Pairs Trade

Pairs trading is a market-neutral strategy that profits from temporary mispricings between two historically correlated securities. The core idea: when two stocks that normally move together diverge, you buy the underperformer and short the outperformer, betting the spread will converge back to its historical mean. The spread is calculated as the price ratio or price difference between the two securities. Entry signals typically trigger when the spread exceeds a threshold of two standard deviations from its rolling mean. This educational content is not financial advice — all trading strategies carry risk.

Not all correlated stocks are suitable for pairs trading. Correlation measures short-term co-movement; cointegration measures a long-run equilibrium relationship. The Engle-Granger test checks whether the residual of a regression between two price series is stationary (mean-reverting). If it is, the pair is cointegrated and suitable for pairs trading. Classic cointegrated pairs include stocks in the same industry with similar business models — for example, major US banks, large-cap oil producers, or ETFs tracking the same index with different constructions. Sector ETF pairs such as XLF/KBE (financials) are popular because the mean-reversion relationship tends to be structurally stable.

The standard pairs trade entry triggers when the z-score of the spread exceeds +2 or -2 standard deviations. A z-score above +2 means Pair A is expensive relative to Pair B: short Pair A, long Pair B. The position is closed when the z-score reverts to zero (mean reversion complete) or a fixed profit target is hit. Stop-losses are typically placed at a z-score of +/-3 to 3.5, where the spread has diverged far enough to suggest the historical relationship may be breaking down. Pair trades are sized so that the dollar value of the long and short legs are equal, creating a dollar-neutral position that eliminates direct market exposure.

The primary risk in pairs trading is relationship breakdown — when two securities that were cointegrated stop behaving as a pair. This can happen due to mergers, regulatory changes, management shifts, or fundamental business model changes. For example, two airline stocks may diverge permanently if one files for bankruptcy. A z-score that keeps expanding instead of reverting is the warning signal. Pairs traders must monitor cointegration statistics (Augmented Dickey-Fuller test) on a rolling basis and retire pairs that show structural breaks. Liquidity is another concern: shorting the outperformer requires locating borrow, and borrow costs reduce profitability on mean-reversion timeframes.

Pairs trading is the simplest form of statistical arbitrage. More sophisticated variants include multi-factor relative value (comparing hundreds of securities on fundamental and technical factors), basket trading (long/short portfolios of multiple names), and ETF arbitrage (trading the discount or premium between an ETF and its underlying basket). Pure statistical arbitrage at quantitative hedge funds operates at microsecond latency with hundreds of pairs simultaneously. Retail traders and platforms like Tradewink focus on a curated universe of structurally-sound pairs with adequate liquidity, holding positions from hours to days rather than milliseconds.