A Cointegration Approach for Selection of Currency Pairs

Abstract

Pairs trading is a statistical arbitrage strategy based on the construction of mean reversion in prices of securities. While these strategies tend to perform well in equities, their effectiveness and performance in the currency market is yet to be tested, which is generally inefficient and predictable. The purpose of this study was to select the pairs for pairs trading in the forex market of select currencies: EUR/USD, GBP/USD, USD/CAD, USD/INR, USD/JPY, and USD/NZD during the period starting from October 1, 2010 – October 20, 2020. The research was organized into three parts to determine the possible pairs of six currencies over different time periods. First, the closeness of potential pairs of six currencies was established using the distance approach for 10 years, 5 years, and 2 years. After that, the Engle – Granger two-step test for cointegration was applied to examine the validity of the top 10 closest pairs of currencies for pairs trading in the study. Based on the empirical results of the cointegration approach, we found very few good pairs in the forex market in our study. USD_INR/USD_NZD was found to be statistically significant at a 10% level of significance over the 10-year sample period, and the same pair of currencies was also found to be statistically significant over a 5-year sample period, but this pair was not found to be statistically significant in the 2-year sample period; whereas, USD_JPY/USD_NZD was found to be statistically significant in the 2 - year period at a 10% level of significance.