Track: Financial Engineering

Abstract

Long-term forecasting exchange rate is one of the inevitable requirements of the organizations that are involved in international economic activities. The concept that is frequently echoed, especially in global stock markets, is short-term exchange rate forecasting. Therefore, to choose the required factors that determine the exchange rates, it is necessary for the chosen factors to cover a very large extent on the economic and political developments. Forecasting exchange rate is usually realized two general ways: 1-by using the factors that are chosen, to predict that there would be a decrease or increase in the current rate.2-by using regression to start analyzing the previous changes in the price of the currency, and continue the discovered model into future and give the new price. The present paper is the result of a quality upgrading, through the industrial engineering analysis, on choosing the most efficient and comprehensive parameters for the prediction. Here, by engaging regression and neural network a new way of forecasting is obtained. In this study a number of data are used, two third of which were used to train the model and the other one third were used for testing the output resulted from implementing the model was very close to the test portion data.