A HYBRID ENSEMBLE FORECASTING MODEL INCORPORATING GLAR AND ANN FOR FOREIGN EXCHANGE RATES

摘要

Exchange rate forecasting is an important and challenging task for both academic researchers and business practitioners. Various theoretical models including both linear and nonlinear approaches have been suggested to model and predict exchange rates. Unfortunately, empirical results often fail to meet theoretical expectations and practical demands. On the basis of these, a hybrid ensemble forecasting model presented in this paper integrating general linear auto-regression (GLAR) with artificial neural networks (ANN) is proposed for obtaining accurate prediction results and ameliorating the forecasting performances. In this study, the performance of the hybrid ensemble model is evaluated by comparing them with two individual forecasting models ― GLAR and ANN, as well as the single hybrid model. Empirical results obtained in this paper reveal that the prediction using the hybrid ensemble model generally performs better than those using the two individual forecasting methods and the single hybrid model in terms of both NMSE and change direction of the exchange rate movement. The paper suggests that the hybrid ensemble model can be used as an alternative forecasting tool for exchange rates to achieve greater forecasting accuracy and improve the prediction quality further.