FLEET: A Framework for evaLuating European options in a parrallEl and disTributed environment

Option pricing is one of the important problem in finance and demands efficient algorithms that produce accurate and fast results. This thesis aims to develop FLEET1: A Framework for evaLuating European options in a parallEl and disTributed environment on a Network of Computers (NoCs). In this thesis, the NoC is an interconnection of a collection of heterogeneous computers and an eight node shared memory machine. We have implemented a multithreaded pricing algorithm on shared memory architecture using Java OpenMP (JOMP [BWKO00, EPC]). FLEET uses the Common Object Request Broker Architecture (CORBA [SGR99, Bol02]) as a client-server model where a client requests for the value of an option (with certain characteristics of the option) over the network to a server. The server computes the option value using the Black-Scholes [BS73] model, a partial differential equation. The server is multithreaded and uses one thread-per-client policy to serve clients over the network. We use the explicit Forward-Time Central-Space (FTCS) finite-difference scheme to solve the Black-Scholes equation on the server side to evaluate the option price. We implemented a database containing the current stock information of the asset of interest, which can be accessed remotely. We compare and analyze the performance results using different scheduling technique on the shared memory machine with eight processors and achieve a speedup of approximately 4 running 16-threads. A key contribution is that the framework integrates the CORBA with the option pricing model.; 1FLEET refers to a number of warships (processors in my case) working together under one command.