Currency correlation: how forex pairs move together and why it matters for risk

Currency correlation measures how consistently two currency pairs move in the same or opposite direction. It is expressed as a coefficient between +1 and −1: close to +1 means the pairs tend to rise and fall together, close to −1 means one tends to rise when the other falls, and close to 0 means there is no reliable linear relationship. For risk management it matters because trades on correlated pairs can turn several apparently separate positions into one large bet.

This guide explains what the number means, how it is calculated, why pairs are correlated in the first place, how to measure the exposure you actually hold, and the ways correlation misleads people. It builds on the position sizing guide, which introduces the idea that two trades can be one risk; here we go into the detail.

Why forex pairs are correlated at all

Every forex price is a ratio of two currencies. EUR/USD is the price of a euro in dollars; GBP/USD is the price of a pound in dollars. If the dollar weakens broadly, because of a change in US interest-rate expectations, say, both pairs tend to rise at the same time. Nothing about the euro or the pound needs to change. The common component, the dollar, does the work.

That is the main reason forex correlations exist, and it is mechanical rather than mysterious. There are only a handful of major currencies, and the most traded pairs share them. The currency pairs guide explains base and quote currencies; the important fact here is that the US dollar is on one side of most of the heavily traded pairs. When the dollar moves, a large part of the market moves with it.

On top of that shared-currency effect sit economic links:

• Neighbouring economies. The euro, the Swiss franc and sterling are all exposed to European growth and trade, so they often move in sympathy against the dollar.
• Commodity links. The Australian, New Zealand and Canadian dollars are often described as commodity currencies because their economies export raw materials. AUD/USD and NZD/USD frequently move together, and the Canadian dollar has historically had a link with oil prices. These relationships are real but loose, and they come and go.
• Risk sentiment. In periods of market stress, traders often buy the yen, the Swiss franc and the US dollar, and sell currencies linked to growth and commodities. That can make pairs as different as AUD/JPY and equity indices move together for a while.
• Interest-rate expectations. Currencies whose central banks are on a similar path tend to drift together. The central banks guide and the guide to what moves currency prices cover these drivers.

What the coefficient means

The usual measure is the Pearson correlation coefficient, which describes how closely two series move together in a straight-line sense. A rough reading guide:

• +0.8 to +1.0: strong positive. The pairs usually move in the same direction on the same days.
• +0.5 to +0.8: moderate positive. Same direction more often than not, with plenty of exceptions.
• −0.5 to +0.5: weak or no consistent relationship.
• −0.5 to −0.8: moderate negative.
• −0.8 to −1.0: strong negative. The pairs usually move in opposite directions.

Three things the coefficient does not tell you:

1. The size of the moves. Two pairs can have a correlation of +0.9 while one moves twice as many pips as the other. Correlation is about direction and consistency, not magnitude. To compare size you also need each pair's volatility.
2. Cause and effect. A high coefficient says the pairs have moved together over the period measured. It does not say one drives the other or that they will keep doing so.
3. Behaviour in the tails. The coefficient summarises ordinary days. Two pairs with a modest correlation can still move violently together on the one day that matters.

Typical relationships, described qualitatively

Correlations are measured from data and change over time, so treat the following as the kind of pattern traders often observe rather than fixed facts. We deliberately give no coefficients here; any number you see quoted for a pair of pairs is only true for the period and timeframe it was calculated on.

• EUR/USD and GBP/USD have usually shown a positive correlation, because both have the dollar as the quote currency and the two European economies are closely linked.
• EUR/USD and USD/CHF have usually shown a negative correlation. The dollar is the quote currency in one and the base currency in the other, and the euro and franc have often moved in a similar direction against the dollar. When EUR/USD rises, USD/CHF often falls.
• AUD/USD and NZD/USD have usually been positively correlated, reflecting the close trading relationship between the two economies and their similar sensitivity to commodity prices and risk sentiment.
• EUR/USD and USD/JPY have tended to be negatively correlated in many periods, because a broad dollar move pushes them in opposite directions, but the yen has its own drivers, so the relationship is less stable.
• Crosses such as EUR/GBP behave differently because there is no dollar in them at all. EUR/GBP can be quiet on days when EUR/USD and GBP/USD both move a long way, because a dollar move cancels out.

You can look at the pair profiles for more on each pair's character, including EUR/USD and AUD/USD.

How correlation is calculated: a worked example

You do not need to calculate correlation by hand, but doing it once shows what the number is made of and stops you misreading it. The figures below are invented for illustration.

The first rule is to use returns, not prices. Two price series that both trend upwards over a year will show a high correlation even if their day-to-day moves have nothing to do with each other. Correlation of price levels is one of the most common errors in amateur analysis. Use the daily (or hourly, or weekly) percentage change instead.

