Foreign exchange market forecasting using evolutionary fuzzy networks

The paper presents an evolutionary fuzzy network method for prediction in foreign exchange markets. The research chooses the financial data used in the Santa Fe time series forecasting competition, 1990-91 (Weigend, 1994). The choice of this data provides a comparative study of the previous attempts to forecast. Fuzzy systems not only provide the mechanism to integrate human linguistic knowledge into logical framework but also provides the means to extract fuzzy rules from an observed data set. For global optimisation, genetic algorithms are used to adapt the parameters of the fuzzy network in order to obtain the best performance.