Short answer: currency correlation is a number between -1 and +1 that describes how closely two currency pairs have moved together over a chosen window. Near +1 they mostly rise and fall in step, near -1 they mostly move in opposite directions, and near 0 the link is weak. Traders use it for risk management, to avoid stacking the same bet twice, and sometimes to hedge. The figure summarises past behaviour, so it drifts and needs rechecking often.
What a correlation coefficient actually tells you
The correlation coefficient compares the direction and size of price changes in two pairs, period by period. A value of +1 means every up move in one pair came with a proportional up move in the other. A value of -1 means the mirror image. Zero means the moves were scarcely related at all.
- A value above roughly +0.7 or below -0.7 is strong enough to change position sizing.
- Anything between -0.3 and +0.3 is noise for most trading strategies.
- The sign tells you direction, not magnitude. GBP/USD often travels further than EUR/USD in pips even when the two correlate closely.
One honest caveat: the coefficient says nothing about which pair moves first. It describes the past and forecasts nothing.
Readings also drift over time, so a figure from one quarter is stale by the next. The table below is a rough guide to what I do with each band when a reading is expressed as a percentage.
| Reading (percent) | What it usually means | How I treat it |
|---|---|---|
| 80 to 90 | Near-identical movement | One position, not two |
| 71 to 77 | Strong shared direction | Halve the second position |
| 62 to 68 | Moderate link | Trade both, cap combined risk |
| 46 to 51 | Loose link | Separate trades, but watch the pair |
| 14 to 28 | Little relationship | Independent trades |
| Negative 66 to 76 | Mirror image | A long plus a short doubles the bet |
How to calculate currency correlation by hand
The standard method is the Pearson coefficient. Take the daily change of each pair, find the average change, and measure how far each day sits from that average. Multiply the two deviations day by day and add them up. Then divide by the square root of the product of the two squared-deviation sums. Any spreadsheet does this with one function; the tedious part is collecting clean data.
A small worked example with five days of pip changes:
| Day | EUR/USD change | GBP/USD change |
|---|---|---|
| 1 | +40 | +50 |
| 2 | -20 | -10 |
| 3 | +10 | +20 |
| 4 | -30 | -40 |
| 5 | +25 | +15 |
Both pairs rose and fell on the same days, and the Pearson result comes out at roughly 0.95. Five days is far too short for a real decision; 20 to 50 periods is the ordinary working range. The Federal Reserve H.10 foreign exchange rates release publishes daily dollar exchange rates for major currencies, which is a tidy source for a weekly recalculation.
Correlated currency pairs worth knowing
| Pair combination | Usual relationship | Why | What breaks it |
|---|---|---|---|
| EUR/USD and GBP/USD | Positive | Both are priced against the USD | UK-specific news and rate decisions |
| EUR/USD and USD/CHF | Negative | Dollar on opposite sides, euro and franc linked | Swiss franc safe-haven flows |
| Gold and USD/JPY | Loosely negative | Both react to real yields and market sentiment | Risk shocks where yen and gold rise together |
The reasons column is more useful than the sign. When you know why two pairs correlate, you can spot the day the reason stops applying. If you also trade metals, our guide to trading gold with a prop firm account covers how gold's dollar link shapes stop placement.
Forex strategies that use correlation
- Avoid doubling up. Long EUR/USD and long GBP/USD amounts to a single dollar bet. Halve the size of each, or pick the pair with the cleaner setup.
- Hedge partially. A long EUR/USD with a smaller long USD/CHF offsets part of the dollar exposure while keeping the euro view.
- Trade the divergence. When a strongly correlated pair lags badly, some traders buy the laggard. This is a slower, patchy edge and it needs wide stops.
Worked numbers: a 10,000 account risking 0.5% per trade puts 50 at risk on each of two trades. Take both EUR/USD and GBP/USD long with 50-pip stops and a shared dollar shock can hit both stops in the same hour. That is 100 lost, or 1% of the account, from one idea wearing two tickets. For very short holds the effect is smaller but not gone, as we note in our list of the best forex pairs for scalping.
Limitations and common mistakes
Correlation is fickle. A pairing that read +0.9 over three months can go stale within a fortnight when economic indicators for one economy surprise. The coefficient also depends heavily on the window: a one-hour reading and a one-month reading for the same pairs often disagree, sometimes sharply.
- Reading a correlation calculator once and treating the result as permanent.
- Assuming correlation implies equal volatility, then sizing both legs the same.
- Hedging with a correlated pair and paying spread and commission cost twice for a position that nets to almost nothing.
- Mixing timeframes, such as a daily correlation figure applied to a five-minute trade.
Correlation also clusters in a crisis. In a disorderly sell-off, pairs that seldom moved together suddenly do, and a book that looked diversified turns brittle. Skip this check and you squander a benefit you never really had.
Correlation and prop-firm evaluation limits
Evaluation accounts make correlated exposure consequential because the daily loss limit does not care how many tickets you opened. Three losing trades that were really one lopsided dollar bet can burn a whole day's allowance before lunch. A static total drawdown is more forgiving than a trailing one, but it still shrinks fastest when correlated positions lose together.
Pivex publishes both limits before you begin: a daily loss allowance that renews each day and a total drawdown that never trails. That lets you size correlated trades against known figures, and the built-in assistant flags when a second trade would quietly push the day's risk past your plan. For the maths behind how repeated correlated losses compound, see our piece on risk of ruin in trading.
A decision framework checklist before you add a correlated trade
- Check the correlation coefficient on the timeframe you actually trade.
- If it is above +0.7 or below -0.7, treat both trades as one exposure for risk management.
- Add up the combined risk in account currency, not in pips.
- Ask bluntly whether the second trade adds information or just repeats the first.
- Recheck weekly, and after any major release, because the reading goes stale.
Next step: pull two months of closing prices for the pairs you trade most, compute the coefficients, and keep the results beside your position sizing rules. If you plan a one-step evaluation with Pivex, do this before the attempt starts. The daily loss limit will not distinguish between one bad idea and three copies of it.
Related guides: best forex pairs for scalping and risk of ruin in trading.
Before participating, review Pivex's current program rules and decide whether a simulated proprietary trading evaluation fits your experience and risk limits. Participation does not guarantee a reward, income, or any particular trading result.