Suppose two pairs, A and B, have these daily returns over five days:

• Day 1: A +0.4%, B +0.3%
• Day 2: A −0.2%, B −0.1%
• Day 3: A +0.3%, B +0.4%
• Day 4: A −0.5%, B −0.4%
• Day 5: A +0.1%, B 0.0%

Step 1: find the averages. A averages +0.02% a day and B averages +0.04%.

Step 2: subtract the average from each day to get deviations. For A: +0.38, −0.22, +0.28, −0.52, +0.08. For B: +0.26, −0.14, +0.36, −0.44, −0.04.

Step 3: multiply the deviations day by day and add them up: 0.0988 + 0.0308 + 0.1008 + 0.2288 − 0.0032 = 0.456.

Step 4: square each set of deviations and add them up. For A the total is 0.548; for B it is 0.412.

Step 5: divide the step 3 total by the square root of (0.548 × 0.412). The square root of 0.2258 is about 0.475, and 0.456 ÷ 0.475 is about 0.96.

So in this made-up sample, A and B have a correlation of about +0.96: they moved in the same direction on four of the five days, and by similar relative amounts. In a spreadsheet the CORREL function does all of this in one go; most charting platforms also offer a correlation indicator.

Two practical points. Five observations is far too few to mean anything; real calculations use dozens or hundreds. And the result depends heavily on the window you choose, which brings us to the biggest caveat of all.

Correlations change

A correlation is a summary of the past. It is calculated over a window, say the last 20, 60 or 250 trading days, on a particular timeframe. Change the window or the timeframe and you get a different number. Run the same calculation a few months later and you may get a very different number again.

There are several reasons relationships drift:

• Policy divergence. If two central banks that had been moving in step start to diverge, the currencies can decouple. A pair of currencies that moved together for a year can stop doing so within weeks.
• Regime changes in sentiment. In calm markets, currencies may trade mostly on their own domestic news. In stressed markets, a single risk-on or risk-off theme can dominate, pushing many correlations towards +1 or −1 at once.
• Commodity cycles. A commodity link can strengthen while a commodity is the main story and fade when it is not.
• Specific shocks. An election, a policy surprise or an intervention can move one currency sharply without affecting its usual partners.

The practical consequence is that correlation is least reliable exactly when you most want it. Many traders find that diversification across apparently different positions disappears in a sell-off, when everything linked to risk appetite moves the same way at once. Plan for correlations to rise in a crisis, not for them to hold at their calm-market levels.

A useful habit is to look at correlation over two windows, such as a short one (around a month of daily data) and a longer one (around a year). If they disagree, the relationship is unstable and you should treat it with suspicion.

Hidden exposure: think in currencies, not pairs

The most useful thing you can do with correlation is often not to calculate a coefficient at all, but to break your open positions down into the currencies they contain. Every forex position is long one currency and short another.

• Long EUR/USD = long euros, short dollars.
• Short USD/JPY = short dollars, long yen.
• Long GBP/USD = long pounds, short dollars.

Here is an illustrative account with three open trades, using example prices (EUR/USD 1.0800, GBP/USD 1.2700, USD/JPY 150.00):

• Long 0.50 lots EUR/USD: long €50,000, short about $54,000.
• Long 0.30 lots GBP/USD: long £30,000, short about $38,100.
• Short 0.40 lots USD/JPY: short $40,000, long about ¥6,000,000.

The trader may think of these as three ideas: a euro view, a sterling view and a yen view. Add up the dollar column and the account is short roughly $132,100. All three trades lose if the dollar strengthens broadly, and all three gain if it weakens. That is one large position on the dollar with three tickets attached.

A simple currency exposure table on paper or in a spreadsheet makes this visible: one row per currency, add the long amounts, subtract the short amounts, and convert everything to your account currency. If one currency dominates the net figure, you have concentrated risk whatever the pair names say.

Stacking risk: the arithmetic

Suppose you risk 1% of your account on each of two trades. How much are you really risking?

If both stops are hit, you lose 2%. That is the worst case and it is the right number to plan around. But correlation tells you how often you should expect the two trades to win or lose together, and how volatile the combined position is.

Statisticians combine the volatility of two positions using their correlation. For two positions of equal size, the combined volatility relative to one position is the square root of (2 + 2 × correlation). Using that formula, two equal 1% positions behave roughly like this (illustrative):

• Correlation +1.0: like a single position of 2.0%. No diversification at all.
• Correlation +0.8: like about 1.9%. Barely better than doubling one trade.
• Correlation +0.5: like about 1.7%.
• Correlation 0: like about 1.4%. This is where genuine diversification starts to show.
• Correlation −0.5: like about 1.0%.
• Correlation −1.0: the positions cancel. You are paying two sets of costs to hold nothing.

The lesson from the top of that list is that "spreading risk" across EUR/USD, GBP/USD and AUD/USD in the same direction usually spreads very little. The lesson from the bottom is that taking opposite positions in strongly correlated pairs does not create a clever hedge; it mostly creates costs.

