Correlation: five positions, two bets: Risk Management

Why this matters

EUR/USD and GBP/USD long are mostly the same dollar bet twice. Sizing each at 1% does not make the day a 1% day, it makes it a 1.7% day.

Correlation is not binary and it is not stable, but between majors it is normally high enough to treat as overlap. During a dollar flush, everything moves together.

The exposure cap is the fix: a total open risk number that counts correlated positions as less than separate but more than one.

Correlation is direction, not ticker

What matters is the underlying direction word, not the symbol. EUR/USD long, GBP/USD long and AUD/USD long are one dollar-short bet wearing three names.

Write the direction word next to each position. Anti-correlation counts too: EUR/USD long against USD/CHF long is still both short dollar, so nothing is hedged.

Gold with silver and indices with indices move together the same way. Cross-asset correlation is high enough to matter in a risk flush.

The effective risk figure

Correlated overlap means two 1% positions behave close to one 1.7% position, not two independent risks. Three behave as about 2.4%, five as about 3.8%; the count flatters the risk by more than double.

Overlap is never one and never zero, so a factor of around 0.7 is a fair working estimate. Effective risk is the number to compare against the cap, never the position count.

If a new position shares the direction word of an existing one, treat it as an add, not as diversification.

The exposure cap

Set a total open risk cap, commonly 2 to 2.5% of equity, and hold it whatever the direction mix looks like.

Once the cap is full nothing new opens until something closes, and no live correlation figures are needed to enforce it. On a normal account 2.5% at 1% risk already allows only two and a half positions, which feels strict, and that strictness is the point.

Reduce an existing position or wait, rather than stacking the cap and calling it diversification.

Netting the book in four steps

Label each open position with one direction word. Count each direction and net them: two longs against one short is a net one long.

Compute effective risk on the net figure with the overlap factor, so overlap is counted rather than ignored, then compare that number with the cap.

A net of one long sitting on four dollar-long positions is still a four position stack, so the direction word alone is not enough. Reduce the largest position first when the cap is breached.

Sector and asset groups

Correlation is not only currency. Group positions by asset class, then apply the same netting and cap logic inside each group. Two metals longs plus two index longs is two separate stacks, not diversification.

Cross-group correlation rises in a risk flush, when everything sells together, so the total cap matters most then. Around a major release, cut the aggregate across every group, not just inside one.

Keep a fixed list of which groups move together so grouping is not re-derived each session.

News collapses correlation

Around a high impact release correlation jumps toward one and separate positions become the same bet for the release window.

Before a central bank decision, flatten the stack or halve the aggregate size. Holding a full stack through a release is one large position with several stop losses, and the combined loss lands at once.

Check the calendar before the third position, not after the fifth. Simulated capital makes flattening free.

Three positions, walked through

An existing EUR/USD long and GBP/USD long are both dollar short at 1% each. Effective risk is 1.7% against a 2.5% cap, so 0.8% of room remains.

A gold long is offered; gold is dollar short too, so it joins the stack. Opening it at a full 1% would take effective risk to 2.4% and leave 0.1% of room.

The correct action is to open gold at 0.6% instead, taking effective risk to 2.1% and keeping 0.4% for a genuine third bet. The cap decides the shape of the book, not impatience.

Worked example

$100,000 account, 1% risk per correlated long, $2,500 total open risk cap (2.5%). Effective risk uses a 0.7 correlation overlap factor.

Risk of one long	$1,000
Effective risk with 2 correlated longs	1.70%
Effective risk with 3 correlated longs	2.40%
Effective risk with 5 correlated longs	3.80%
Positions allowed by the 2.5% cap at 1%	2.5
Effective risk with 3 longs at 0.8% instead	1.92%
Money at risk on that 3 position stack	$1,680
Room left under the cap after the stack	$820

Five separate 1% longs read as 5% to the eye and behave as about 3.8%. The cap says only two and a half fit.

Common mistakes

Counting positions as independent risk	Five correlated longs feel like five small trades but behave as one 3.8% bet, four times a normal trade.
Using correlation as a reason to stack	Low correlation between EUR and JPY does not help if the first four positions are already the same dollar trade. Four dollar-short positions are an effective 3.1% bet, not 4% of diversification.
Ignoring the calendar	A full stack held through a central bank release is a single large position with five stop losses. Five 1% stops firing together is a 5% day, the whole daily limit.
Hedging and calling it flat	EUR/USD long and USD/CHF long are still both short dollar, so nothing is neutralised. A 2% dollar move against both loses 2% on each leg for 4% total.

Checklist

State the direction of each open position in one word: dollar long or short.
Sum effective risk, not position count.
Keep total open risk under the 2.5% cap.
Before a fifth position, check the economic calendar.
Around a high impact release, halve the stack or flatten it.
If two positions hedge each other, reduce both rather than add a third.
Never open a fifth correlated position just because the setup is clean.

Key terms

correlation: How closely two instruments move together.
effective risk: Risk after adjusting for overlap between correlated positions.
exposure cap: The maximum total open risk allowed at once.
stack: Several open positions in the same effective direction.

Trading involves risk. Educational only, not advice. Mark it complete to bank progress.

Correlation: five positions, two bets