COMPUTING EFFICIENT HEDGING STRATEGIES IN DISCOUNTINUOUS MARKET MODELS

Wolfgang J. Runggaldier
(joint work with Sara Di Emidio)

Abstract

We consider an incomplete, discontinuous market model of a simple but basic form: a geometric Poisson process model with two independent Poisson processes. The problem is that of determining an investment strategy that minimizes the expectation of a convex function of the hedging error for a given claim. The method of Dynamic Programming is suitable solution approach, but leads to various computational problems especially when the intensities of the driving Poisson processes are not fully known and are updated according to a Bayesian procedure. We present a computationally feasible approximation approach and show its convergence. Numerical results are also discussed.