Remember this formula describes ordinary variation. The stop-loss maths, both stops hit, is the floor you plan for. And neither accounts for gaps, which is why the guide on gap risk and weekend risk matters too.

Practical ways to use correlation in your risk rules

None of these require advanced maths. They are habits that make your sizing honest.

• Set a limit per currency, not just per trade. For example: no more than 2% total risk on positions that are all long or all short the same currency. Under that kind of rule, a third dollar trade means reducing the first two.
• Split risk across correlated ideas. If you want to express a view on dollar weakness through two pairs, take 0.5% on each rather than 1% on each. The position sizing guide shows how to turn a risk amount into lots, and the position size calculator does the arithmetic.
• Check the net currency exposure before each new trade. It takes a minute, and it is often the moment you realise the new trade is the same bet again.
• Remember non-forex instruments. Gold, equity indices and oil CFDs carry currency and risk-sentiment exposure too. Gold is priced in dollars, so a long gold position often behaves partly like a short dollar position. A long position on a US index and a long AUD/JPY may both be bets on risk appetite.
• Record it. Note the net currency exposure in your trading journal when you open a trade. Over time you will see whether your worst days came from concentrated exposure you did not notice.

Correlation and hedging

Some traders try to reduce risk by opening an offsetting position in a correlated pair rather than closing the original trade. For example, long EUR/USD and long USD/CHF at the same time, on the basis that they usually move in opposite directions.

There are problems with this:

• The hedge is imperfect. A correlation of −0.8, say, still leaves a lot of independent movement. When the relationship weakens, both sides can lose at once.
• You pay twice. Two spreads, two commissions and two sets of overnight financing. Swaps on the two legs may both be charges rather than offsetting each other; the swap and overnight financing guide explains how those are set.
• It obscures the decision. If you no longer want the risk of the original trade, closing it is cleaner, cheaper and easier to evaluate later.

There are legitimate reasons for hedging, mostly in professional settings where a business has real currency exposure. For a retail trader, a correlation hedge is usually a way of avoiding the decision to take a loss.

Common mistakes

• Treating trades in different pairs as different risks. EUR/USD, GBP/USD and AUD/USD in the same direction are largely one dollar trade.
• Correlating prices instead of returns. Two trending price series always look correlated. Use percentage changes.
• Trusting one number. A coefficient from one window and one timeframe is a snapshot. Check at least two windows and look at a chart of the rolling correlation if your platform offers one.
• Assuming correlations hold in a crisis. Stress tends to push correlations towards extremes. Diversification often shrinks exactly when it is needed.
• Ignoring magnitude. A correlation of +0.9 between two pairs with very different volatility still means very different pip moves, and therefore different money at risk for the same lot size.
• Building hedges from correlations. Two positions that partly cancel each other cost twice as much and often fail together.
• Forgetting gold and indices. Exposure to the dollar and to risk sentiment does not stop at forex pairs.

The most important point sits underneath all of these. Correlation does not change the basic odds of trading leveraged products: most retail CFD accounts lose money, as our risk warning explains, and the guide to why most retail CFD traders lose looks at the causes. Concentrated, unrecognised exposure is one of the ways a single bad day becomes a large loss.

Frequently asked questions

What is currency correlation in forex?

It is a measure of how consistently two currency pairs move in the same or opposite direction, usually given as a coefficient between +1 and −1. Positive values mean the pairs tend to move together, negative values mean they tend to move in opposite directions, and values near zero mean there is no consistent relationship. See the correlation glossary entry for a short definition.

Which forex pairs are most correlated?

Pairs that share a currency on the same side, such as EUR/USD and GBP/USD, or AUD/USD and NZD/USD, have often shown strong positive correlation. Pairs where the shared currency sits on opposite sides, such as EUR/USD and USD/CHF, have often shown strong negative correlation. These relationships change over time, so check current data rather than relying on a published table.

Is it safe to trade two correlated pairs at the same time?

It is not a question of safe or unsafe, but of how much you are really risking. Two positively correlated trades in the same direction behave much like one larger trade, so most traders reduce the size of each so that the combined risk stays within their normal limit for one idea.

How often do currency correlations change?

Continuously. They depend on the window and timeframe used, and they can shift noticeably within weeks when central-bank policy diverges, a major shock hits one economy, or market sentiment swings between risk-on and risk-off. Many traders recheck the relationships they rely on at least monthly.

Can I use negative correlation to hedge a trade?

You can, but it is an imperfect hedge that costs two spreads and two sets of overnight financing, and it can fail when the relationship weakens. If you no longer want the risk of a position, closing or reducing it is usually simpler and cheaper.

Does correlation matter if I only trade one pair?

Less, but it still matters if you hold gold, indices or other CFDs alongside your forex trade, since many of those share exposure to the dollar or to risk sentiment. It also matters if you hold the same pair in more than one position, which is the strongest correlation of all.