Option pricing, maturity randomization and distributed computing

We price discretely monitored options when the underlying evolves according to different exponential Lévy processes. By geometric randomization of the option maturity, we transform the $n$-steps backward recursion that arises in option pricing into an integral equation. The option price is then obtained solving n independent integral equations by a suitable quadrature method. Since the integral equations are mutually independent, we can exploit the potentiality of a grid computing architecture. The primary performance disadvantage of grids is the slow communication speeds between nodes. However, our algorithm is well-suited for grid computing since the integral equations can be solved in parallel, without the need to communicate intermediate results between processors. Moreover, numerical experiments on a cluster architecture show the good scalability properties of our algorithm.

We price discretely monitored options when the underlying evolves according to different exponential Levy processes. By geometric randomization of the option maturity, we transform the n-steps backward recursion that arises in option pricing into an integral equation. The option price is then obtained solving n independent integral equations by a suitable quadrature method. Since the integral equations are mutually independent, we can exploit the potentiality of a grid computing architecture. The primary performance disadvantage of grids is the slow communication speeds between nodes. However, our algorithm is well-suited for grid computing since the integral equations can be solved in parallel, without the need to communicate intermediate results between processors. Moreover, numerical experiments on a cluster architecture show the good scalability properties of our algorithm. (C) 2010 Elsevier B.V. All rights reserved.