Key takeaways
- Correlation runs from -1 to +1 and is a historical statistic, not a guarantee.
- Pairs share currencies, so several trades can add up to one large bet on the same currency.
- With three equal positions at 0.8 correlation, combined risk is about 2.8 times a single trade, not 1.7 times.
- Track net exposure per currency and cap total risk per currency or theme.
- Correlations change and often rise in stress; recheck them regularly.
Opening three different pairs feels like diversification. Often it is the same trade three times. Currency pairs share currencies, so many of them move together, and a portfolio of positions that look separate can behave like one oversized position when the market turns.
What correlation actually measures
Correlation is a number between -1 and +1 describing how two price series move together. At +1 they move in lockstep, at -1 they move in opposite directions, and near 0 there is no consistent relationship. The figure is calculated from a window of historical returns, for example the last 60 daily changes.
Two points are often missed. Correlation describes direction, not size: two pairs can be highly correlated yet differ in how many pips they move. And it is a statistical description of the past window, not a law, so it changes over time.
| Combination (historical tendency) | Typical relationship | What it means in practice |
|---|---|---|
| EUR/USD and GBP/USD | Often positive | Both lean on the dollar; being long both is mostly a short-dollar bet |
| EUR/USD and USD/CHF | Often negative | Long EUR/USD and long USD/CHF partly cancel; long both can be a hedge |
| AUD/USD and NZD/USD | Often strongly positive | Two similar economies; close to the same trade |
| USD/CAD and oil | Moderate, unstable | Link to commodity prices varies by period |
| EUR/USD and USD/JPY | Weak, changes sign | Little consistent link, so combining them may spread risk |
The table describes tendencies seen in some periods and nothing more. Relationships between pairs can weaken, strengthen or flip, especially around major events.
Where the doubling comes from
Every pair is two currencies. Long EUR/USD means long euro and short dollar. Long GBP/USD means long pound and short dollar. Hold both and you are short dollar twice. If the dollar rallies broadly, both trades lose at once.
The reverse also confuses people. Long EUR/USD and short GBP/USD cancels the dollar leg and leaves a bet on the euro against the pound, which is a different position than it looks, and you pay spread on two trades to hold it.
A worked example
Take a 10,000 USD account and a rule of risking 1% (100 USD) per trade. You open three long positions: EUR/USD, GBP/USD and AUD/USD, each risking 100 USD to its stop. If they were unrelated, a typical combined swing would be the square root of three times the single-trade risk, about 1.73 times 100 USD.
Now suppose the pairwise correlation is 0.8. Using the standard formula for three equal positions, the combined risk is the single-trade risk times the square root of (3 + 6 x 0.8) = the square root of 7.8, which is about 2.79. That is roughly 279 USD, close to the 300 USD you would lose if all three stopped out together. Your real exposure is nearer to 2.8% than 1%.
Illustrative figures from the formula above: risk as a multiple of one trade's risk, for different correlations between the pairs.
To keep total risk near 1% in the 0.8 case, each position would need to risk about 1% / 2.79 = 0.36% (36 USD), instead of 1%.
Checking correlation before you add a trade
- List the net currency exposure. Write each open position as long or short in each currency. Look for the same currency appearing on the same side more than once.
- Check a correlation matrix. Many platforms and websites provide one. Use returns over a window that matches your holding period, such as 60 daily bars for swing trades.
- Compare windows. A pair of numbers from 20 days, 60 days and a year shows whether the link is stable or drifting.Tip: If the readings disagree widely, treat the relationship as unreliable.
- Cap risk per currency, not just per trade. Set a limit on the total you will risk on any single currency or theme, such as the dollar.
- Scale down correlated positions. If you want two similar trades, reduce each so the combined risk fits your rule.
Limits of correlation
- Correlations spike in stress. In sharp risk-off episodes, many pairs start moving together and apparent diversification shrinks.
- Statistical calculation depends on the window and data frequency; two tools can give different values.
- Correlation does not show causation. Two pairs may move together because of a third driver, such as dollar strength or commodity prices.
- Opposite-direction trades that cancel out may pay spreads and swaps without much exposure.
Frequently asked questions
Is a negative correlation a good hedge?
It can offset risk while the relationship holds, but a hedged pair of trades still pays two spreads and swaps, and the link can break. A hedge reduces exposure rather than removing it.
How often should I update correlation figures?
Review them whenever your holding period or the market backdrop changes, and at least before adding positions. Weekly checks suit swing traders; day traders may use shorter windows.
Does trading unrelated markets, like indices and gold, avoid the problem?
It can reduce it, but relationships exist across asset classes too, particularly in stressed markets. Check the dollar and risk-sentiment links.
Can I use a single volatility measure instead?
Volatility tells you how big each pair's moves are; correlation tells you how they combine. A complete picture uses both.