CMStatistics 2021: Start Registration
View Submission - CFE
Title: Not all words are equal: Sentiment and jumps in cryptocurrency market Authors:  Francesco Poli - University of Padova (Italy) [presenting]
Massimiliano Caporin - University of Padova (Italy)
Oguzhan Cepni - Copenhagen Business School (Denmark)
Ahmet Faruk Aysan - Hamad Bin Khalifa University (Qatar)
Abstract: The aim is to decipher the relation between price jumps and investors' sentiment on cryptocurrencies. We use intraday (1-minute) data on a large set of cryptocurrencies prices and Thomson Reuters MarketPysch Indices (TRMIs) which are monitoring sentiment on the same cryptocurrencies by applying complex natural language processing to news and social media. We detect jumps at the intra-day level and correlate their occurrence with TRMI-scored events through logistic regressions. Our results show moderate evidence that the release of information, as monitored by the TRMIs, increases the probability of both positive and negative price jumps, especially within 60 min from the TRMI event. In particular, we find that while sentiment on topics limited to emotions such as optimism, anger, conflict and fear increases the probability of positive jumps occurrence, a more extensive set of topics including both emotional factors and risks perceptions related to volatility, scam, violence and price forecast is identified as the possible causation of negative jumps. In addition, the size of the jumps seems to be less related to the TRMIs compared to the jump occurrence, which has some impact on the occurrence of further TRMIs events. Furthermore, our findings suggest the presence of a significant self-excitation, that is, the occurrence of price jumps (TRMIs events) strongly increases the probability of the occurrence of price jumps (TRMI events).