Neural Network Pricing of All Ordin Aries SPI Options on Futures

摘要

Superior pricing accuracy over conventional option pricing model is demonstreated using an artifical enural network. This occurs in an important subset of the input parameter space. A neural network is applied to pricing American-style call otpions on Australian All Ordinaries share Price Index (SPI) Furtures, using data objtained from the Sydney Futures Exchange (SFE). The Neural Network Pricing is comapred, not only with the Blac-Scholes model, but also with the Barone-Adesi/Whaley model~3, which is a more appropriate model for American-style options on futures.