Abstract
ABSTRACT Purpose: The objective of this article is to model a minute series of exchange rates for the EUR/USD pair using the singular spectrum analysis (SSA) and ARIMA-GARCH methods and evaluate which one offers better forecasts for a five-minute horizon. Originality/value: Despite being a successful technique in other branches of science, the application of SSA in finance is quite new. Furthermore, exchange rate modeling is a complex problem, comprising statistical concepts and properties. However, despite the complexity, the analysis of this series is extremely important for several agents playing, directly or indirectly, a role in the economy and the financial market. Design/methodology/approach: Time series models were estimated using the ARIMA-GARCH and SSA techniques, taking into account three samples of the ask exchange rate (closing): uptrend, downtrend, and no well-defined trend. Findings: The forecasts carried out by the SSA were the ones closest to the original observations for the three cases. Regarding the quality measurements, SSA obtained the best results for both uptrend and downtrend samples; for the sample with no well-defined trend, the findings indicated that the ARIMA-GARCH technique attained better results. However, it was concluded that the SSA forecasts, regarding exchange rates during the studied period, are more appropriate than the ones obtained by the ARIMA-GARCH model, regardless of the market movement.
Highlights
In the last two decades, it has been observed that with the intense increase in globalization, international trade flow has reached levels previously unimaginable, wherein transactions involving different currency pairs have become fundamental for the various economy agents
According to the results shown in the figures, the forecasts made by the singular spectrum analysis (SSA) were the ones closest to the original observations for the three cases in which the technique was applied and for all quality measures, except in sample 3, in which root mean square error (RMSE) and Theil’s inequality coefficient (TIC) were better for ARIMA-GARCH
This fact allows concluding that the SSA forecasts, with respect to the exchange rates in this period, are more adequate than those obtained by the ARIMA-GARCH model independently of the market movement, that is, for both uptrend and downtrend when there was no well-defined trend
Summary
In the last two decades, it has been observed that with the intense increase in globalization, international trade flow has reached levels previously unimaginable, wherein transactions involving different currency pairs have become fundamental for the various economy agents. Many factors have influenced this growth, such as international cooperation treaties and new statistical approaches made possible by advances in information technology for measuring transactions in terms of added value. In this context, Atkočiu- nas, Mačiulis, Klimavičiene., and Kalendiene. (2010) believe that exchange rates play a central role, since they allow an easy comparison of prices of goods and services produced in different countries and because governments, companies, and individuals participating in global scale commerce are exposed to the risk of exchange fluctuations. The forecast of exchange rates is a relevant object of research
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
