Title: Investor sentiment and intraday bitcoin returns
Authors: Thomas Renault - Université Paris 1 Panthéon-Sorbonne (France) [presenting]
Dominique Guegan - Universite Paris 1 - Pantheon-Sorbonne (France)
Abstract: The purpose is to use a dataset of several million messages sent on Twitter and on Stocktwits to explore the relation between investor sentiment on social media and intraday Bitcoin returns. Computing returns and investor sentiment at various frequencies, from 1 minute to 24 hours, we do not find any strong evidence that lagged sentiment or lagged returns predict Bitcoin returns on recent periods (2016-2019). This result is robust to the method used to compute investor sentiment (lexicon-based approach and machine learning) and to the inclusion of news from the website CoinDesk.com. The Bitcoin market seems to be more efficient than is commonly understood.