Foreign exchange rate forecasting by artificial neural networks
Maya Markova · AIP conference proceedings · 2019
Forecasting exchange rates are an important financial problem that is receiving increasing attention especially because of its difficulty and practical applications. Artificial neural networks (ANNs) have been widely used as a promising alternative technique for a forecasting task because of several distinguishing features. Neural networks were originally developed in cognitive science and later were used in engineering for pattern recognition and classification. Neural networks are used because they can model nonlinear behavior in financial markets, in contrast to traditional linear models which are more restrictive. Neural networks can approximate any nonlinear function and are capable of dealing with “noisy” data. In this work, we present an approach of forecasting the exchange rate of the Euro against the US dollar by Nonlinear Autoregressive with Exogenous Input (NARX) Neural Network. It was used the Neural network toolbox of Matlab 2016 software. Different kinds of algorithms and network structures are tested to find the best model for the prediction of foreign currency exchange rates. It is shown how to receive closed price one step ahead.