Trading Strategies Aug 1, 2026 19 min read

Pairs Trading: How the Market-Neutral Spread Trade Works

Reviewed by Nina Carr

Pairs trading is a market-neutral strategy. Instead of betting on where a single market is heading, you trade the gap between two instruments that normally move together. You buy the one that looks cheap and short the one that looks rich, then wait for the two to converge. Because you are long one leg and short the other, a broad move up or down in the whole market mostly cancels out. What you are really trading is the spread between the pair. Traders usually track that spread as a z-score, opening a position when it stretches to about two standard deviations from its average and closing as it drifts back toward the middle. It is a close cousin of statistical arbitrage and a form of mean reversion, applied to a pair rather than one chart. This guide covers how pairs trading works, the common ways to run it, and where it quietly breaks, across forex, gold and crypto.

What pairs trading actually is

Most strategies pick a direction. Pairs trading refuses to.

You hold two positions at once, long one instrument and short a related one, and you profit when the distance between them closes.

Pairs trading anatomy chart showing two normalised legs, Leg A in navy and Leg B in orange, with the spread plotted below as a z-score oscillating between a plus two sigma and minus two sigma band, and a sell-spread signal marked where the spread pushes above plus two sigma — The anatomy of a pairs trade. The top panel shows two legs rebased to 100, so both start at the same value and you can compare them fairly. Leg A is the navy line, Leg B the orange one, on every chart in this guide. The bottom panel is the spread between them, expressed as a z-score. The red dashed line is +2 sigma (written σ on the charts) and the teal dashed line is -2 sigma. When the spread pushes above +2 sigma, the two have stretched apart, and the signal is to sell the spread: short the leg that ran ahead (Leg A) and buy the one that lagged (Leg B).

Read the chart above from the bottom panel up. The two price lines wander together for a while, then Leg A rips higher while Leg B stalls.

That gap is the spread, and the z-score in the lower panel measures how unusual it is.

The z-score is just the spread measured in standard deviations from its own average. A standard deviation is a way of measuring how unusual a move is, so two of them is genuinely rare. Zero means the pair is at its normal distance apart.
+2 sigma (the red line) means the spread is stretched wide, wider than it usually gets. Leg A has run too far ahead of Leg B.
-2 sigma (the teal line) is the mirror image. The spread has collapsed, so Leg A is unusually cheap against Leg B.
The shaded band between the two lines is the normal zone. Most of the time the spread lives in here, which is why you only act at the edges.
The bet is that a stretched spread snaps back toward zero. You fade the extreme and wait for the convergence.

You are not asking “will gold go up.” You are asking “have these two drifted too far apart, and will they close the gap.” That is the whole idea.

The core mechanic, in one table

Two setups, two mirror actions. The exit is the same either way: the spread coming home.

Spread reads	What it means	The trade
Above +2 sigma	Leg A too rich vs Leg B	Sell the spread: short A, buy B
Below -2 sigma	Leg A too cheap vs Leg B	Buy the spread: long A, short B
Back near zero	Pair converged	Close both legs, take the gap

“Buy the spread” and “sell the spread” trip people up. It just means the two-legged position as a whole.

Buying the spread is betting the gap widens in your favour by the cheap leg catching up; selling it bets the rich leg falls back. You always hold both legs at once.

Three ways traders run pairs trading

The core idea never changes. What changes is how you pick the pair and how you decide the spread is “stretched.” Three approaches cover most of what you will see.

1. The distance and z-score method

This is the one on the anatomy chart, and the one most beginners meet first. You rebase both instruments to the same starting point, measure the spread, and standardise it into a z-score.

Then you trade the ±2 sigma edges. It works like Bollinger Bands wrapped around the spread instead of price.

How it looks on a chart:

Two lines that track each other closely, then pull apart and rejoin.
A spread oscillator below that swings through a middle zone and pokes past the outer bands.
Clean signals when the lines are genuinely tied together, messy ones when they are not.

Role	How you use it	Best read
Entry trigger	Spread hits ±2 sigma	Fade the extreme
Exit cue	Spread returns toward zero	Close near the mean
Stop / bail	Spread runs past ±3 sigma	Relationship may be broken
Best fit	Quick screening	Any two correlated markets

The catch is honest and worth stating up front. The z-score assumes the average spread is stable.

