On a class of optimization problems emerging when hedging with short term futures contracts (Q2482688)

The paper of G. Leobacher is located on the interface between continuous optimization theory, some probability theory and stochastic calculus, and financial mathematics. This combination lets the paper be a real and valuable contribution to Operations Research and its modern methods. Indeed, the paper is excellently motivated, organized and written, witnessing OR as an excellent host of deep mathematics. This article generalizes earlier work by G. Larcher and G. Leobacher about hedging with short-term futures contracts, a problem which is considered in connection with the debacle of the Germany company Metallgesellschaft. While the original problem corresponded to the simplest possible model for the price process, i.e., Brownian motion, the author gives solutions to more general models now, i.e., mean reverting model (Ornstein-Uhlenbeck process) and geometric Brownian motion. Furthermore, this article allows for interest rates greater than 0. The regarded stochastic differential equation is very shortly introduced, without a deep excursion into the stochastic calculus of Ito. It represents a price process of the commodity. Then, the discounted payoff from a heding strategy holding certain features at discrete times is regarded, and an assumption made on the intergral form of the difference (``error'') between the cash flow and its expected value. This is a stochastic process, and so is the hedging strategy which is included into the integral term. The problem on how to select the hedging strategy is discussed, e.g., in terms of some vanishing variance, by rolling stock, or by minimizing (with respect to the heding strategy) the supremum of the variance taken over the time iterval. This problem is then transformed into the optimization problem which this paper analyzes. The problem variable is a measurable function; the target functional is the supremum of an integral (the upper bound being the time) with an integrand that is quadratic in the difference between time and unknown. Various nice examples show and motivate this. The author very carefully studies and proves existence, uniqueness and various basic properties of the solution. Among the latter ones there are being piecewise constant, differentiability properties, forms of the derivative, etc. The paper is completed by nice examples, and remarks on open problems. Indeed, this precious paper can serve as a basis for future scientific studies, but also for an understanding of real financial markets and ways how to avoid risk and even debacles in decision making there.