Dresden 2003 – scientific programme
Parts | Days | Selection | Search | Downloads | Help
AKSOE: Physik sozio-ökonomischer Systeme
AKSOE 11: Finanzm
ärkte und Risikomanagement IV
AKSOE 11.4: Talk
Thursday, March 27, 2003, 17:30–18:00, BAR/205
Parallelisation and Vectorisation of Simulation Based Option Pricing Methods — •Jürgen Schumacher1, Uwe Jaekel2, and Achim Basermann2 — 1Institut für Informatik III,Uni Bonn — 2NEC Europe Ltd, Sankt Augustin
In the last years, an increasing demand for complex derivative products arose in financial markets. These new products have features which are adapted to the hedging goals of a specific investor or a small group of clients. Many of these products are dependent on a set of risk factors e.g. basket options, spread options or quantos. In order to price these products and hedge the comprised risks, methods have to be applied which can easily be adapted to different option types and which show a competitive convergence speed. Pricing methods based on Monte Carlo simulation in the case of European style options and the simulation-based stochastic mesh method developed by Broadie & Glasserman are the first choice for high dimensional pricing problems since their convergence rates are independent of the dimension of the underlying vector. We have parallelised and vectorised both pricing methods in order to speed up the computation of price estimates such that the time requirements of market participants are met. Our experimental results confirm that high speedup ratios are reached by these methods.