Page 42 - Contributed Paper Session (CPS) - Volume 8

CPS2176 Chiraz KARAMTI et al.
through the adoption of appropriate monetary policies. In this context, and as
part of a preventive approach, the prediction of exchange rates helps
authorities to plan monetary policies to be adopted in the future in order to
maintain and strengthen country competitiveness. However, exchange rate
prediction has long been recognized as a puzzle. Many models have been
developed by theorists and analysis in order to determine the exchange rate.
This paper tries to develop a new approach for exchange rate prevision using
the wavelet method.
Wavelet theory has provided statisticians with powerful new techniques
for nonparametric inference by combining conventional forecasting methods
with insights gained from applied signal analysis. Only recently few research
(Tan et al. (2010); Ismail et al., 2016) used this novel forecasting method based
on wavelet transform approaches combined with ARIMA and GARCH models,
and it was compared with some of the most recently published price
forecasting techniques. The comparative results clearly showed that the
proposed forecasting method was far more accurate than the other
forecasting method.
This article suggests using this novel technique for forecasting the
exchange rate EURO/USD, based on Wavelet transforms and ARIMA/GARCH
model. High frequency return data (5 minutes) from 01/05/2017 until
12/12/2016 with a total sample of 44508 observations is used for this study.
Results show that the predictability of exchange rates varies along the
different frequencies. Besides, there is significant improvement in
predictability as we move from a short forecast horizon to a long forecast
horizon. The omission of these features in forecasting the REER indicators for
the euro may have serious consequences since the REER is used as indicator
for monetary and exchange rate policies, and as policy makers may use it to
forecast current account and trade balance in the country.

2. Methodology
Based on prior evidence of unparallel performance of wavelet technique
and following Tan et al. (2016) and Ismail et al. (2016) methodology, we
implement the wavelet transformation in conjunction with Autoregressive
Moving Average (ARMA) and the Exponential Generalized Autoregressive
Conditional Heteroscedasticity (EGARCH) model to accurately predict
exchange rates.

2.1 Wavelet transform
Wavelet theory is a powerful mathematical tool for time series analysis. It
provides a time-frequency representation of a time series () (in our study,
() is the exchange rate return), and it can be used to analyze non-stationary
time series, which are very common in finance, given the continuous presence

31 | I S I W S C 2 0 1 9