Detection of volatility regime-switching for crude oil price modeling and forecasting

Abstract For crude oil price modeling and forecasting, time-discrete models like GARCH and HAR-RV have been further developed with Markovian regime-switching in recent years. Questioning on the ubiquity of regime-switching, we establish a time-continuous diffusion model governing the oil prices and detecting whether volatility regime-switching exists in different time horizons over the past decade. Model analysis and comparison with existed methods show that, during the second period we investigated, there exists no obvious volatility regime-switching, hence combining with regime-switching contributes little to its modeling. Besides, shown by this paper, existence of regime-switching and the transition rate of regime depict the long-term structure of volatility, which could be an intrinsic market property.