Capital at risk. Leveraged forex is a high-risk product and most retail accounts lose money on it. Nothing here is advice or a trading recommendation.
A trader has six positions open. Different pairs, different entries, different setups, each risking one percent of the account. Total risk on the table: six percent, comfortably within any sensible limit.
Then one currency moves, and the account is down eleven percent in an afternoon.
Nothing malfunctioned. The six positions were never six independent bets. They were substantially the same bet, entered six times, and the risk calculation that said six percent was measuring tickets rather than exposure.
Every pair is two positions
The mechanism is not subtle once you see it. A currency pair is a ratio, and holding it means being long one currency and short the other at the same time.
Buy EUR/USD and you are long euros and short dollars. Sell GBP/USD and you are short pounds and long dollars. Buy AUD/USD and you are long Australian dollars and short dollars again.
Two of those three positions are short the dollar. If the dollar strengthens broadly, both lose, and they lose together, because they are the same trade wearing different labels.
This is why the pairs on a platform are a presentation choice rather than a list of independent instruments. Your real position is the net exposure to each individual currency, and the platform does not show you that anywhere.
How to actually measure it
The useful exercise takes about two minutes and is worth more than any correlation table.
Write out every open position. Split each into its two legs. Add up how much of each currency you are long and short, in a common unit of size. What comes out is your actual portfolio.
| Position | Long | Short |
|---|---|---|
| Buy EUR/USD | EUR | USD |
| Buy GBP/USD | GBP | USD |
| Buy AUD/USD | AUD | USD |
| Sell USD/JPY | JPY | USD |
| Sell USD/CAD | CAD | USD |
Five tickets, five different pairs, and every single one is short the US dollar. This is not a diversified book. It is one dollar-short position at five times the intended size, and it will move as one.
The same trader would never knowingly open a single position five times larger than their risk limit. They did it by accident, because the interface counted trades and the market counts currencies.
The two kinds of correlation
Mechanical correlation is the one above: pairs sharing a currency leg. It is arithmetic, not statistics, and it is the one that causes most retail damage because it is completely invisible in a list of tickets.
Economic correlation is the softer kind: currencies that tend to move together because their economies respond to the same conditions. Commodity-linked currencies often move with the commodities their countries export. Currencies of closely integrated economies tend to track each other. Some currencies are bought during market stress and others sold, which links them without any shared leg at all.
Economic correlation is where correlation tables earn their keep — for pairs that do not share a currency and where the relationship is statistical rather than structural.
Correlations move, and they move at the worst time
The single most dangerous assumption in this area is that a measured correlation is a property of the pairs. It is not. It is a description of a past period, and it changes when the drivers change.
Two forces move them, and neither is predictable.
Policy divergence. When two central banks move in different directions, currencies that tracked each other stop doing so, sometimes abruptly.
Stress. During a broad risk event, distinctions between assets compress and many things that were independent start moving together. A portfolio whose risk calculation depended on them being independent discovers that dependency precisely when it can least afford to.
That second one is the reason a diversification argument built on correlation should always be stress-tested by asking a simpler question: if the currency I am most exposed to moves three percent against me, what is the account down? If the answer is uncomfortable, the correlation table is not going to save you.
The hedging trap
A common response, on discovering concentration, is to hedge it with a correlated pair rather than to reduce it. This is almost always worse.
You now pay two spreads instead of one and two swap charges instead of one, permanently. The hedge is approximate because the pairs are related rather than identical, so a residual remains that can move against you — and it will do so precisely when the correlation changes, which is when you needed the hedge.
Cutting the position size achieves the same reduction in exposure, costs nothing, and cannot come apart. It is less interesting, which is most of why it is less popular. Risk management and position sizing is the framework, and how to calculate position size is the arithmetic.
Why this compounds with leverage
Correlation and leverage multiply each other, and the combination is what turns a bad afternoon into a closed account.
Leverage means a small percentage move in the price produces a large percentage move in the account. Correlation means several positions deliver that move simultaneously rather than independently. Together they mean an ordinary daily range in one currency can take an account through the margin close-out level in a way that no individual position's risk calculation predicted.
The regulatory backstops help and do not solve it: the firm must close positions when equity falls below 50% of the margin requirement, and a retail client's liability is limited to the funds in that account (FCA COBS 22.5.13R and 22.5.17R). Both cap the damage at your balance. Neither prevents it. Understanding leverage and margin covers the arithmetic properly.
The one habit worth building
Before opening any position, ask what currency exposure it adds to the book you already hold — not whether the setup looks good in isolation.
That single question converts a list of trades into a portfolio, and it is the difference between risking one percent six times and risking six percent once without noticing.
Nothing in this article names a specific correlation value, because correlations are period-dependent and change. Measure them over the period and pairs you actually trade, and re-measure them.
Frequently Asked Questions
What is currency correlation?
The tendency of two currency pairs to move together or in opposite directions. It arises mechanically when pairs share a currency — every pair quoted against the US dollar carries a dollar leg — and economically when two economies respond similarly to the same conditions.
How do I know if my positions are correlated?
Start by listing every position as its two separate currency legs and adding up the net exposure to each currency. If one currency appears on the same side of most of your trades, that currency is your actual position regardless of how many tickets are open. Correlation tables are useful afterwards, for pairs that do not share a currency.
Is correlation stable?
No, and relying on it as if it were is a common and expensive error. Correlations shift with interest rate policy, commodity prices and risk sentiment, and they tend to converge towards one during stress — which is exactly when a portfolio built on their independence needs them not to.
Does hedging with a correlated pair reduce risk?
It reduces some risk and introduces basis risk in its place: the two pairs are related but not identical, so the hedge is approximate and the residual can move against you. It also doubles the transaction costs and swap charges. Reducing position size achieves the same risk reduction more cheaply and more reliably.
