Pairs trading and statistical arbitrage
You trade two securities that normally move similarly as soon as they drift apart, hoping the gap will close again. That was well documented for a long time. Since the early 2000s, though, it has brought much less.
In pairs trading you look for two securities that normally move very similarly, such as Coca Cola and Pepsi. If they drift unusually far apart, you buy whichever is currently cheaper and short the more expensive one. You're hoping the gap closes, whether the whole market rises or falls.
The key study comes from the economists Evan Gatev, William Goetzmann and Geert Rouwenhorst. They tested the strategy with daily prices from 1962 to 2002 and found a clear extra return per year for the best pairs, even after carefully estimated trading costs.
What matters is how this developed over time. The authors themselves noticed that it brought less in the last years of their study. Later studies with data up to 2009 found a further clear decline. One reason: more and more money flowed into similar strategies, and the gaps became smaller and shorter as a result.
It's a textbook example of a pattern that recurs with many strategies: an effect is discovered, published and used by more and more professionals, and that makes it weaken itself. What was a reliable source of income in the 1960s is today a fiercely contested field of specialised hedge funds with extremely fast technology.
Gatev, Goetzmann and Rouwenhorst (2006) formed pairs from stocks whose prices had moved most similarly in the past. For the twenty best pairs they found about eleven percent extra return a year over the whole period from 1962 to 2002. To rule out that this was just a chance find in the data, they also tested execution a day later.
Do and Faff (2010, 2012) repeated the method with American stocks up to 2009 and found a further clear decline. Above all, more and more pairs didn't come back together. Zhu (2024), in a new replication with data from the last twenty years, still finds positive but smaller extra returns. Part of them can be explained by the companies' default risk. That points more to payment for risk carried than to a pure market error.
Economically, the strategy counts as statistical arbitrage. You exploit two closely related securities drifting apart, without either of them having to be over or undervalued on its own. Gatev and his colleagues themselves point out that returns fell exactly when a great deal of money flowed into similar hedge funds that wanted to earn independently of the market. That supports the explanation that more competition makes the opportunities themselves smaller.
Data as text
The extra return of the best pairs was 11 percent per year from 1962 to 2002 (Gatev, Goetzmann, Rouwenhorst, 2006) and 6.2 percent per year over the last 20 years (Zhu, 2024).
Summary
- Pairs trading bets that two closely related securities will return to their usual gap.
- The original study found clear extra returns over forty years up to 2002.
- Later studies show a clear decline, probably because of growing competition.
Did you get it?
What is the basic idea of pairs trading?
Trading two closely related securities as soon as they drift unusually far apart, betting that they'll return to their usual gap.
What did Gatev, Goetzmann, and Rouwenhorst find for 1962 to 2002?
For the best pairs, about eleven percent extra return a year, even after carefully estimated trading costs.
What do Do and Faff largely attribute the post-2002 decline in profitability to?
To more and more pairs not coming back together.
Sources and further reading
- Gatev, E., Goetzmann, W. N. and Rouwenhorst, K. G. (2006), Pairs Trading: Performance of a Relative-Value Arbitrage Rule, Review of Financial Studies View source ↗
- Do, B. and Faff, R. (2010), Does Simple Pairs Trading Still Work?, Financial Analysts Journal
Related
- Strategy types overviewStrategy Types
- Mean reversionStrategy Types
- Compound interestStage 1