How do I set up a pairs trading strategy using cointegration?

Question

AcadiFi · Accepted Answer

Pairs trading exploits mean reversion between two related stocks. Modern implementation uses cointegration rather than simple correlation because correlated prices can still drift apart permanently. Cointegrated pairs share a long-run equilibrium. Step one: screen candidate pairs. Within an industry, test every pair using the Engle-Granger two-step procedure. Regress stock A on stock B: A_t = α + β × B_t + ε_t. Test residuals ε_t for stationarity via Augmented Dickey-Fuller (ADF). If ADF rejects unit root at 5%, the pair is cointegrated. Example: Cresthaven Logistics (CHL) and Ironbridge Freight (IBF) both trucking. OLS gives α = 2.1, β = 0.87. ADF statistic -3.85, p = 0.02. Cointegrated. Step two: compute spread. Spread_t = CHL_t - 0.87 × IBF_t - 2.1. Standardize: z_t = (spread_t - μ) / σ. Step three: trading rules. Enter short CHL / long IBF when z > +2 (CHL relatively expensive). Enter long CHL / short IBF when z < -2. Exit at |z| < 0.5. Stop loss at |z| > 4 (structural break likely). Step four: sizing. Beta-weight positions: for every $1M long CHL, short $870k IBF so dollar-neutral at the cointegration vector. Leverage typically 3-5x. Step five: re-estimate cointegration every 60-120 days. Breakdowns happen when one company's fundamentals diverge (merger, lawsuit, management change). Performance: well-designed pairs books deliver 8-12% returns, 6-9% vol, Sharpe 1.0-1.5, with low market beta.

How do I set up a pairs trading strategy using cointegration?

Master Level III with our CFA Course

Related Questions

Practice Questions