Computational intelligence based hybrid approach for forecasting currency exchange rate

Abstract

A new and robust hybrid model is presented here for the purpose of forecasting currency exchange rate. Initially forecasts are obtained from three different models: linear-trend model, autoregressive moving average model as well as from artificial neural network. Because of its non-linear features, results obtained from artificial neural network outperform rest of the two linear models. With the goal to further improve the performance of forecasting models, forecasts obtained from three models are merged together so as to form a hybrid model. In order to do so, optimal weights are required which are generated using an optimization model and solved using genetic algorithms. The proposed hybrid model has been tested on real-world data; the results confirm that this approach can be a promising method in forecasting currency exchange rate.