BUILD: COURSE 2 | LESSON 5
Correlation and portfolio-managed risk
Learning objectives
Explain how correlated positions multiply real risk while each trade looks individually well-sized.
Identify the major structural correlations in FX and CFDs (shared currencies, USD blocs, risk-on/risk-off, gold–dollar) without needing a correlation matrix from memory.
Apply a portfolio risk cap that counts correlated trades as one exposure, and audit open positions by shared driver.
You now size every trade to risk 1% (P2.2), you know your survivable drawdown (P2.3), and you manage exits by fixed rules (P2.4). Here is how traders who do all of that still blow through their risk limits: five open trades, each a textbook 1%, that are secretly the same trade. When the shared driver moves against them, all five stops hit in the same hour — a planned-for −1% day becomes a −5% day, the kind P2.3 told you to make impossible. Portfolio risk is the last layer of this course because it sits on top of everything else, and because it is invisible if you only ever look at one chart at a time.
The classic double-exposure: EUR/USD + GBP/USD
EUR/USD and GBP/USD typically show a strong positive correlation — often in the +0.70 to +0.90 range on daily returns — because both pairs have the same currency on the same side: they are largely two prices for "the dollar, inverted". (Correlation runs from +1, moving in lockstep, to −1, moving exactly opposite; 0 means unrelated. It is measured on returns over a window, commonly 30 trading days, and it changes over time.)
Now the worked example. Account $10,000, risk 1% per trade:
- Long EUR/USD, 72-pip stop, sized to risk $100.
- Long GBP/USD, 90-pip stop, sized to risk $100.
On paper: two trades, 2% total risk, diversified across two pairs. In practice, with correlation ≈ +0.85, the scenario that stops one out — broad USD strength, say a hot US inflation print — almost certainly stops out both. Your realistic loss cluster is −$200 from a single event: you are effectively running one dollar-short position at double size. Stack on a long AUD/USD and short USD/JPY (also both dollar-short) and your "4 × 1%" portfolio is close to one 4% bet against the US dollar. Correlation also works at −1 in disguise: long EUR/USD plus short USD/CHF is the same double-exposure, because the negative correlation times the opposite direction makes the positions reinforce, not offset. What matters is not the sign of the correlation but sign × your trade directions.
Where correlation hides: think in drivers, not tickers
You do not need to memorise a 28×28 matrix. Almost every large co-movement in a retail CFD portfolio comes from a handful of structural drivers. Audit your open trades against this list:
- Shared currency legs. Any two pairs sharing a currency are mechanically linked. Long EUR/USD + long EUR/JPY + short USD/CHF = triple euro-and-anti-dollar exposure. Count legs: sum your net exposure per currency, not per pair.
- The USD bloc. Most majors are dominated by the dollar side. Four majors traded "independently" are usually one DXY view. (Cross pairs — EUR/GBP, AUD/NZD — are the tool for expressing a view without the dollar.)
- Risk-on / risk-off. In stress episodes, correlations lurch toward ±1: AUD, NZD, equity indices and copper fall together while JPY, CHF and often USD rise. Long AUD/JPY + long US500 + short gold can behave as one "risk-on" trade precisely when it hurts most. Diversification is weakest on the worst days — plan with crisis correlations, not calm ones.
- Commodity linkages. Gold and EUR/USD often correlate positively (both anti-dollar); oil ties to CAD; AUD to metals and China data. XAU/USD long + EUR/USD long is a milder version of the GBP example above.
- Same instrument, multiple timeframes. Two EUR/USD setups (an H4 swing and an M15 intraday) in the same direction are correlated at +1.0. Obvious when stated; common in journals anyway.
Because correlations drift, check a live matrix rather than trusting folklore — the Currency Correlation Matrix in our Tools section shows rolling 30-day values, colour-coded. Anything above |0.7| deserves treatment as linked; |0.5–0.7| as partially linked.
The portfolio risk cap, upgraded
P2.2 gave the raw cap: total open risk ≤ 4–6% of equity. The upgrade is to count correlated risk as one bucket:
- Group open (and pending) trades by dominant driver: net-USD, per-currency legs, risk-on/off, gold-bloc.
- Sum risk within each group, treating |correlation| > 0.7 with reinforcing directions as 100% overlapping (conservative and simple — proper portfolio maths would scale by correlation, but at 2–5 positions the conservative rule is both safer and actually usable mid-session).
- Apply two limits: ≤ 2% per correlated group, ≤ 6% across all groups.
Worked audit. Open positions, each risking 1%: long EUR/USD, long GBP/USD, short USD/JPY, long XAU/USD, long AUD/NZD.
- Anti-USD group: EUR/USD + GBP/USD + short USD/JPY + (partially) gold = 3.5–4% in one bucket → breaches the 2% group cap.
- AUD/NZD: a genuine cross with low dollar loading → its own 1% bucket. Fine.
- Fixes, in preference order: (a) don't take the third and fourth anti-dollar trade — pick the best setup in the group; (b) halve size on each so the group sums to 2%; (c) express one idea via a cross instead (e.g. keep the euro view as EUR/GBP if the pound leg was the weaker conviction).
Note what this rule quietly enforces: when several charts "all look great" in the same direction, that is usually one macro move photographed from different angles — the cap converts that observation from a temptation into a selection discipline.
Course wrap: the four numbers that run your trading
This completes Risk & Trade Management. Your risk framework is now four written numbers plus one habit: expectancy you measure (P2.1), risk per trade set by ATR-sized stops (P2.2), drawdown circuit-breakers (P2.3), one frozen exit scheme (P2.4), and a 2%-per-group / 6%-total portfolio cap with a driver audit before every new position (this lesson). Course P3 puts an engine inside this chassis: strategies with explicit rules whose expectancy you can actually test — on demo, where the tuition is free.
Key takeaways
Correlation makes separately-sized trades share the same loss event: long EUR/USD + long GBP/USD at +0.85 correlation is close to one double-sized dollar-short bet.
What matters is correlation × direction: short USD/CHF reinforces long EUR/USD despite the negative correlation sign.
Think in drivers, not tickers: net currency legs, the USD bloc, risk-on/off, and commodity blocs explain most hidden overlap — and correlations tighten toward ±1 exactly in crisis conditions.
Cap risk per correlated group (≈2%) and across the book (≈6%), treating |ρ| > 0.7 reinforcing positions as one exposure; prefer taking only the best setup in a group.
Correlations drift — verify on a rolling 30-day matrix before assuming either overlap or diversification.