Abstract

Since its inception, bitcoin has occupied a prominent role in the cryptocurrency market. As bitcoin received increased attention, researchers have examined popular sentiment as a predictor of bitcoin price fluctuation. However, the popular sentiment may differ from the sentiment of financial professionals. We ask whether the presence and news sentiment from prestigious business journals would affect the bitcoin return. We collected the daily bitcoin price data from 01/01/2017 to 03/31/2018 from the third-party cryptocurrency exchange website CoinMarketCap. The headlines of news mentioning bitcoin on Wall Street Journal and Financial Times are collected from LexisNexis Database. We calculated the sentiment embedded in the news headlines using a bag-of-words model and a dictionary-based approach. We estimate a regression of financial news sentiment on bitcoin daily return. Our main findings are that positive sentiment contributes significantly negatively to bitcoin return on the same day; negative sentiment day contributes positively, although not significantly. We interpret these findings through the disposition effect that, investors tend to sell high-performed assets and hold underperformed assets. The findings extend our understanding of disposition effect to the cryptocurrency market, and we plan to further examine if there is evidence of the disposition effect.

Download

Share

COinS
 

Disposed of Bitcoin? Using the Disposition Effect to Understand Financial News Sentiment and Bitcoin Returns

Since its inception, bitcoin has occupied a prominent role in the cryptocurrency market. As bitcoin received increased attention, researchers have examined popular sentiment as a predictor of bitcoin price fluctuation. However, the popular sentiment may differ from the sentiment of financial professionals. We ask whether the presence and news sentiment from prestigious business journals would affect the bitcoin return. We collected the daily bitcoin price data from 01/01/2017 to 03/31/2018 from the third-party cryptocurrency exchange website CoinMarketCap. The headlines of news mentioning bitcoin on Wall Street Journal and Financial Times are collected from LexisNexis Database. We calculated the sentiment embedded in the news headlines using a bag-of-words model and a dictionary-based approach. We estimate a regression of financial news sentiment on bitcoin daily return. Our main findings are that positive sentiment contributes significantly negatively to bitcoin return on the same day; negative sentiment day contributes positively, although not significantly. We interpret these findings through the disposition effect that, investors tend to sell high-performed assets and hold underperformed assets. The findings extend our understanding of disposition effect to the cryptocurrency market, and we plan to further examine if there is evidence of the disposition effect.