Is pairs trading (buying the loser and selling the winner of close-substitute stocks that have diverged unusually in price) profitable after accounting for reasonable trading frictions? In the November 2010 version of their paper entitled “Are Pairs Trading Profits Robust to Trading Costs?”, Binh Do and Robert Faff examine the impact of trading friction (commissions, market impact and short selling fees) on pairs trading profitability. Their baseline pairs trading strategy consists of: (1) finding a partner for each stock that minimizes normalized price spread during a 12-month formation period; (2) screening the best pairs based on lowest tracking error; (3) within six months after pairs identification, opening long/short positions in the underpriced/overpriced members of the best pairs with normalized price divergences of at least two standard deviations; and, (4) closing the trade at the first price convergence or, otherwise, at the end of the six-month trading interval. They also consider 29 strategy refinements that address industry affinity, frequency of past price divergence-convergence and/or magnitude of past price divergence. Using prices and industry designations for relatively liquid U.S. stocks over the period July 1962 through December 2009, they find that:
- Estimated one-way trading frictions are 0.81% for the 1963-1988 subperiod and 0.33% for the 1989-2009 subperiod. Estimated cost of shorting is an annualized 1% prorated over the life of each trade.
- The baseline hedge strategy typically results in a portfolio of 20 equally weighted pair trades. This strategy is unprofitable after accounting for trading friction.
- For the 29 refined hedge strategy portfolios:
- Gross monthly returns range from 0.70% to 1.15% (average 0.93%), with very low standard deviations.
- Net monthly returns range from -0.07% to 0.35% (average 0.12%).
- The top four portfolios (all intro-industry) generate an average monthly return of 0.29% (3.5% annualized). Limiting these four strategies to the largest 30% of stocks by market capitalization reduces average monthly return to 0.19%.
- Over the 1989-2009 subperiod, the top strategy refinements are profitable (0.45% monthly net return), but most of the profitability concentrates during the 2000-2002 bear market.
- Limiting pairs to intra-industry matches generally improves profitability, with finer segmentation outperforming coarse segmentation.
- Excluding pairs with extremely tight tracking (such that two standard deviations represents a small absolute price difference) generally improves profitability.
- A short-term contrarian hedge strategy that buys/sells extreme losers/winners sorted within
industries generates higher net monthly returns than pairs trading (0.65%), but with higher portfolio volatility.
In summary, evidence from tests on reasonably liquid U.S. stocks over the past 47 years indicates that pairs trading offers at best low profitability (but also low risk) to investors who can quickly and cheaply exploit temporary relative mispricings.
Cautions regarding findings include:
- As noted in the paper, the authors do not correct for any data snooping bias introduced by testing 30 strategy variations on the same set of data. Results for the best portfolios therefore overstate expectations.
- Trading frictions would be substantially higher for most individual investors than the estimates used in this study.
