An Artifical Neural Network Based Trade Forecasting System for Captial Markets

摘要

This paper reports on an Artificial Neural Network (ANN) based system that is used to provide trading decisions. The system, a colalborative effort betwee SEDAL and Hongking Bank of Australia (HKBA), was initially tested on historical data and a trading simulator. These results were sufficient to encourage the allocation of a capital fund which was traded over a period of 7 months. The system was used to trade the Aus-tralian/US Dolalr exchange rate and Austrialian 10 Year Bond Futures. This provided valuable "real world" data from which to analyse the system at minimum cost and in the shortest time. The validity of the hostorical simulations was justified as they accurately predicted the actual trading results.