earticle

영어

We examine how investor emotions and Bitcoin price influence each other using high-frequency data and NLP analysis. Emotions regarding Bitcoin are extracted from an online chatting window at one of the largest cryptocurrency exchanges in Korea. To control for global factors, we analyze relative Bitcoin prices, Korean premium, and differences between Korean exchange and other global prices. We aggregate intraday high-frequency data at five-minute intervals from October 8, 2017 to January 23, 2018. The identified emotions predict the return and volatility of Korean premium five minutes ahead. The results are economically significant: simple trading strategies using the relationship between emotions and Bitcoin prices generate profits. Consequently, investor emotions drive Bitcoin prices, suggesting irrational crypto-markets that rational speculators can exploit, but policy makers need to address.