Financial Engineering Applications of Receding Horizon Control

James A. Primbs
Assistant Professor
Management Science and Engineering
Stanford University

Abstract:  In this talk we begin by presenting a couple of core financial engineering problems and show how they may be formulated as control problems.  In general, these control problems are characterized by the significant presence of noise and constraints.  We use these problems as motivation for the development of stochastic receding horizon control methods. In particular, the stochastic nature of the problem  leads to new challenges for receding horizon control that don't arise prominently in a deterministic setting.  We detail some of our newly developed stochastic receding horizon methods that use on-line semi-definite programs and incorporate constraints.  Additionally, we outline important theoretical properties of these schemes.  Finally, we return to the financial engineering problems and demonstrate the effectiveness of the receding horizon control approach.