified through news co-mentions — exhibit abnormal returns of the opposite sign. This mis-pricing persists for several weeks and enables profitable trading strategies. Our findings suggest that investor overreaction to news drives these dynamics, highlighting the role of financial media in shaping prices. The proposed methodology extends beyond crypto, offering a generalizable approach to studying peer effects and news-driven pricing distortions.