Correlation of 12 Pairs: A Measured Matrix Instead of Generalities
Currency pair correlation is usually discussed in generalities: the majors are linked, the crosses move together. We computed coefficients for all 66 combinations of twelve pairs and got a more complicated picture: a strong link exists in only 7 of the 66.
The currency pair correlation matrix: what we computed
The Pearson correlation coefficient between daily returns on closes, for each of the 66 possible instrument pairings. We used returns rather than prices: correlation of levels is inflated by shared trends and is useless for risk assessment.
The strongest links
- AUDUSD and NZDUSD: r = 0.9
- EURJPY and GBPJPY: r = 0.83
- EURUSD and GBPUSD: r = 0.8
- AUDJPY and GBPJPY: r = 0.76
- GBPUSD and NZDUSD: r = 0.74
- AUDJPY and EURJPY: r = 0.71
Measured across 518 daily bars for each of 12 currency pairs. The sample is public, the methodology is in how we run backtests, and the calculation is reproduced by a script in the repository.
Inverse links
- EURUSD and USDCHF: r = -0.81
- NZDUSD and USDCAD: r = -0.69
- AUDUSD and USDCAD: r = -0.68
- GBPUSD and USDCHF: r = -0.63
- EURUSD and USDCAD: r = -0.61
- EURUSD and USDJPY: r = -0.6
The main finding: strong links are rarer than assumed
Only 7 of 66 combinations showed |r| ≥ 0.7, roughly one in ten. The rest are moderately or weakly linked. That matters, because the common claim that all majors move together is not supported by the data: EUR/USD and USD/JPY, for instance, give -0.6 — a noticeable link, but far from rigid.
What to do with it when sizing risk
Add risk where the link is strong. Simultaneous positions in AUD/USD and NZD/USD are effectively one position of double size, and treating them as two independent trades at one percent risk each is wrong. Pairs with correlation near zero can be counted separately: more on total portfolio risk.
Limitations
Correlation is unstable. It strengthens during market stress, when almost everything starts moving together, and weakens when the market focuses on individual central bank policy. Our coefficients describe an average over 518 bars, not today's state, and are worth recomputing regularly.
This material is for educational purposes and is not individual investment advice.
Frequently asked questions
Which currency pairs correlate most strongly?
In our measurement the strongest link is AUDUSD and NZDUSD at r = 0.9, followed by EURJPY and GBPJPY (0.83) and EURUSD and GBPUSD (0.8).
Is it true that all majors move together?
No. Of 66 combinations only 7 showed a strong link (|r| ≥ 0.7). The rest are moderately or weakly linked, so counting their risk separately is justified.
Why compute correlation on returns rather than prices?
Because correlation of levels is inflated by shared trends and does not reflect joint risk. Judging whether positions lose together requires returns.
New backtest studies — to your inbox
When a new reproducible study comes out or the data is updated, we send a short email. No spam, no "signals", no selling your address.
By leaving your email you agree to receive occasional emails from the portal. Unsubscribe in one click from any of them.