Pairs Trading Research: Methods, Evidence, and Strategy Limits
Summary
This review surveys research on selecting and trading equity pairs, comparing distance-based matching, cointegration, correlation, and other selection criteria. A common design forms candidate pairs over one period and trades them during a subsequent, shorter period. The approaches also differ in spread modeling, hedge-ratio estimation, entry thresholds, and use of high-frequency data.
The cited evidence is mixed and often weakens in more recent samples. The review reports that Gatev-style strategies showed excess returns earlier in their sample but decayed, while later studies found naive pairs trading unprofitable in examined periods. Results also depend on transaction costs and trade timing; one high-frequency study reports strong results with relatively wide thresholds. Other findings suggest that convergence delays, stock-specific news, shared institutional ownership, and common analyst coverage can affect outcomes. Several theoretical treatments lack broad empirical validation, and at least one proposed minimum-profit method lost money when tested on wider universes. The article presents this as a research survey, not a single validated strategy.
Key ideas
- Many pairs strategies select securities during a formation period and trade them in a later period.
- Distance matching, cointegration, and correlation are among the reviewed pair-selection methods.
- The reviewed studies report decaying or negative returns in some later samples, with results varying by method and market.
- Transaction costs and execution timing can materially change high-frequency pairs results.
- Slow convergence and security-specific news may help explain why some apparent divergences do not revert.
Tags
Cited by
- Strategies Gold Miners vs Gold Beta-Hedged Residual Reversion: fade a >=2-sigma 10-session GDX-vs-GLD idiosyncratic move, dollar/beta-neutral long-short, 15-session time stop (USEQ 1-DAY)
- Hypotheses Gold Miners vs Gold Beta-Hedged Residual Reversion: fade a >=2-sigma 10-session GDX-vs-GLD idiosyncratic move, dollar/beta-neutral long-short, 15-session time stop (USEQ 1-DAY)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.