can computers beat the market?: testing pairs trading using a simulated trading strategy

Abstract: The purpose of this study is to increase the understanding of quantitative trading and to contribute to the general knowledge relating to market efficiency. In current paper the authors develop and simulate a popular trading methodology known as pairs trading. They simulate trading of over 30000 possible pairs during 10 years; specifically testing whether the excess returns previously achieved applying a similar trading strategy on US shares by Gatev, Goetzmann and Rouwenhorst (1998), are possible to achieve on Swedish stocks. In the study Ibbotson and Nygren do not achieve positive returns using a basic trading simulation, predominantly due to large losses incurred during the IT bubble of 1999. However, even excluding 1999 leaves the total return only just positive and not significantly so. The authors believe that the most interesting area for further research on pairs trading relates to the choice of pairs, which they found to be more important than trading strategy.