The South-East Asia Crisis, Neural Networks and Market Behavior: An Exploratory Study

Abstract

A model of market behavior as a learning process was explored using artificial neural networks. Market and currency data from the period before and during the South-East Asia crisis of 1997 relating to Indonesia, the Philippines, Malaysia, Thailand, Australia, and New Zealand was used to train neural networks. Time series of changes in neural networks' connection weights were generated whilst making a series of forecasts over time. Changes in connection weights captured the changing importance of these Asian markets to those of Australia and New Zealand as the crisis unfolded, and could be regarded as a measure of market learning.