If the two instruments quietly stop moving together, the spread can push to +3, +4 sigma and keep going, and the “cheap” leg never catches up. That is the classic way a pairs trade bleeds.

2. The correlation method, in forex

Here you pick the pair first, using a rolling correlation coefficient, then trade the spread the same way. Forex is a natural home for this, because currencies share drivers.

Two dollar pairs, or a metal and a currency, often lean on the same macro story.

The chart below is deliberately a loose pair, so you can see what “too loose” looks like: EUR/USD against spot gold (XAU/USD) on the daily.

Daily pairs trading chart of EUR/USD versus XAU/USD, both normalised to a base of 100, with the spread z-score plotted below and a buy-spread signal marked where the spread drops below minus two sigma — EUR/USD (navy) against spot gold, XAU/USD (orange), daily, both rebased to 100. The lower panel is the spread z-score with the +2 sigma and -2 sigma guides. At the teal marker the spread dropped below -2 sigma, so EUR/USD had fallen unusually cheap against gold, and the signal was to buy the spread: long EUR/USD, short gold. Notice how far the pair wanders. This is a loose relationship, not a tight one.

What the chart is really teaching:

Gold went on a long one-way run while the euro sat flat. The two are related, but not glued together.
The spread makes big, slow swings and touches both bands more than once.
A signal here can stay stretched for weeks, so this pair rewards patience and punishes a tight stop.

Role	How you use it	Best read
Pair selection	Rolling correlation coefficient	Trade pairs above roughly 0.8
Regime read	Is the correlation holding	Falling correlation, stand aside
Timeframe	Daily or 4-hour	Slower bars, cleaner spread
Best fit	Forex crosses, sector pairs	Same-driver instruments

Pairs trading in forex lives and dies on that correlation number. A pair of currencies that shares a central bank or a commodity link will hold together.

A macro pairing like the euro and gold drifts, so treat a chart like this one as a caution, not a green light. If you want the pieces of a good forex pair, the forex pairs guide covers how the majors relate.

3. Cointegration, and why crypto suits it

Correlation asks whether two markets move together day to day. Cointegration asks a stricter question: do they keep returning to a stable relationship over time, even if they wander in the short run.

It is the more rigorous test, and it is what serious desks lean on. Crypto majors are a good fit because they are driven by the same risk appetite and rarely decouple for long.

Here is Bitcoin (BTC/USDT) against Ether (ETH/USDT) on the 4-hour.

Four-hour pairs trading chart of BTC/USDT versus ETH/USDT, both normalised to a base of 100, with the spread z-score below and a sell-spread signal marked where the spread rises above plus two sigma — Bitcoin (navy) and Ether (orange), 4-hour, both rebased to 100. The two track each other tightly, which is exactly what you want in a pair. The lower panel spread z-score pokes above +2 sigma at the red marker, where Bitcoin had run ahead of Ether, and the signal was to sell the spread: short Bitcoin, long Ether. The spread then falls back toward zero. That convergence is what a pairs trade aims to capture.

Why this is the textbook version:

The two lines are almost stitched together, so the spread is well-behaved.
When it hits +2 sigma it actually comes back, rather than trending away.
Crypto trades around the clock, so there are no session gaps to distort the spread.

Role	How you use it	Best read
Pair selection	Cointegration test on history	Stable long-run relationship
Entry trigger	Spread hits ±2 sigma	Fade the tighter the better
Confirming gauge	Correlation still high	Both tests agreeing
Best fit	Crypto majors, index pairs	Instruments that rarely decouple

Pairs trading in crypto is popular for a reason. The majors are so tied to one risk cycle that the spread genuinely mean-reverts, which is the behaviour the strategy needs.

The trade-off is that funding costs on a short leg and the odd violent decoupling can still catch you.

How a pair fits together, by role

Whichever method you use, the moving parts do the same jobs. Reading a pair by role, rather than as one signal, is what keeps you out of a broken relationship.

Role	How you use it	What good looks like
Entry trigger	Spread stretched to ±2 sigma	A clear, rare extreme
Regime read	Rolling correlation of the legs	Still moving together
Confirming gauge	Cointegration over history	Relationship returns to mean
Filter / context	News on a single leg	No standalone story on one side
Exit cue	Spread back toward zero	The gap closes, you are done

Skip the filter and you learn the hard way. If one leg has its own catalyst, an earnings miss, a rate decision, a protocol upgrade, the pair is no longer trading its spread.

