A Research on High Frequency Foreign Exchange Rate Data

Abstract

Financial markets, particularly foreign exchange (FX) markets, are a rich source of high-frequency data. Because of the fast growth of computer power, collecting financial data has become much easier. This helps researchers understand market behaviour better (via price movements) by analysing high-frequency data.

Foreign exchange rates have been the subject of many studies. For decades, researchers have been performing a lot of studies for both statistical properties and technical analysis in FX market. The aim of this paper is to explore behaviour of Euro-US dollar (EUR-USD) exchange rates. Many previous studies in FX markets show that the empirical distribution of exchange rate returns can be considered non-normal, especially at high frequency. Conducting Kolmogorov-Smirnov goodness of fit tests with application of Khmaladze transformation and an autocorrelation function analysis on the returns of our dataset over seven years, we conclude that weekly returns for EUR-USD exchange rate are independent, stationary, and normally distributed which indicates that EUR-USD prices follow geometric Brownian motion. Examination of fractal behaviour of EUR-USD prices shows that prices follow a random walk process.

Previous empirical studies for technical analysis in FX market can not confirm the profitability and stability in performance of technical analysis. Application of two technical trading rules, Kagi and Renko, in EUR-USD market shows unstable, small returns. Those two rules therefore are not profitable for this market.