Risk Academy

Correlation: One Trade in Five Costumes

Ask a trader with five open positions what happens if the Fed surprises, and watch the face change. Long NQ, long ES, long BTC, short DXY, long gold — that's not a portfolio. On a macro headline it's one position at five-times size, and it eats a daily loss limit in the time it takes to read the alert.

The one-driver problem

Instruments look different on the screen — different tickers, different charts, different tabs. Underneath, most of what retail traders hold is priced off a handful of drivers: the dollar, rates, and risk appetite. When one of those moves hard, your “separate” positions vote together. And the cruel part: correlations tighten in stress — the day you most need diversification is the day it disappears.

Long NQLong ESLong BTC“three trades, three markets, nicely diversified”one driverrisk sentimentone headline → −3R in one minuteCorrelations converge exactly when it hurts: in risk-off, “different” markets become one market.
The costume test: if one headline moves all your positions the same direction, you don't have several trades — you have one trade wearing several costumes.

The math on a funded account

You risk 1R = $500 per position, daily loss limit $1,500. Three “independent” longs: NQ, ES, BTC. Feels like three separate 1R bets with 3R of room.

A hot CPI print hits. All three gap the same direction, all three stops fill late (news slippage) — −$1,650 in ninety seconds. The daily limit is gone in one candle, on what your journal will record as “three losing trades”.

It was one losing trade. You just paid for it three times — and the account paid with its life.

Know your clusters

ClusterTravels togetherThe real driver
US indicesES / NQ / YM / RTY (+ DAX in sympathy)Rates & risk sentiment
USD blocEURUSD, GBPUSD, AUDUSD (inverse: DXY)The dollar
MetalsGold, silver (XAU, XAG, GC, SI)Real yields & USD
Crypto majorsBTC, ETH, SOL — and most alts, amplifiedBTC itself / risk appetite
Oil complexCL / Brent, USDCAD (inverse), energy stocksCrude

Correlations aren't constant — they drift in quiet markets and snap to 1 in stressed ones. Treat the table as the stress-day map, because that's the day that decides your account.

Where traders fall

  • They count tickets, not themes: five positions, one idea (usually 'risk on') — sized as if the idea were five ideas.
  • They add a 'hedge' that isn't one: long NQ + long gold often both bleed on a rates shock. A hedge you haven't seen work in stress is a hope.
  • They scale into the same cluster as confirmation arrives — NQ works, so add ES, then YM. That's averaging up on one trade while calling it three.
  • They diversify across accounts instead of across drivers — two funded accounts long the same index is the same trade with two breach letters.

Risk per theme, not per ticket

  • Before adding a position, ask the costume question: 'if one headline moves everything I hold, which direction am I?' If the answer is 'all the same way', it's one theme.
  • Cap total open risk per cluster at ~1R — split it across tickets if you like, but the cluster budget is the budget.
  • Run the stress test nightly: 'Fed surprise / CPI shock — what's my combined hit with news slippage?' If it exceeds the daily loss limit, you're pre-breached and don't know it.
  • On funded accounts, remember the DLL doesn't care how many journals entries the loss is split across — it sums.

The rules this collides with

Educational content only — not financial advice.