The risk that is invisible on every individual ticket
Each position shows its own risk. Nothing on your platform shows the risk you actually carry, because that depends on how the positions relate to each other.
A trader with five open positions at 1% each does not have five 1% risks. They have somewhere between 1% and 5%, depending entirely on correlation — and the platform will never tell them which.
Counting exposure by currency
Every FX position is two bets: long one currency, short another.
Position
Long
Short
Buy EURUSD
EUR
USD
Buy GBPUSD
GBP
USD
Sell USDJPY
JPY
USD
Buy XAUUSD
Gold
USD
Four different tickets. Four short-dollar bets. At 1% each, that is 4% riding on one variable, and every one of them loses on the same headline.
The fix is a second ledger. Before entering, total your existing exposure per currency:
`` USD: −4% (short in four positions) EUR: +1% GBP: +1% JPY: +1% ``
A fifth dollar-negative trade takes USD to −5%. Whether that is acceptable is a decision — but it must be a decision, not a discovery made during a Fed press conference.
Correlation strength decides how much overlap counts
Not all shared exposure is equal:
Pair combination
Typical correlation
Counts as
EURUSD + GBPUSD
+0.85
Almost one position
EURUSD + AUDUSD
+0.60
Substantially overlapping
EURUSD + EURGBP
−0.30
Partly offsetting
EURUSD + USDCAD
−0.75
Nearly opposite — genuine hedge
The practical simplification: treat anything above +0.7 as the same trade. Two positions at 1% each with 0.85 correlation is closer to a single 1.85% risk than to two independent ones.
The stress problem
Correlations strengthen under stress, and this is the part that catches experienced traders.
EURUSD and GBPUSD run around +0.85 in normal conditions. In a genuine risk-off event they approach +0.98 — everything becomes one trade at exactly the moment diversification was supposed to help.
So the honest planning assumption is: whatever correlation you measured, assume higher when it matters. A portfolio that looks diversified at +0.7 is a single position at +0.95.
Margin compounds it
Five correlated positions require five lots of margin. When they all move against you together, free margin falls five times as fast, and the broker's stop-out does not wait for your analysis to be vindicated.
This is the mechanism behind most forced liquidations: not one oversized position, but several "small" ones that were the same bet.
The three-limit structure
One limit is not enough. Three, written down:
Limit
Typical
Prevents
Per trade
1%
One bad trade
Per currency
3%
One view expressed repeatedly
Total open
5%
One bad session
The second is the one almost nobody has, and it is the one that catches the failure mode above.
What genuine diversification looks like
Positions that can win independently — meaning they depend on different variables:
EURGBP (euro versus sterling, dollar-neutral) alongside USDJPY
A gold position driven by real yields alongside a cross with no dollar leg
Different timeframes on the same instrument is *not* diversification — it is the same instrument
The test: could one of these win while the other loses, for reasons that make sense? If you cannot describe such a scenario, you hold one position in two tickets.
Worked example
$20,000 account. The trader is long EURUSD, long GBPUSD and long AUDUSD, 1% each. They believe they risk 3%, spread across three trades.
Correlations: EURUSD/GBPUSD +0.85, EURUSD/AUDUSD +0.62, GBPUSD/AUDUSD +0.58. All three are short-dollar.
Currency ledger: USD −3%. Already at the cap.
A hawkish Fed surprise sends the dollar up 1.1%:
EURUSD stops out: −$200
GBPUSD stops out: −$200
AUDUSD, higher beta, gaps through its stop: −$260
Total: −$660, or 3.3% in one release. The risk model said 3% and the outcome exceeded it, because the third stop slipped — which is what happens when everything moves at once.
Now the version with a currency cap of 3% applied before entry: the third trade is refused. Loss is $400, or 2%.
And the more important difference is what happens next. At −2% the trader continues normally. At −3.3%, with three simultaneous stop-outs on screen, they are in exactly the state the tilt lesson described — and the next trade is the dangerous one.
What to take away
Your platform shows position risk, never portfolio risk. Keep a currency ledger.
Treat correlations above +0.7 as the same trade.
Assume correlations rise toward 1 under stress, when it matters most.
Correlated positions multiply the margin drain as well as the loss.
Three limits: per trade, per currency, total open. The middle one is the one most people lack.
That completes Risk as a System. Next: turning all of it into a written, testable plan.
