Abstract This paper investigates the dynamics of crude oil markets proxied by the USO ETF. Based on a preliminary analysis of the time series of implied volatility surfaces, we propose a model accounting for multiple stochastic volatilities, co-jumps, and self-exciting jump intensity. We estimate the model on a time series of option prices using a quasi-Bayesian methodology that exploits sequential Monte Carlo methods. We compare the performances of the proposed model with several nested and non-nested specifications. Our results show that both multiple variance factors and self-excitation in the jumps process are needed for accurate model fitting and to improve out-of-sample forecasting.
Brignone et al. (Wed,) studied this question.