Correlation risk has its own little cluster of threads now: prop-focused, gold-focused, yen-focused, indices, and a couple on netting your currency exposure. They all start from the same observation and then go in useful different directions, so here's the combined version.
The core problem
The prop correlation thread puts it in one line: long EURUSD, long GBPUSD and long AUDUSD at the same time is mostly one trade, short USD. The mentee notes thread uses the same example with gold swapped in, and turns it into a sticky-note rule: one USD idea per hour. The point both make is that correlation rises in stressed markets, so pairs that look loosely related on a quiet week can move almost identically after a surprise US release, which is exactly when stops cluster.
How to see it
The netting thread gives the simplest tool. Write each currency on a line and add +1 or -1 for each trade. Long EURUSD, short USDJPY, long GBPUSD comes out as EUR +1, USD -3, GBP +1, JPY +1. The biggest number is your real position. It takes seconds and it's surprisingly sobering.
The weekly audit thread went further and the replies turned into a code thread, with MT4 and cTrader versions of a small dashboard that measures rolling correlation between your theme symbols and warns above a threshold. A reply there sums up the mindset: treat the market as themes rather than individual tickets.
Specialist edge cases
- Gold. The gold specialists thread describes the slow creep: gold, then silver, then a miners index, then AUDUSD because it's "a gold currency", and within a month it's one big position on four tickets. Gold and silver count as one risk during London hours.
- Yen. The yen thread uses a combined cap: USDJPY plus any yen cross can't exceed 1.5R. On risk-off days almost every yen pair moves together, so the cap tightens.
- Indices. The US30 thread noticed holding NAS100 alongside US30 in the same hour, and less obviously USDJPY, which on afternoons when US yields are moving can travel with US indices.
The several-accounts thread makes a point people avoid. Copying the same strategy across three evaluations isn't diversification. If the trade loses, it loses on every account at once. A reply there suggests a worst-case check: take your largest daily loss from your journal and multiply it by the number of accounts. Another flags that different firms can reset daily loss at different hours, so a trade fine on one account can breach another.
Where the threads differ
Mostly on the size of the cap. One USD idea per hour is simple and blunt. A 1.5R combined cap allows some overlap. A rolling-correlation dashboard is precise but depends on the lookback, and correlation measured over calm weeks will understate how things move on a bad day. My lean is the blunt rule for discretionary scalping, because it works in the exact moment you're most likely to talk yourself out of it. The dashboards make more sense for EA portfolios.
Either way, the starting point is the same: before adding a trade, ask what currency or theme it's really a bet on, and whether you already own that bet.
What cap do you use for correlated trades, and is it per hour, per session or always on?
Threads this pulls from
- Correlation Risk on Prop Accounts: When Three Trades Are Really One
- Running one strategy across several prop accounts, the correlation risk people ignore
- Correlation caps for prop candidates: weekly audit
- Correlation caps for gold specialists: beginner mistakes
- Correlation caps for yen specialists: advanced edge cases
- Correlation caps for discretionary scalpers: mentee teaching notes
- Correlation caution: US30 plus related risk in same hour
- Currency Exposure Netting: Seeing What You Really Own
