Correlation risk is what happens when several open positions that look diversified are actually driven by the same underlying factor, so they win or lose together instead of independently. A strategy running EURUSD, GBPUSD, and AUDUSD at once looks like three separate bets - but if all three are really a bet on the US dollar weakening, a single broad USD move can push all three the same direction on the same day. The account then carries far more risk than "three 1% trades" suggests, because there's no diversification actually happening - it's one large position wearing three different symbol names. This isn't about avoiding multiple positions; it's about knowing when multiple positions are genuinely independent versus when they're the same underlying trade counted three times. The fix isn't complicated once it's visible: size correlated positions as a group, not individually, and check the relationship between instruments before adding a new one to a strategy that already trades something related.
What correlation actually measures
Correlation measures how closely two instruments' price movements track each other, on a scale from -1 (they move in exact opposite directions) through 0 (no relationship) to +1 (they move in exact lockstep). It isn't fixed forever - the relationship between two instruments can strengthen, weaken, or even flip depending on what's driving the market at the time - which is exactly why "these two pairs are correlated" needs periodic checking rather than being assumed once and forgotten.
Why "different" pairs often aren't independent
Most major currency pairs share a currency with several others, and that shared currency's own strength or weakness can dominate the move. EURUSD and GBPUSD both have the US dollar on one side, so a broad dollar move tends to push both the same direction at once - not because EUR and GBP are related to each other, but because both are being pushed by the same third factor. The same pattern shows up beyond forex: commodity-linked currencies moving with the commodities they're linked to, or index instruments moving with the sectors that dominate them. The instruments look different on the watchlist; the underlying driver is often the same one.
The portfolio math: what correlation does to combined risk
For three equally-sized positions each risking 1% of the account, the combined risk depends entirely on how correlated they are with each other - not on the fact that there are three of them:
| Correlation between the 3 positions | Combined risk (3 × 1% each) |
|---|---|
| 0 (fully independent) | ≈ 1.73% (√3 × 1%) |
| 0.5 (moderately correlated) | ≈ 2.45% |
| 1.0 (fully correlated) | 3% - no diversification benefit at all |
Independent positions get a real diversification discount - the combined risk is less than simply adding the three together, because they don't all go wrong at once. Fully correlated positions get none of that discount: the "3 separate 1% trades" framing was never true risk-wise: it was always one 3%-risk position, just split across three tickets.
What this looks like in practice
A strategy that opens a long position on EURUSD, GBPUSD, and AUDUSD whenever its signal fires isn't necessarily wrong to do so - but if the signal is really reacting to broad USD weakness, it's placing one trade three times, not three trades. On a day the dollar reverses instead of weakening further, all three positions can hit their stops within minutes of each other, and the "three small losses" the position sizing was built around arrive as one large one.
Managing it, instead of discovering it live
Correlated positions should be sized as a group, the same way a single trade's risk gets capped - if two or three instruments in a strategy are known to move together, the combined risk across all of them should stay within the limit normally applied to one trade, not be multiplied by however many symbols happen to be involved. See position sizing for how the single-trade version of that cap works, and why martingale and grid trading blows up for a different, more direct way position sizing gets undermined. In a multi-rule, multi-asset builder, this means treating a strategy that trades several correlated instruments as one combined risk budget when setting position size - not as several unrelated trades that each get the full per-trade allowance.
Common questions
What is correlation risk in trading?
Correlation risk is the risk of treating several open positions as independent bets when the instruments actually tend to move together. If three "different" trades are all driven by the same underlying factor - like broad USD strength - they can win or lose together, which means the real risk on the account is closer to one large position than three separate small ones.
Which currency pairs are commonly correlated?
Pairs that share a currency on the same side often move together - EURUSD and GBPUSD both tend to rise when the US dollar weakens broadly, for example. Pairs on opposite sides of a shared currency, like EURUSD and USDCHF, often move in opposite directions instead. The exact relationship shifts over time and isn't fixed, which is exactly why it needs checking rather than assuming.
How much does correlation actually increase risk?
With three equally-sized positions each risking 1% of the account, fully independent positions combine to roughly 1.73% total risk (the square root of 3), while fully correlated positions combine to a full 3% - no diversification benefit at all. The exact number depends on the real correlation between the specific instruments, but the direction is always the same: higher correlation means less risk reduction from holding multiple positions.
How do I manage correlation risk in an automated strategy?
Treat correlated positions as a group for sizing purposes rather than sizing each one independently - cap the combined risk across correlated instruments the same way a single trade's risk gets capped, and check correlation before adding a new instrument to a strategy that already trades a related one.