Home / Futures & Directional Strategies / Pair Trading
Neutral Advanced Undefined risk Margin-based

Pair Trading

Long one, short a related other — profit from the spread reverting, not from either direction.

Quick answer: Pair Trading is a futures approach that goes long one instrument and short a correlated other, aiming to profit from the spread between them reverting to normal — a relative-value bet whose central risk is that the relationship breaks when most needed.

In simple words

Pair trading means buying one instrument and short-selling a related one at the same time, so you are not betting on the market going up or down, but on the gap between the two returning to normal. If two things usually move together and one gets unusually expensive relative to the other, you short the expensive one and buy the cheap one, expecting the gap to close. The money comes from the spread, not the direction. The great danger is that the relationship you rely on can break exactly during a crisis — the moment you most need the two to move together — and then the hedge stops hedging.

Not to be confused with: Pair trading rests on cointegration, not correlation, and the two are routinely confused. Correlation measures whether two series move together day to day; cointegration means their difference stays bounded and reverts to a stable level even as each series wanders. A pair can be highly correlated yet not cointegrated — co-moving now but with a spread that can drift apart permanently — which is exactly the pair that ruins a trader who selected it on correlation alone.

What it looks like

Pair Trading — the market condition it assumes

A spread between two correlated instrumentsThe relative value of two legs mean-reverts even when both drift. Illustrative synthetic series — not market data.
Market outlook
Neutral
Risk
Undefined risk
Difficulty
Advanced
Undefined risk. The maximum loss on this position is not capped by its own structure. The only thing bounding the loss is the underlying reaching zero, which caps it at a large finite figure — a cap so far away it offers no practical protection. Losses can exceed the premium collected by a large multiple and can exceed the margin posted. This page explains the mechanics so the risk is understood; it is not a suggestion to hold the position.

Professional explanation

The bet is on the spread, not on direction

Pair trading is a relative-value strategy: long one instrument, short a correlated other, sized so the position is roughly neutral to the overall market. What the trader is betting on is the spread — the difference between the two prices — returning to its normal level after it has widened or narrowed abnormally. If both instruments rise or both fall together, the long and short offset and the market direction washes out; the profit or loss comes only from how the spread moves. This is why pair trading is described as market-neutral, and why it can make money in a falling market as easily as a rising one, provided the spread behaves.

Cointegration, and how it differs from correlation

The property pair trading really needs is cointegration, not mere correlation. Cointegration means two price series can each wander far on their own, yet their difference stays bounded and tends to revert to a stable level — the spread has a mean it returns to. Correlation only measures whether the two move together day to day, and two series can be highly correlated for a while without their spread being stable, or can drift apart permanently despite short-term co-movement. Pair trading fades deviations in a spread on the assumption it is cointegrated; if the pair is merely correlated and not cointegrated, the spread can widen without ever coming back.

The real risk: correlation breaks when you need it most

The central danger of pair trading is that the relationship holds in calm markets and breaks in stressed ones — precisely when the trader is relying on it. During a crisis, instruments that normally move together can decouple violently: one is sold for liquidity while the other is not, a sector-specific shock hits one leg, or a corporate event severs the link. At that moment the spread blows out instead of reverting, and the short leg and long leg move against the position together. The hedge that made the trade feel market-neutral stops hedging exactly when the neutrality was supposed to matter.

Executing pairs with Indian stock or index futures

In India, pair trading is typically done with stock futures — two correlated stocks in the same sector — or between index futures. Each leg carries its own SPAN plus exposure margin, and because the two futures are not a recognised spread, the margin benefit may be limited, so a pair ties up margin on both legs. Sizing must equalise the rupee exposure of the legs, which is fiddly when lot sizes and price levels differ, and both legs must have enough liquidity to enter and exit without slippage. A pair is only as tradeable as its less liquid leg, and stock-futures liquidity in India thins quickly outside the most active names.

Construction

  1. Identify two instruments whose spread has historically been stable — ideally cointegrated, not merely correlated — accepting that past stability does not guarantee future stability.
  2. Wait for the spread to deviate abnormally from its normal level, one leg rich relative to the other.
  3. Go long the relatively cheap leg and short the relatively rich leg, sizing the two so their rupee exposures match and the pair is roughly market-neutral.
  4. Define the spread level at which the relationship is judged broken rather than merely stretched, recognising a break can gap and the legs can decouple fast.
  5. Exit as the spread reverts to normal, since the edge is the spread's return, not the direction of either instrument.

