Currency Exchange Rate Forecasting with Neural Networks


Bona P Nasution and Arvin Agah


Ernst & Young LLP, Kansas City, Missouri, U.S.A. Department of Electrical Engineering and Computer Science, The University of Kansas, Lawrence, Kansas, 66045, U.S.A.




This paper presents the prediction of foreign currency exchange rates using artificial neural networks. Since neural networks can generalize from past experience, they represent a significant advancement over traditional trading systems, which require a knowledgeable expert to define trading rules to represent market dynamics. It is practically impossible to expect that one expert can devise trading rules that account for, and accurately reflect, volatile and rapidly changing market conditions. With neural networks, a trader may use the predictive information alone or with other available analytical tools to fit the trading style, risk propensity, and capitalization. Numerous factors affect the foreign exchange market, as they will be described in this paper. The neural network will help minimize these factors by simply giving an estimated exchange rate for a future day (given its previous knowledge gained from extensive training). Because the field of financial forecasting is too large, the scope in this paper is narrowed to the foreign exchange market, specifically the value of the Japanese Yen against the United States Dollar, two of the most important currencies in the foreign exchange market.




Artificial neural networks, prediction, financial markets