Gustav Montgomerie-Neilson: Selecting the Worst-Case Portfolio: A proposed pre-trade risk validation algorithm of SPAN

The thesis outlines a possible pre-trade risk validation algorithm for portfolios of commodities futures and options. A method is proposed that, given an order book of unmatched orders, determines the particular order selection, or portfolio, with the maximum margin requirement, as calculated by the risk analysis methodology SPAN. The method consists of a selection algorithm, where all orders in the order book are either included or discarded according to a specified criterion. The selection criterion approach reduces the problem from exponential to linear time complexity, complying with pre-trade risk validation requirements. Further, three different selection criteria are proposed and evaluated by accuracy and time performance. Simulations indicate that one of the criteria has considerable accuracy in determining the worst-case portfolio in linear time, without relying on approximations of the orders it includes. This makes it a particularly suitable candidate for pre-trade risk validation.