A beginner opens long EUR/USD, long GBP/USD and long AUD/USD and feels diversified: three pairs, three ideas. In practice, all three are short US dollar. If the dollar rallies sharply, all three can lose together. This is correlation at work, and it is one of the least appreciated sources of hidden risk in retail FX.
What correlation means
Correlation measures how closely two price series move together, on a scale from -1 to +1. A reading near +1 means they tend to move in the same direction; near -1 means they tend to move opposite; near 0 means there is no stable relationship. Importantly, correlation describes the direction of moves over a chosen window, not the size, and it says nothing about cause.
Why FX pairs are correlated by construction
Every FX pair is a ratio of two currencies. Pairs that share a currency inherit part of its behaviour. EUR/USD and GBP/USD both have the dollar on the bottom, so a broad dollar move pushes them in the same direction. EUR/USD and USD/CHF both involve the dollar but on opposite sides, and the euro and Swiss franc are economically close, so they have often moved in opposite directions, a negative correlation. Commodity currencies such as the Australian, New Zealand and Canadian dollars often move with risk appetite and commodity prices, which links them to each other and, to varying degrees, to oil and metals.
Three common relationships (historical, not guaranteed)
- EUR/USD and GBP/USD: historically positive, because both are mostly a dollar story.
- EUR/USD and USD/CHF: historically negative, for the reasons above.
- AUD/USD and NZD/USD: historically strongly positive, since the two economies and their trade links are similar.
The word "historically" matters. Correlations change with the macro regime. A pair that moved together for a year can decouple within weeks when central banks diverge. See how central banks move FX for why policy divergence is such a powerful driver.
Reading a correlation table
Many sites publish a matrix of correlation coefficients over 1-week, 1-month and 3-month windows. When you read one, bear in mind:
- Short windows are noisy; long windows are slow to reflect a regime change. Compare more than one.
- Correlations are usually computed on returns. Correlating raw price levels tends to produce misleadingly high numbers, because most trending series look related.
- A coefficient of 0.7 does not mean the pairs move 70 percent together. It describes the strength of a linear relationship; the relationship explains roughly half of the variation (0.7 squared is about 0.49).
- Correlations tend to rise toward 1 in a crisis, when everything trades as one risk-on or risk-off factor. The diversification you thought you had disappears just when you need it.
The real risk: stacking exposure
Suppose you risk 1 percent of your account on each of three trades, all of which are effectively short the dollar. If the dollar surges, the three stops can trigger together, and you lose roughly 3 percent on what you thought was three independent ideas. A cleaner way to think about it: count your exposure by currency, not by pair.
Illustrative example: long EUR/USD, long GBP/USD, short USD/JPY. Net exposure: long EUR, long GBP, short USD (twice), long JPY. Three trades, but a lot of dollar. If the dollar is not what you actually want a view on, you have taken more of that view than intended.
Using correlation constructively
- Cap currency exposure. Decide in advance a maximum of total risk you are willing to carry on any single currency, and treat highly correlated pairs as one position for sizing. Methods are covered in position sizing and risk per trade.
- Look for confirmation, cautiously. If a dollar-short idea shows up consistently across several dollar pairs, that is information about the dollar, not three independent confirmations.
- Beware of "hedging" with correlated pairs. Going long EUR/USD and long USD/CHF to hedge is not a clean hedge: the correlation is not -1, spreads and swap costs apply on both legs, and you pay twice for the privilege. If you want to reduce risk, usually the simpler action is to reduce size.
- Know your session. Relationships can shift between the Asian, European and US sessions; see forex trading sessions explained.
Correlation is not causation, and not a signal
When one pair moves and a correlated pair has not yet moved, it is tempting to assume the laggard will catch up. Sometimes it does; sometimes the relationship is breaking. Treat correlation as a risk-management tool first. It tells you how much you are exposed, far more reliably than it tells you what will happen next.
A quick weekly routine
- List every open position and rewrite it as currency exposures (long/short each currency).
- Total risk per currency in percent of account.
- Check a correlation matrix over two windows for pairs you hold or plan to hold.
- If any currency exceeds your limit, reduce or skip the new trade.
- Note the outcome in your journal, including cases where correlation surprised you.
This article is educational and not investment advice. Leveraged FX trading carries a high risk of loss. Correlation examples are illustrative and historical relationships may not persist.