Statistical Arbitrage Using Cross-Market Pairs Trading

Abstract: Pairs trading is a statistical arbitrage strategy that offers appealing properties for the sophisticated investor. The concept relies on the creation of a mean-reverting spread between two assets, where there is assumed to exist a long-term equilibrium relationship. This paper applies cointegration testing to model such equilibrium relationships between different pairs of 27 European equity indices. A selection algorithm based on the Engle-Granger two-step procedure picks the five most mean-reverting pairs in a formation period that are consequently traded in a trading period. The process is shifted in time in a rolling window fashion to obtain out-of-sample results for the period January 2006 to December 2017. Performance measures after transaction costs are encouraging with annualized excess returns between 3.9\% and 13.3\%, as well as information ratios between 0.52 and 1.29 for different parameter sets. The returns do not load on a conventional systematic risk-factor, but are almost completely beta neutral during the entire sample period.