An approximation pricing algorithm in an incomplete market: a differential geometric approach (Q1776017)

The authors provide an algorithm to approximate the density of the minimal distance equivalent martingale measure (EMM) \(Q^\ast\) and the fair price of any contingent claim in an incomplete market. Using Wiener-Itô chaos expansions, they approximate the infinite dimensional space \({\mathcal M}\) by an \(n\)-dimensional manifold \({\mathcal S}\) of EMMs, where \(n\) can be arbitrary large. Such a finite-dimensional approximation is motivated by the need of practical implementation of incomplete market pricing models using computers. An approximate problem is formulated as maximization of the financial entropy over the finite-dimensional space \({\mathcal S}\). A concept of cross entropy is introduced (new for the financial literature) as a pseudo-distance between EMMs. In the main theorem it is shown that there is a natural Riemannian geometric structure on \({\mathcal S}\). An important application of the theorem (approximation pricing algorithm) is provided by the gradient descent method on a Riemannian manifold. Financial interpretation of the geometry of EMMs is given in terms of a quantitative measure of pricing model risk. It is shown that the Riemannian distance between two EMMs is related to the pricing model risk between two pricing systems. A solution to the approximate problem is also shown to be the minimal pricing model risk approximation to \(Q^\ast\).