Forecasting Stochastic Volatility Characteristics for the Finan-cial Fossil Oil Market Densities

Abstract

This paper builds and implements multifactor stochastic volatility models for the international oil/energy markets (Brent oil and WTI oil) for the period 2011-2021. The main objective is step ahead volatility predictions for the front month contracts followed by an implication discussion for the market(-differences), the data dependence and therefore predictability important for market participants. The paper estimates multifactor stochastic volatility models for the contracts giving access to the reported posterior chain. The model vector realization establishes a functional form of the conditional distribution, which is evaluated on observed data convenient for step ahead volatility predictions. Applying the nonlinear Kalman filter technique, the calibrated condition distribution is evaluated on the observed data series giving projected values for the volatility factors at the data points. For both contracts one factor is slow moving persistent factor while one factor is fast moving mean reverting factor. The negative mean and volatility correlation suggest higher volatilities from negative price movements, suggesting holding volatility as an asset class on its own may insure market participants against market crashes and provide them with an excellent diversification instrument. Moreover, for especially the WTI oil but also Brent oil contracts, the data dependence BDS measure for volatility is strong suggesting predictability. Hence, the multifactor SV models visualize the latent volatility, and their predictions extend the available market information especially interesting for derivative trading (including swaps).