Forecasting exchange rates procedures: artificial intelligence or statistical techniques?

Abstract

Purpose - The purpose of this study is to demonstrate the power of Artificial Intelligence, specifically ANN, in forecasting exchange rates, and comparing the results to the statistical techniques' results. Design/methodology/approach - The purpose of this research is to compare between these two techniques, thus to find out which one is more accurate. Considering the fact that a slight difference among exchange rates influences business and trading, it is crucial to accurately predict exchange rates. The research will take the Euro Dollar exchange rate as a working example, although the procedure can be applied on other rates. Findings - The results of our research showed that using ANN with the right parameters and variables rather than using a regression model will yield a result with a lower error margin. Research limitations/implications - Some limitations are the scarcity of some information, the shortcomings of the ANN model, and the juvenility of the Euro Zone (Number of used data sets is 140). The result of this research has a great impact on many fields such as worldwide economies, decision makers, international companies, governments, etc... Practical implications - This model can be practically used by decision makers and financial analysts who are interested in knowing the general trend of the exchange rate between any two currencies. Originality/value - The originality of this research is the actual combination of a financial issue and a methodology that can be considered as a software mimic of the human brain. Many points can be considered as new or original in this research, like the uniqueness of the comparison, the uniqueness of the model, the uniqueness of the parameters used in the models, etc...