Forex Correlation Matrix

How to read it

Each cell shows the correlation between two pairs' daily returns over the chosen window. +1.0 means the pairs move in lockstep. −1.0 means they move opposite each other. 0 means independent.

The biggest insight for retail traders is the same-side trap: long EUR/USD + long GBP/USD + long AUD/USD looks like three trades but is essentially three legs of one short-USD bet. Their correlations are usually strongly positive, often +0.6 to +0.9. A loss on one is usually a loss on all three. The matrix makes that obvious.

How forex correlation is measured

The matrix uses the correlation coefficient (Pearson) of daily returns, not of prices. Two currency pairs that both trended up for a month would show a high price correlation even if their day-to-day price movements were unrelated; returns remove that trend. As a rule of thumb:

• ±0.8 to ±1.0: very strong. Treat the two pairs as one position.
• ±0.5 to ±0.8: strong. Size them together.
• ±0.2 to ±0.5: weak. Some shared exposure.
• Below ±0.2: close to independent.

Choosing the window

The selector sets how many business days go into each figure. 14 days gives 13 daily returns, so it reacts fast but a single big day can swing a cell. 90 days is steadier but slow to show a change. Compare 30 and 90: if a cell is similar in both, the link is stable; if they differ a lot, it is shifting.

Why some pairs show negative correlation

A negative number often just means the dollar sits on opposite sides of the quote. EUR/USD rises when the dollar weakens; USD/CHF falls when the dollar weakens. So a strong negative currency correlation between them mostly reflects the shared USD leg, not two unrelated markets moving apart. Flip USD/CHF to CHF/USD and the sign turns positive. Read each cell with the quote order in mind before calling it a hedge.

Forex correlation and position size

Correlation changes how much you really have at risk. Take two trades risking 1% each:

• Long EUR/USD and long GBP/USD, correlation +0.88: together they swing about as much as a single 1.9% position, not two independent 1% bets (which would behave like about 1.4%).
• Long EUR/USD and long USD/CHF, correlation −0.86: the two mostly cancel, leaving the swing of about a 0.5% position, while you pay the spread and swap on both.

Before adding a second trade, find the pair in the matrix. If the cell is above about +0.7 in the direction you are trading, cut both sizes so the combined risk stays where you want it. The position size calculator sizes each leg from its stop.

Currency pairs that often move together

• EUR/JPY and GBP/JPY: both carry the yen leg, so yen moves hit both at once.
• AUD/USD and NZD/USD: two commodity-linked economies with close trade ties to China and each other.
• EUR/USD and USD/CHF: mirror images through the dollar, and the euro and franc often move together against it.
• USD/CAD and oil: not in this matrix, but a higher oil price tends to strengthen the Canadian dollar and push USD/CAD down.

These are tendencies, not rules. The window selector shows whether a link is holding right now.

Practical uses

• Avoid double-exposure. If two pairs you want to trade have correlation above ~0.7, sizing them as independent positions is hidden over-leverage.
• Pair selection for diversification. A portfolio of pairs with mutually low (or negative) correlations is more resilient to single-currency shocks.
• Hedging. Strongly negative correlations (e.g. EUR/USD vs USD/CHF, often −0.85+) can hedge dollar exposure in either direction.
• Spot regime shifts. When historically uncorrelated pairs suddenly correlate, the market is being driven by a single macro factor (rates, risk-on/off). Cross-check the US Treasury yield curve, the DXY dollar index, and gold to identify the driver. Worth knowing before sizing up.

Why daily and not intraday?

Daily closes are taken at the same moment for every pair, so each day's returns line up exactly, and one close a day filters out the noise of individual ticks and broker spreads. Intraday correlation drifts with the trading session (Tokyo, London, New York) and swings much more from hour to hour.

FAQ

Why these ten pairs and not all 28?

They are the seven US-dollar majors plus three widely traded crosses. The eight major currencies make 28 pairs; a 28×28 matrix is unreadable on a phone, and most of the extra crosses can be read off the USD legs shown here.

Are these correlations stable?

No. Correlation regimes shift with macro conditions. EUR/USD and GBP/USD typically run +0.7 to +0.9 but can briefly drop to +0.3 during Brexit-style UK-specific events. Use the window selector (14d / 30d / 60d / 90d) to see how stable a given pair-pair relationship is across horizons.

Does correlation tell me which pair will move next?

No. It describes how two pairs' daily returns moved together over the window. It doesn't show which one leads, and a relationship that held for 90 days can break in a week.

How is the calculation done?

We take the daily European Central Bank reference rates for the seven US-dollar majors, keeping completed business days only, and build the three crosses from their dollar legs. For each pair we compute the log return on each day from the previous day's close, then the Pearson correlation coefficient for every pair-pair across the window. All four windows are recomputed on our server several times a day, and a new day enters once the ECB publishes, around 16:00 CET. Full formula in the methodology.

Pair this with our Currency Strength Meter to pick the cleanest side, and the Trade Plan Builder to size the position.