Pairs Trading with Copulas

Abstract

Pairs trading is a well-acknowledged speculative investment strategy that is widely used in the financial markets, and distance method is the most commonly implemented pairs trading strategy by traders and hedge funds. However, this approach, which can be seen as a standard linear correlation analysis, is only able to fully describe the dependency structure between stocks under the assumption of multivariate normal returns. To overcome this limitation, we propose a new pairs trading strategy using copula modeling technique. Copula allows separate estimation of the marginal distributions of stock returns as well as their joint dependency structure. Thus, the proposed new strategy, which is based on the estimated optimal dependency structure and marginal distributions, can identify relative undervalued or overvalued positions with more accuracy and confidence. Hence, it is deemed to generate more trading opportunities and profits. A simple one-pair-one-cycle example is used to illustrate the advantages of the proposed method. Besides, a large sample analysis using the utility industry data is provided as well. The overall empirical results have verified that the proposed strategy can generate higher profits compared with the conventional distance method. We argue that the proposed trading strategy can be considered as a generalization of the conventional pairs trading strategy.