Application of Seasonal Holt-Winter model for the prediction of exchange rate volatility

Abstract

Prediction of exchange rates volatility is a challenging problem in investigation of financial markets. Various econometrical models are widely used in predicting of foreign exchange rates. Construction of optimal model is complicated problem due to many factors which can influent volatility of foreign exchange rate. Values and impact of factors on the exchange rate are changing over time; therefore investors should be cautious for the fact that there always be a certain level of risk investing in foreign currency. The article investigates the dynamics of hourly exchange rates of the Euro (EUR) against USA dollar (USD). Foreign exchange (forex) rate dynamic is a complex process that can be better understood through investigation of its characteristics. The idea of describing fluctuations by calculating sum of absolute differences (SAD) of a time series is developed. SAD value time series are supposed to be easier predicted since it takes into account of only the magnitude of exchange rate oscillation ignoring its direction. For the analysis and investigation of exchange rates volatility is applied both additive and multiplicative versions of Holt – Winter exponential smoothing techniques. These methods are appropriate for series with a linear trend and seasonal variations. The analysis of EUR/USD exchange rates data in major financial centers showed that they obey intraday periodical variations, so it was used the mentioned model. Two different statistics: mean absolute error (MAE) and root mean squared error (RMSE) were applied for selection of optimal parameters of the model. The modeling results of models were compared. After investigation it was noted that the best prediction results are given by simplified multiplicative model.DOI: http://dx.doi.org/10.5755/j01.ee.26.4.5210