FX forecasting using B-WEMA: Variant of Brown's Double Exponential Smoothing

Abstract

A new variant of B-DES (Brown's Double Exponential Smoothing), as a type of classical MA (Moving Averages) method commonly used in time series data forecasting, had been introduced and known as B-WEMA. It has proven to have a better accuracy and robustness level compare to the other moving average methods, such as WMA and B-DES. However, B-WEMA implementation on a real financial time series data such as foreign exchange (FX) had never been done. Therefore, in this research we try to implement B-WEMA as a variant of MA method on FX forecasting and compare the results with other moving average methods using the MSE and MAPE forecast error measurements criteria. Results from the experiments conducted show that B-WEMA has a better accuracy level compared to WMA and B-DES methods.