What is Pair Trading? Definition, Formula, and Example
Pair trading is a market-neutral strategy where a trader goes long one security and short a correlated security, betting on convergence of the spread rather than the direction of either name.
Pair trading is a market-neutral strategy where a trader goes long one security and simultaneously short a correlated security, betting on convergence of the price spread rather than the absolute direction of either name. The position carries roughly zero net market exposure (beta ≈ 0), so it profits regardless of whether the broader market rises or falls — provided the historical relationship between the two assets reverts. The strategy was popularized in the 1980s by Morgan Stanley's quant group under Nunzio Tartaglia.
How Pair Trades Are Identified
A robust pair trade requires more than visual correlation:
Step 1: Identify candidates. Same industry, same business model, similar market cap. Classic examples: KO vs PEP, GM vs F, MA vs V, XOM vs CVX.
Step 2: Test for cointegration. Apply the Engle-Granger or Johansen test. Cointegration is stronger than correlation — it requires a linear combination of the two prices to be stationary (mean-reverting). Two assets can be 95% correlated yet drift apart forever; only cointegrated pairs reliably converge.
Step 3: Compute the spread. Spread = Price(A) − β × Price(B), where β is the hedge ratio from an OLS regression. β determines dollar-neutral position sizing.
Step 4: Z-score the spread.
z = (current spread − rolling mean spread) / rolling standard deviation
Entry rule: open the trade when |z| > 2.0. If z is positive, A is rich vs B → short A, long β × shares of B. If z is negative, reverse. Exit when z reverts to roughly 0. Hard stop when |z| > 3.5, which suggests a structural break.
Worked Example
Consider KO and PEP over the trailing 252 trading days. OLS regression yields β = 0.36 (each $1 move in PEP corresponds to a $0.36 move in KO). Trailing spread = KO − 0.36 × PEP has mean of $1.40 and standard deviation of $0.85.
Suppose KO = $68.00 and PEP = $182.00. Current spread = 68 − 0.36 × 182 = $2.48. Z-score = (2.48 − 1.40) / 0.85 = +1.27. Below the +2.0 entry threshold, so no trade yet.
A week later the spread reaches $3.10 (z ≈ +2.0). The trader shorts 100 shares of KO ($6,800 notional) and buys 36 shares of PEP ($6,552 notional). The trade is approximately dollar-neutral. Profit is realized when the spread reverts toward $1.40 — typically within 5 to 20 trading days for liquid, mean-reverting pairs.
When Traders Use Pair Trading
Pair trading is deployed in:
- Choppy, range-bound markets where directional strategies underperform
- Earnings season for relative-value plays between sector peers
- Merger arbitrage as a structured form (long target, short acquirer at the deal ratio)
- Statistical arbitrage at scale — hundreds of pairs run in portfolio form to diversify idiosyncratic risk
Hedge funds like Renaissance Technologies, Two Sigma, and DE Shaw built early track records on portfolios of thousands of overlapping pair trades, smoothing single-pair noise into consistent returns.
Limitations and Common Misconceptions
Pair trades are not riskless. The dominant failure mode is a structural break: one of the two names gets acquired, restates earnings, cuts a dividend, or pivots its business model. The spread then widens indefinitely and never reverts. The 2015 Volkswagen emissions scandal blew up countless VW-vs-BMW pair books.
Pairs are also vulnerable to crowded-trade unwinds. When many quants hold the same trade and risk is reduced simultaneously, the spread moves against the position even with no fundamental news — the August 2007 quant crisis was exactly this dynamic.
Finally, correlation is not cointegration. Two assets can show 0.95 return correlation yet have prices that drift apart at unbounded rates. Test for cointegration before sizing the trade — visual co-movement is not sufficient evidence.