Market outlook

A trader may study pair trading when two instruments that normally track each other have diverged abnormally and the trader believes the divergence is temporary — a spread stretched by a transient flow or event rather than a permanent change. The approach assumes the spread is cointegrated and reverts. It is invalidated when the relationship itself changes — a structural break, a corporate event, or a crisis that decouples the legs — after which the spread widens without returning. Because a temporary divergence and a permanent break look alike while they happen, the condition that would confirm reversion is only visible after the spread has, or has not, come back.

Risk profile

Pair trading carries undefined risk on each leg. Neither futures leg has a long option cap; a short leg's loss is unbounded and a long leg's reaches zero. The market-neutral construction reduces exposure to overall direction but not to the spread, and the defining risk is that the correlation breaks in stress — the legs decouple, the spread blows out, and both legs move against the position at once. A gap in either leg can carry the spread past the level where the trader intended to exit. The apparent safety of neutrality is exactly what tempts larger size into a relationship that fails when it is most needed.

Reward profile

The reward is the spread reverting to its normal level: as the rich leg cheapens and the cheap leg richens back toward their usual relationship, the short and long legs both work and the spread narrows in the trader's favour. Because the bet is on the spread rather than direction, the reward is available in rising, falling or flat markets, provided the relationship holds — which is the appeal of a market-neutral return. The gain per trade is bounded by how far the spread was stretched, so it is typically modest, and it depends entirely on the relationship reverting rather than breaking.

Margin requirement

Each futures leg is held against full naked SPAN plus exposure margin, marked to market daily. Because the two legs are separate futures rather than a recognised spread, the margin benefit may be limited, so a pair ties up margin on both legs at once. If the relationship breaks and both legs move adversely, the combined mark can strain margin sharply and risk a square-off of one or both legs at the worst moment. Sizing to match rupee exposure across different lot sizes and price levels is fiddly, and margin percentages and lot sizes are revised periodically.

Greeks exposure

Δn/a

Delta is exactly 1 per unit on the long leg and −1 on the short, and constant. Sized to match rupee exposure, the pair's net delta is near zero — that is what makes it roughly market-neutral — but each leg individually carries clean linear exposure, and the neutrality holds only as long as the two legs keep moving together.

Γn/a

Gamma is zero on both legs. The futures deltas do not change as prices move, so the pair carries no gamma — its neutrality does not curve or accelerate, and a decoupling is felt linearly through the spread rather than being amplified or cushioned by any second-order term.

Θn/a

Theta is zero. A pair does not decay with time, so holding it across the weeks a spread may take to revert costs nothing per day. Expressing relative value through options would introduce theta on each leg; the futures pair simply waits for the spread, indifferent to the clock, paying only carry and roll.

Vn/a

Vega is zero. A pair trader in futures is indifferent to the options market's expected-move pricing on either leg; implied volatility does not move a futures price. The bet on the spread reverting is expressed purely in the realised prices of the two legs, not in any option premium.

ρn/a

Rho is negligible. Rates enter only through each leg's basis, and to the extent both legs share similar carry, rate effects largely offset in the spread, so a pair trader can effectively disregard rho.

Volatility impact

Implied volatility does not apply to the futures legs — there is no vega — so the options market's expected-move pricing does not move either futures price, and the 'IV regime' field is not meaningful here. What matters is realised volatility, and specifically the volatility of the spread rather than of either leg alone. A pair can be built from two volatile instruments yet have a quiet, stable spread, which is the ideal. The danger is that in a stress event the spread's own realised volatility spikes as the legs decouple, so the volatility that matters is the spread's, and it tends to be lowest just before it matters most.

Time decay

There is no time decay. The futures legs carry no theta, so holding a pair across the weeks a spread may take to revert costs nothing per day. This suits pair trading, which often needs patience for a stretched spread to return to normal and would be penalised by an option's daily theta bleed on each leg. The only holding costs are the roll spreads on both legs if the pair is carried across expiry, and the financing embedded in each future's basis — real but modest compared with an option's time decay.

