Pairs Trade
A pairs trade is a strategy where a trader simultaneously buys one asset and sells short a second, closely related asset, betting not on which direction the market moves overall but on how the price relationship between the two assets changes. If the pair's prices normally move together and temporarily diverge, the trader bets the gap will close again - buying the one that fell behind and shorting the one that ran ahead - and profits from the "spread" narrowing rather than from either stock rising or falling in absolute terms.
Because it holds one long and one short position at the same time, a pairs trade is a form of market-neutral strategy: if the whole market drops 5%, both legs tend to drop together and the losses on the long side are largely offset by gains on the short side, leaving the trade's profit or loss dependent mainly on the two assets' relative performance rather than the market's overall direction. This is also why it's classified as a type of statistical arbitrage - it exploits a statistical relationship (historical correlation) rather than any fundamental mispricing.
The strategy is credited to a Morgan Stanley quantitative trading group in the 1980s, pioneered by analyst Gerry Bamberger and later scaled up under the desk run by Nunzio Tartaglia. Their original approach tracked pairs of historically correlated stocks and, whenever the price spread between them widened beyond its normal range, shorted the outperformer and bought the underperformer, expecting the two to converge again. The approach reportedly produced strong returns for the bank in its early years and helped seed an entire branch of quantitative trading that grew into modern statistical arbitrage.
A concrete, simplified example: two large domestic airlines historically trade with a price ratio close to 1:1.2. If Airline A trades at $60 and Airline B, which usually trades about 20% higher, drops to $64 after a company-specific piece of bad news unrelated to the industry as a whole, a pairs trader might short Airline B and buy Airline A, expecting the ratio to snap back toward its historical 1:1.2 relationship - profiting if B underperforms A going forward, regardless of whether airline stocks broadly rise or fall.
Choosing the pair is the hardest part of the strategy. Traders look for two assets with a strong statistical relationship - often measured using correlation or a more rigorous test for cointegration, which checks whether the spread between two prices tends to revert to a stable long-run average rather than simply moving in the same direction by coincidence. Common candidates include two companies in the same industry (two airlines, two banks), two share classes of the same company, or two commodities with linked supply chains, such as gold and silver. Crypto traders have adapted the same logic to correlated coins, such as two large-cap layer-1 tokens that historically move together.
Pairs trading isn't risk-free. The central danger is that the historical relationship breaks down permanently rather than temporarily - a scenario often described as the spread "not reverting." This can happen when one company has a fundamental change (a merger, a regulatory shock, a product failure) that the other doesn't share, meaning the old correlation was never guaranteed to hold going forward. Because both legs use margin (one long, one short), losses can also compound quickly if the divergence widens further before it reverses, which is why position sizing and stop-loss rules on the spread itself are standard practice among pairs traders.
The strategy's academic record is unusually clear, including about its decline. The definitive study, published in 2006, tested a simple mechanical version on American shares from 1962 to 2002 and found annualised excess returns of up to about 11 percent, comfortably above conservative estimates of transaction costs, which is a strong result for a rule that involves no forecasting at all. Follow-up work published in 2010 extended the test and documented what the original had begun to show in its later years: a continuing downward trend in profitability, with the notable exception of periods of sustained turbulence, where the returns reappeared.
Both halves of that finding make sense and both are worth carrying. The decline is what happens to any published rule: the paper described a mechanism, capital followed it, and the spread that the strategy depended on was arbitraged toward the cost of executing it. The survival during turbulence is the more interesting part, because it says what the returns actually were. A trade that pays well precisely when markets are dislocated and pays little when they are calm is being compensated for providing liquidity under stress rather than for spotting a mispricing. That reframes the risk entirely: the strategy's bad periods are not random, they cluster in exactly the conditions that force a leveraged holder to close, which is the same shape of risk that destroyed the convergence trades of 1998.