Dynamic Cointegration Based Pairs Trading

Mads Hofstedt Jæger

Student thesis: Master thesis

Abstract

This article implements the pairs trading algorithm by Gatev et al. (2006) on recent data on the S&P 500 Index. The finding is that this version of the pairs trading strategy have become unprofitable. Several authors has suggested the use of cointegration to pairs trading. Employing cointegration in pairing and use of the cointegrating regression as basis for the trading rule in the Gatev et al. (2006) algorithm does not alleviate recent negative performance. A suggestion to a new pairs trading strategy is made which essence is to dynamically update the equilibrium model capturing fundamental equilibrium changes. Portfolio selection of pairs is based on trend following, of allocating to pairs with the strongest recent backtest performance. This strategy has had an average annual excess return of 17% but returns are decreasing in recent years. Many of the tested portfolios has shown unprofitable since 2010.

EducationsMSc in Advanced Economics and Finance, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2016
Number of pages81