Practical examples

NIFTY example

A common Indian pair is long one index future against short another, or two sector stock futures; here, take long NIFTY futures against short BANKNIFTY futures when NIFTY is judged cheap relative to BANKNIFTY. Size to match notional: one NIFTY lot is 24,000 × 75 = ₹18,00,000, so a comparable BANKNIFTY exposure at 52,000 needs about ₹18,00,000 ÷ (52,000 × 30) ≈ 1.15 lots — rounded to one lot, the hedge is slightly imperfect. If the spread reverts — NIFTY gains 150 points (₹11,250) while BANKNIFTY is flat — the pair earns ₹11,250 gross before brokerage, STT, exchange charges, stamp duty and GST on both legs. If instead the relationship breaks and NIFTY falls 200 (−₹15,000) while BANKNIFTY rises 300 (short loses ₹9,000), the pair loses ₹24,000 — the neutrality gave no protection because the legs decoupled.

BANKNIFTY example

Taking a stock-futures pair in spirit but illustrating on BANKNIFTY as one leg: long BANKNIFTY futures (lot 30 at the time of writing) against a short in a correlated index future, entered when BANKNIFTY looks cheap on the spread. If the spread reverts and BANKNIFTY gains 250 points while the short leg is flat, the long earns 250 × 30 = ₹7,500 gross before costs on both legs. If the relationship breaks and BANKNIFTY falls 300 (−₹9,000) while the short leg rises against you, the combined loss can exceed ₹15,000. These are illustrative round levels; the arithmetic is points × lot size per leg, and matching the two legs' rupee exposure across different lot sizes is exactly the fiddly, error-prone part of building a real pair.

Lot sizes used above (NIFTY 75, BANKNIFTY 30) are those in force at the time of writing; NSE revises them periodically. Figures exclude brokerage, STT, exchange charges, stamp duty and GST, all of which materially affect small spreads.

Common mistakes

  • Treating correlation as if it were cointegration, and fading a spread that is merely co-moving rather than mean-reverting, so it can widen without ever coming back.
  • Assuming market-neutral means low-risk, and sizing up into a relationship that fails in stress exactly when the neutrality was supposed to matter.
  • Sizing the two legs by lots or points rather than matching rupee exposure, leaving the pair unintentionally directional rather than neutral.
  • Holding a broken pair on the belief the spread must revert, when a structural break or corporate event has permanently changed the relationship.
  • Ignoring the liquidity of the weaker leg, so exiting in stress incurs heavy slippage precisely when the pair needs to be unwound.
  • Underestimating the doubled transaction and margin costs of two legs, so a spread that reverts only slightly ends in a loss after costs.
  • Forgetting that a gap in either leg can blow the spread past the intended exit, so the loss on a break is realised well beyond the plan.

Advantages & disadvantages

Advantages

  • Bets on the spread rather than direction, so a well-behaved pair can profit in rising, falling or flat markets alike — a market-neutral return.
  • Sized to match exposure, the pair's net delta is near zero, reducing sensitivity to overall market moves that a single-instrument trade carries fully.
  • Expresses the view with constant deltas and no time decay, so waiting weeks for a spread to revert costs nothing per day, unlike an option pair.
  • The reward is defined by the spread's reversion to a normal level, giving clear exit logic that does not depend on an open-ended move.
  • Forces a rigorous distinction between correlation and cointegration, which sharpens a trader's understanding of what actually makes a spread tradeable.

Disadvantages

  • Undefined risk on each leg, and the market-neutral construction caps overall direction but not the spread, so a decoupling can inflict a large loss.
  • The relationship the whole trade relies on tends to break in stress — exactly when it is most needed — and the hedge then stops hedging.
  • Correlation is not cointegration, so a pair chosen on co-movement alone can have a spread that widens permanently without reverting.
  • Two legs mean doubled transaction and margin costs, so the modest spread gains are eroded and a small reversion can end in a loss.
  • Matching rupee exposure across different lot sizes and price levels is fiddly and error-prone, and the weaker leg's liquidity limits the whole pair.

Adjustments & exits

  • A trader may test a pair for cointegration rather than correlation before trading it, reducing the chance of fading a spread that never reverts, at the cost of a smaller universe of qualifying pairs.
  • A trader may cap the spread level at which the pair is abandoned as broken, accepting a defined loss on a decoupling rather than holding a widening spread in hope of reversion.
  • A trader may weight the legs by notional and re-balance as prices move, keeping the pair neutral at the cost of extra transaction costs from adjusting the hedge.

