Designing Perceptron Three-layered Neural Network For Predicting Dollar- Franc Currency Pair In International Exchange Market

Abstract

International financial markets like exchange market are among the most dynamic markets in the capital markets available in which investing needs enough knowledge and experience. So we decided to invent a system so that it can forecast trend created in exchange market. There are a lot of wide transactional strategies for buy and selling of exchange symbols which here we used science of artificial neural network in this article. Final results of the predictions and percent of Mean square error of the network is calculated after defining neural network and training by Matlab software. Training of desired neural network is achieved by 2010 price data in order to mean square error decrease to 2.15e-6 and at comparison against newest of research show proper improvement.