It is trading that one story, and your market-neutral hedge is suddenly a naked directional bet.

Which method to use when

A quick way to choose, from fastest to most rigorous.

Method	Picks the pair by	Best for
Distance / z-score	Standard deviation of the spread	Screening, first pairs
Correlation	Rolling correlation coefficient	Forex crosses, sector pairs
Cointegration	Long-run statistical relationship	Crypto majors, index pairs

Most people start with the z-score because it is visual and needs no coding. The correlation and cointegration checks are the upgrades that stop you trading a pair that only looked related.

You do not have to pick one forever. The z-score is your screen; correlation and cointegration are the checks you run before you trust it.

How to place a pairs trade

The mechanics, without the jargon. No numbers to memorise, just the sequence.

Pick two instruments that normally move together, and confirm it with a rolling correlation.
Build the spread, either the ratio of the two prices or the difference once both are rebased.
Turn the spread into a z-score so you can see when it is unusually wide or narrow.
Wait for the z-score to reach about +2 or -2. Below that, there is no edge, just noise.
At +2, short the leg that ran ahead and buy the one that lagged, in roughly equal dollar value.
At -2, do the reverse.
Close both legs when the z-score comes back near zero. Do not hold out for the opposite extreme.
Bail early if the correlation breaks or one leg gets its own news. A broken pair does not converge.

The equal dollar value in step five is the part beginners skip. On a small account, say $700, that might mean roughly $300 on the long leg and $300 on the short one, kept even in dollars.

Market-neutral only works if the two legs are balanced by size, so a broad market move really does cancel out. Weight one leg heavier and you have quietly built a directional trade with extra steps.

One more expectation to set. The ±2 sigma edge is rare by design, so plan for only a handful of setups and do not force a trade while the spread sits in its normal zone.

Where pairs trading breaks

The honest limitations, because this strategy fails in specific, predictable ways.

The correlation snaps. Two markets that moved together for a year decouple, the spread trends instead of reverting, and the “cheap” leg never catches up. This is the number one killer.
A single leg gets its own story. News on one instrument turns your hedge into a naked bet on that one thing.
You pay two spreads and two sets of costs. Every pairs trade is two positions. On tight-margin pairs, the round-trip cost can eat most of a small convergence.
Financing on the short leg. When you short something you are effectively borrowing it, and you pay a small daily fee to hold that short. Over days or weeks it grinds against a slow-converging trade.
It caps your upside on purpose. Because you are hedged, a huge one-way move you would have loved to ride passes you by. Market-neutral cuts both tails.

None of this makes pairs trading bad. It makes it a specialist tool.

It shines in choppy, sideways markets where a directional strategy struggles, and it lags badly in a strong one-way trend, the exact mirror of a breakout system.

What to remember

A pairs trading strategy comes down to three things.

Trade the spread, not the market. You are betting two related instruments converge, not that either one goes up.
Pick the pair with care. Correlation and cointegration are what separate a real pair from two charts that happened to rhyme.
Respect the ±2 sigma edges and the exit at zero. The discipline is the edge, and a broken correlation is your cue to leave.

Pairs trading sits alongside the carry trade as one of the few genuinely market-neutral plays a retail trader can run. It will not make headlines, and that is the point.

It earns its keep in the quiet, ranging markets where everything else is spinning its wheels.

Reproduce these charts

You can build the spread on free tools without any code.

On TradingView, type a ratio symbol straight into the search, for example EURUSD/XAUUSD or BTCUSDT/ETHUSDT, to plot the pair as one line. Add the “Correlation Coefficient” indicator to watch the relationship, and a “Bollinger Bands” or standard-deviation study on the ratio to see the ±2 sigma edges.
On MT4 or MT5, open both instruments and use a spread or correlation indicator from the “Insert, Indicators” menu. There is no one-click ratio chart, so most traders eyeball the two normalised lines side by side.
The z-score panel itself is a custom calculation. If your platform does not ship one, a community script or a simple spreadsheet on the price difference does the job.

