Pair Trading

Pair trading, also known as statistical pairs trading or spread trading, was pioneered by quantitative traders at Morgan Stanley in the 1980s. The fundamental idea was that two securities sharing common fundamental or operational drivers — such as two competing banks, two refineries, or two IT services companies — tended to move together over time. When their price relationship temporarily diverged from its historical norm, a pair trade involved going long the relatively cheap security and short the relatively expensive one, with the expectation that the relationship would revert to its mean.

The statistical framework underpinning pairs trading required evidence of cointegration — a specific type of long-run relationship between two price series that meant, despite each being individually non-stationary (trending), the spread between them was stationary (mean-reverting). The Engle-Granger or Johansen cointegration tests were standard tools for screening potential pairs. The spread was then modelled as a mean-reverting process, with z-score thresholds (typically ±1.5 to ±2.0 standard deviations from the mean) used to trigger entries and exits.

In Indian equity markets, pairs trading found natural candidates among sector peer groups. HDFC Bank and ICICI Bank, Infosys and TCS, Reliance and ONGC, and Nifty Bank versus Nifty IT were among the pairs that quantitative analysts at domestic hedge funds and proprietary desks screened for cointegration and traded when spreads reached statistically extreme levels. The strategy was appealing for its inherent market neutrality: because one leg was long and the other short, the pair was theoretically insulated from broad market moves, though sectoral events could affect both names in the same direction and temporarily widen spreads beyond expected bounds.

NSE's securities lending and borrowing (SLB) mechanism was a critical enabler of pair trading for Indian participants, providing a regulated avenue to borrow shares for the short leg. However, the Indian SLB market was thin relative to global standards, with limited availability in many mid-cap names, which constrained the universe of tradable pairs for institutional participants. Index futures provided an alternative vehicle for the short side in index-level pair strategies, avoiding the friction of single-stock SLB entirely.

Risk management was a central concern in pairs trading because convergence was not guaranteed within any fixed timeframe. A pair that traded at a two-standard-deviation spread and was entered as a mean-reversion trade could widen to three or four standard deviations before reverting — a phenomenon known as 'spread blowout' — requiring either the capital to sustain the position or a disciplined stop-loss policy. Practitioners typically sized pairs positions as a fraction of maximum historical drawdown of the spread series, and monitored for structural breaks in the pair relationship (mergers, major operational changes, regulatory shifts) that would render the cointegration assumption invalid.