Finance & Financial Econometrics : 
Time: 11.00 am
Date: 12th of October 2023
Room 3001 + Zoom

Daniel VELASQUEZ-GAVIRIA (Maastricht University) “From Past to Future: A Study on noncausal Forecasting for Bitcoin’s Realized Volatility”

Abstract : Bitcoin has experienced significant volatility since its inception, complicating efforts to predict its price. Although its daily returns remain unpredictable, Bitcoin’s variance displays patterns characterized by conditional heteroskedasticity. This study aims to deepen the understanding and forecasting of Bitcoin’s realized volatility by exploring the capabilities of noncausal and mixed causal-noncausal (MAR) autoregressive models. These models account for both historical data and forward-looking expectations, positioning them to effectively capture Bitcoin’s distinctive dynamics influenced by past events and future projections. We examine Bitcoin’s 5-minute realized volatility data spanning from January 2017 to 2022. For in-sample model fitting, we utilize high-order spectral estimation techniques. In contrast, our out-of-sample forecasts are formulated using an approximation of the predictive density, facilitated by the Metropolis-Hastings sampling method. By contrasting various non-Gaussian distributions, our research offers insights into the complexities of Bitcoin’s volatility, enriching both academic discourse and practical financial forecasting in the cryptocurrency arena.

Joint work with Alain Hecq (Maastricht University)

Organizers:

Jean-Michel ZAKOIAN (CREST)

Sponsors:
CREST and ILB