Adjustment is a decision about risk, not a way to rescue a losing view. See Adjustments and Exit Planning.

Professional usage

Statistical-arbitrage desks run pair and basket trades systematically across hundreds of relationships, testing for cointegration, sizing each spread small, and diversifying so no single decoupling dominates, with cross-margin and financing retail cannot access. They monitor relationships continuously and cut a pair the moment its statistical basis weakens, rather than holding in hope. The concept is replicable by retail with two stock or index futures, but the cointegration testing, the breadth that dilutes a single break, and the execution across two liquid legs are the hard parts, so a retail pair trader is far more exposed to the one relationship breaking than a stat-arb desk is.

Key takeaway

Pair trading profits from a spread reverting rather than from market direction, but it depends on cointegration — not mere correlation — and its defining risk is that the relationship breaks in a crisis, precisely when the hedge was supposed to protect you.

Frequently asked questions

What is pair trading in futures?
Pair trading goes long one instrument and short a correlated other, betting the spread between them reverts to its normal level. It profits from the spread, not market direction, so it is roughly market-neutral — but its central risk is that the relationship breaks when it is most needed.
What is the maximum loss on a pair trade?
It is not capped by structure. If the relationship breaks, both legs can move against the position at once — the short leg's loss unbounded, the long leg's reaching zero — and a gap can carry the spread past the intended exit. Market-neutrality removes direction, not this spread risk.
What is the maximum profit?
It is bounded by the spread's reversion, not by structure. The pair profits as an abnormal spread returns to its normal level, so the gain is limited by how far the spread had stretched. The trader targets the spread's mean, not an open-ended move in either leg.
What is the difference between correlation and cointegration?
Correlation measures whether two series move together day to day. Cointegration means their difference stays bounded and reverts to a stable level even as each series wanders. Pair trading needs cointegration; a merely correlated pair can have a spread that widens permanently without reverting.
Why is pair trading called market-neutral?
Because long one leg and short a correlated other, sized to match exposure, offsets overall market direction — if both rise or both fall together, the legs cancel and only the spread matters. It can profit in rising, falling or flat markets, provided the relationship holds.
What is the biggest risk in pair trading?
That the correlation breaks in stress — exactly when you rely on it. In a crisis, instruments that normally move together can decouple violently, the spread blows out instead of reverting, and both legs move against you at once. The hedge stops hedging when neutrality was supposed to matter.
Does time decay affect a pair-trading futures position?
No. Futures have zero theta on both legs, so holding a pair for the weeks a spread may take to revert costs nothing per day. Expressing relative value with options would add theta on each leg; the futures pair waits for the spread, indifferent to time, paying only carry and roll.
Does implied volatility matter for pair trading?
Not directly. Futures have no vega, so option-market implied volatility does not move either leg. What matters is the realised volatility of the spread itself — a pair of volatile instruments can have a quiet spread, and the danger is the spread's volatility spiking as the legs decouple in stress.
How much capital do I need for pair trading?
Enough to cover full naked SPAN plus exposure margin on both legs, which may not net into a spread benefit, plus a buffer for a decoupling that hits both legs. Because there are two legs, margin and costs are doubled, so a pair ties up more capital than a single-leg trade.
Is pair trading suitable for beginners?
It is generally advanced. It requires distinguishing cointegration from correlation, sizing two legs to match rupee exposure across different lot sizes, managing two liquidity profiles, and accepting that the relationship can break in stress. The market-neutral label makes it feel safer than its undefined spread risk warrants.
How do I size the two legs of a pair?
By matching rupee exposure, not lots or points, because the two instruments have different lot sizes and price levels. One NIFTY lot at 24,000 is ₹18,00,000 of notional, so the other leg must be sized to roughly the same rupee exposure, which usually means an imperfect round-lot approximation.
Can a stop protect me if a pair decouples?
Only partially. A stop on the spread becomes market orders on the legs when hit, and if either leg gaps, the spread can open past the level, so the fill is worse than intended. A decoupling can also be fast within the session, so the stop manages but does not guarantee the loss.
How is pair trading related to mean reversion?
Pair trading is mean reversion applied to a spread rather than to a single price. It fades an abnormal spread on the assumption it reverts to a stable level, just as mean reversion fades a stretched price. The difference is that the mean-reverting object is the difference between two instruments.
How is pair trading done in India?
Typically with stock futures — two correlated stocks in the same sector — or between index futures. Each leg carries its own SPAN plus exposure margin, the legs must be sized to match rupee exposure, and both must be liquid enough to enter and exit, since the weaker leg's liquidity limits the whole pair.
What horizon does pair trading use?
It is typically a multi-week approach, because a stretched spread can take time to revert to its normal level. Futures carry no time decay, so the pair can wait for the spread without a daily cost, unlike an option-based relative-value trade that would bleed theta while waiting.
Does leverage make pair trading more dangerous?
Yes. Each leg is a leveraged future, and the market-neutral label tempts larger size. If the relationship breaks, leverage amplifies the adverse move on both legs at once, so a decoupling can produce a loss large relative to the modest spread gains the strategy usually collects.
What is a cointegrated pair?
A cointegrated pair is two instruments whose individual prices can each wander far, yet whose difference stays bounded and reverts to a stable mean. That reverting spread is what pair trading needs; a pair that is correlated but not cointegrated can drift apart without the spread ever returning.
Should I hold a pair whose spread keeps widening?
A widening spread may be a deeper temporary stretch or a permanent break, and the two look alike while they happen. Holding assumes reversion; the spread may never return. The trade-off is patience versus a structural break, and naming that trade-off matters more than a rule to hold or fold.
Does pair trading predict which leg will outperform?
Not in absolute terms. It bets, without confirmation, that the spread between the two reverts to normal, regardless of whether both rise or both fall. It makes no directional forecast on either leg alone, and whether the spread reverts or breaks is only known after the fact.
How do transaction costs affect pair trading?
Every entry and exit pays brokerage, STT, exchange charges, stamp duty and GST plus the spread on both legs, so costs are doubled versus a single-leg trade. Because the spread gains are typically modest, these costs eat a meaningful share, and a small reversion can end in a loss after them.
Why does the correlation break exactly when I need it?
Because calm markets let normal relationships hold, while stress breaks them — one leg is sold for liquidity, a shock hits one instrument, or an event severs the link. The neutrality that made the pair feel safe depends on co-movement that a crisis is precisely the thing to destroy.