Key terms

Spread: the gap between the two legs, either their ratio or the difference once rebased.
Z-score: the spread measured in standard deviations from its average. Zero is normal, ±2 is stretched.
Market-neutral: long one leg, short the other, so a broad market move cancels out.
Correlation: how closely two instruments move together, scored from -1 to +1.
Cointegration: a stricter test, whether two markets keep returning to a stable relationship over time.
Convergence: the spread coming back toward its average, which is the profit in a pairs trade.
Leg: one of the two positions in the pair.

FAQ

What is pairs trading, in plain terms?

Pairs trading is a market-neutral strategy where you trade two related instruments at once. You buy the one that looks cheap and short the one that looks rich, then wait for the gap between them to close. Because you are long one and short the other, a broad market move mostly cancels out, so what you are really trading is the spread between the pair.

How do you do pairs trading step by step?

Pick two instruments that normally move together and confirm it with a correlation check. Build the spread between them and turn it into a z-score. Wait for the z-score to reach about plus or minus two standard deviations. Short the leg that ran ahead and buy the one that lagged, in equal dollar value. Close both legs when the spread returns toward zero, and bail if the correlation breaks.

Does pairs trading actually work?

It works best in choppy, sideways markets on pairs that genuinely move together, like two crypto majors or two same-sector instruments. It struggles when a correlation breaks down, because the spread then trends away instead of converging. It is a specialist tool for range conditions, not an all-weather strategy, and it deliberately caps your upside because you are hedged.

What is the spread and the z-score?

The spread is the gap between the two legs, measured either as their ratio or as the difference once both are rebased to the same start. The z-score expresses that spread in standard deviations from its own average. Zero means the pair sits at its normal distance apart, while plus or minus two means it has stretched unusually wide, which is the entry signal.

How do you pick a good pair?

Look for two instruments driven by the same thing, so they move together for real reasons rather than by coincidence. Check a rolling correlation coefficient and favour pairs that stay well correlated, often above roughly 0.8. For a stricter test, use cointegration, which checks whether the two keep returning to a stable relationship over time. Crypto majors and same-sector instruments tend to pair well.

Can you do pairs trading in forex?

Yes, and forex is a natural home for it because currencies share macro drivers. Two dollar pairs, or a currency and a linked commodity, often lean on the same story. The catch is that some forex pairings, like the euro against gold, are loose relationships that wander for weeks, so the correlation number matters more here than anywhere. Trade the tightly correlated pairs and stand aside when the correlation falls.

Is pairs trading good for crypto?

Crypto suits pairs trading well because the majors are tied to one risk cycle and rarely decouple for long, so the spread genuinely mean-reverts. Bitcoin against Ether is a common example. The trade-offs are funding costs on the short leg and the occasional violent decoupling, so you still need the correlation filter and a bail-out rule when the relationship breaks.

What is the difference between pairs trading and statistical arbitrage?

Pairs trading is the simplest form of statistical arbitrage. Pairs trading works one relationship, long one instrument and short one related instrument. Statistical arbitrage is the broader, model-driven version that trades many such relationships across a basket at once. If you understand a single pair, you understand the building block of stat arb.

What is the best timeframe for pairs trading?

The daily and 4-hour charts tend to work best, because slower bars give a cleaner, less noisy spread. On very fast timeframes the z-score snaps between extremes constantly and the trading costs of two legs eat the small moves. Most pairs traders operate as swing traders, holding a convergence trade for days rather than minutes.

What are the main risks of pairs trading?

The biggest risk is a correlation breakdown, where the two markets stop moving together and the spread trends away instead of reverting. Other risks are news hitting one leg and turning your hedge into a directional bet, paying two spreads and two sets of costs on every trade, and financing charges on the short leg. It also caps your upside on purpose, since being hedged means you miss big one-way moves.

Reader Reviews

0.0 No reviews yet

Be the first to review this — tell other traders what actually helped, or where it fell short.

Leave a Review

James Hartwell

Forex Analyst & Senior Trader

Former FX desk trader with 8 years in institutional forex. Works in multi-timeframe analysis and order flow, turning desk experience into systematic, testable rules across forex and metals.

Forex AnalysisMulti-Timeframe AnalysisOrder FlowSystematic Rules