Optimal trading strategy from an agricultural producer perspective:Calibration and regularization

We study the decision problem of storing commodity for an agricultural producer who aims to sell the production in the futures market under the continuous time speculative storage model in the infinite horizon time. To do so, first we consider the log-Ornstein-Uhlenbeck process as the demand process and provide a novel demand function, capable of being an Snell envelope over all stopping times. Second, we propose novel calibration method which combines the quasi maximum likelihood method and Milstein method. Then we apply our method on both simulated and actual data for two families of value functions. Eventually, we use the Likelihood Ratio Test (LRT) for comparison of two models (storage model and no-storage model). Our results show that the proposed storage model is more efficient than the no-storage model.