Exchange Rates Forecasting Using a Hybrid Fuzzy and Neural Network Model

Abstract

Artificial neural networks (ANNs) are promising approaches for financial time series prediction and have been widely applied to handle finance problems because of its nonlinear structures. However, ANNs have some limitations in evaluating the output nodes as a result of single-point values. This study proposed a hybrid model, called fuzzy BPN, consisting of backpropagation neural network (BPN) and fuzzy membership function for taking advantage of nonlinear features and interval values instead of the shortcoming of single-point estimation. In addition, the experimental processing can demonstrate the feasibility of applying the hybrid model-fuzzy BPN and the empirical results show that fuzzy BPN provides a useful alternative to exchange rate forecasting