Voice search & related questions

Natural-language questions people ask about the Pair Trading.

What is pair trading?
Pair trading buys one instrument and short-sells a related one at the same time, betting on the gap between them returning to normal rather than on the market's direction. If one gets unusually expensive relative to the other, you short it and buy the cheap one, expecting the gap to close.
What's the difference between correlation and cointegration?
Correlation just measures whether two things move together day to day. Cointegration means their difference stays tethered and comes back to a stable level even as each one wanders. Pair trading needs cointegration — a merely correlated pair can drift apart forever without the gap ever closing.
Is pair trading really market-neutral and safe?
It's roughly market-neutral because the long and short offset overall direction, but that's not the same as low-risk. Its big danger is that the relationship breaks in a crisis — exactly when you need it — and then both legs move against you and the hedge stops hedging.
What's the biggest risk in pair trading?
That the correlation you rely on breaks under stress. In a crisis, instruments that normally move together can decouple hard — the spread blows out instead of coming back, and both legs lose at once. The neutrality that made it feel safe fails right when it was supposed to matter.
Is pair trading good for beginners?
It's generally advanced. You have to tell cointegration from correlation, size two legs to match rupee exposure across different lot sizes, handle two liquidity profiles, and accept the relationship can break in stress. The market-neutral label makes it feel safer than its real spread risk.
How is pair trading done in India?
Usually with stock futures — two correlated stocks in the same sector — or between index futures like NIFTY and BANKNIFTY. Each leg has its own margin, you size them to match rupee exposure, and both legs need enough liquidity, since the less liquid one limits the whole pair.

Sources & references

Last reviewed 9 July 2026. Educational content only — not investment advice.

← Previous
Gap Trading
Educational content only — not investment advice. Payoff diagrams and Greek curves are computed from the illustrative legs shown, not from live quotes. Options and futures carry substantial risk, including loss exceeding your deposit on undefined-risk positions. See our Risk Disclosure and SEBI Disclaimer.