Three-factor commodity forward curve model and its joint P and Q dynamics

In this paper, we propose a new framework for modeling commodity forward curves. The proposed model describes the dynamics of fundamental driving factors simultaneously under physical ( P ) and risk-neutral ( Q ) probability measures. Our model is an extension of the forward curve model by Borovkova and Geman (2007), into several directions. It is a three-factor model, incorporating the synthetic spot price, based on liquidly traded futures, stochastic level of mean reversion and an analog of the stochastic convenience yield. We develop an innovative calibration mechanism based on the Kalman filtering technique and apply it to a large set of Brent oil futures. Additionally, we investigate properties of the time-dependent market price of risk in oil markets. We apply the proposed modeling framework to derivatives pricing, risk management and counterparty credit risk. Finally, we outline a way of adjusting the proposed model to account for negative oil futures prices observed recently due to coronavirus pandemic.