You size each trade at a careful 1% and feel comfortably spread across three positions. But if all three ride the same driver, you don’t hold three 1% bets — you hold one 3% bet wearing three different tickers. That’s correlation risk, and it’s how textbook sizing still empties an account.
The trades look independent on your order ladder. Your P&L knows better.
What correlation does to your true risk
Correlation measures how tightly two instruments move together, from +1 (lockstep) to −1 (perfect mirror). Position sizing quietly assumes each trade is a separate roll of the dice. Correlation breaks that assumption: when instruments move together, their risks add instead of averaging out.
Three long positions that are highly positively correlated behave, on a bad day, like one position at triple the size. The stop-outs land together, the drawdown arrives all at once, and the “diversification” you thought you had evaporates at the exact moment you needed it. True risk isn’t the sum of your position sizes — it’s the sum of your correlated exposure.
Positive and negative correlation across pairs and indices
You don’t need a stats package to see the obvious clusters. Ask what driver each position depends on:
- Positive correlation (risk stacks). EUR/USD, GBP/USD and AUD/USD are each partly a bet against the US dollar; long all three is one big anti-dollar position. In futures, ES, NQ, YM and RTY are all the same risk-on equity trade.
- Negative correlation (risk offsets — sometimes). A pair and its inverse, or an equity index against a safe haven, can partly cancel; long ES and long gold may hedge on a calm day. But in a genuine panic “everything correlates to 1” — the relationships you leaned on snap together and the hedge stops hedging.
- Shared macro drivers. Anything sensitive to the same rate decision, the same CPI print, or the same risk sentiment is correlated for your purposes, even across asset classes.
The practical move: group your open and candidate trades by driver, not by instrument name, and treat each group as a single exposure.
The hidden concentration in ES and NQ together
Futures traders meet this constantly with ES and NQ. They feel like two markets — one broad index, one tech-heavy — so holding both reads as diversification. In reality they’re deeply correlated: on most sessions they rise and fall together, and a sharp risk-off move takes both down at once.
Long ES and long NQ at “1% each” is closer to a 2% directional bet on US equities than to two independent trades. The same trap hides in holding a stock-index future beside a long position in a correlated single name, or two currency pairs that share a common leg. Count correlated futures as one position for risk purposes, and size the cluster, not each leg.
Adjusting size when positions move as one
Once you’ve grouped by driver, size the group as though it were one trade. Two workable approaches:
- Budget the cluster. Decide the total risk you’ll allow the whole correlated group, then divide it across the legs — three anti-dollar longs share one dollar-risk budget instead of each drawing a full allocation.
- Discount for concentration. If you insist on multiple correlated legs, cut each leg’s size so the combined worst case equals what a single full-size position would have risked.
A position size calculator turns whichever budget you pick into exact contracts or lots per leg. The hard part isn’t the arithmetic — it’s honestly admitting the three tickers are one bet before you total the risk.
Setting a correlated-cluster risk cap
Write this into a rule so you’re not eyeballing it live: no single correlated cluster may exceed a set fraction of your total risk budget. One driver — the dollar, US equities, oil, rates — gets one allocation, however many instruments you use to express it.
On a prop-firm account this cap guards a subtle breach path: several “small” correlated trades hitting their stops together can trip a daily loss limit that no single trade could reach. The limits themselves vary by firm — confirm them with yours — but the defense is identical everywhere: cap exposure per driver, not just per ticker.
Monitoring correlation live, not from a static table
Correlations aren’t fixed. Two instruments that drift apart for months can lock together the second a macro event dominates, and a correlation table you saved last quarter won’t warn you. The pairs that hurt are the ones that become correlated right when volatility spikes.
So this is a live-monitoring problem, not a lookup. Shibiki auto-journals every fill across your connected accounts, so it can measure your true aggregate exposure — including the same driver held on more than one account at once, which no single platform shows you. Connect your book through the Tradovate or MT5 integration and open risk is read as one picture rather than per-terminal. And because risk ceilings can be enforced as hard limits at the broker, a cluster cap you set actually holds when everything starts moving together — the moment good intentions usually fail.
Related: Position size calculator · Tradovate integration · MT5 integration