A0605
Title: Forecasting cryptocurrencies log-returns: A Bayesian approach using social media sentiment indexes
Authors: Federico DAmario - Sapienza University of Rome (Italy) [presenting]
Milos Ciganovic - Sapienza University of Rome (Italy)
Abstract: Academics are increasingly acknowledging the contribution of social media information to make predictions in many areas, particularly in financial markets and economics. We leverage the predictive power of Twitter and Reddit sentiment together with Google Trends indexes to forecast the log-returns of ten cryptocurrencies divided into three tiers according to their market capitalization. We evaluate the performance of three Bayesian VARs specified with hierarchical shrinkage priors using daily data from November 2017 to January 2022. We perform a four-step-ahead forecast and we find a significant improvement in mean directional accuracy compared with some state-of-the-art forecasting models.
