Correlation Risk: Why Three 1% Trades Are Often One 3% Trade

You are long EURUSD, long GBPUSD and long AUDUSD, each risking 1% of your account. Three setups, three instruments, three separate decisions, three stop losses. Your risk is 1% per trade — that is what your position size calculator said, and it is what your journal will record. Then the dollar catches a bid on a Fed headline and all three stops fill inside four minutes. You are down 3% on a day where you never risked more than 1%. Nothing malfunctioned. You simply took one trade three times and paid a commission for the privilege of thinking it was three.

The Arithmetic You Think You Are Doing

The 1% risk rule is sound, and every position sizing calculator implements it correctly. The problem is not the rule. The problem is that the rule governs a single position, and traders apply it position by position while quietly assuming the results will spread out.

That assumption is doing enormous work. Three genuinely independent 1% bets are a reasonable portfolio: sometimes one loses, occasionally two, rarely all three. Three copies of the same bet at 1% each is a 3% bet with extra steps. The account cannot tell the difference between "one 3% trade" and "three correlated 1% trades" — only your journal can, and only if you record what drives each position.

The uncomfortable part is that the illusion is strongest exactly where retail traders concentrate: major FX pairs (which share a dollar leg by construction), index CFDs (which track overlapping baskets of the same companies), and altcoins (which mostly express a view on bitcoin with extra volatility).

What Correlation Actually Does to Your Risk

For equally sized positions, combined risk is not the sum and it is not the average. It scales with the square root of the correlation-adjusted total:

The second column below is the more revealing one. "Effective independent positions" answers the question you actually care about — how many genuinely separate bets you are holding, as opposed to how many tickets you bought.

Read the bottom row against the top row. Both are described in your journal as "three trades at 1% risk." One of them is a diversified book. The other is a 3% directional bet that you have disguised from yourself using three ticker symbols. Your position sizing was never the problem — your position sizing was correct for every individual trade and wrong for the account.

One precision worth stating plainly: the combined figure is a volatility measure, not a promise. The literal worst case is always 3% — every stop hit — regardless of correlation. What correlation changes is not the size of the disaster but how frequently it arrives.

How Often They All Lose Together

This is where the abstraction becomes an account balance. The table below simulates three positions, each with a 60% chance of hitting its stop (a 40% win rate system), linked by a shared driver — 500,000 runs per row.

At 0.85, close to half of your trading days that carry three correlated positions end at the full −3%. You planned for that to be a rare event and priced your emotional tolerance accordingly. It is not rare. It is a coin flip, and it will arrive in clusters that feel exactly like a broken strategy — which is how correlation risk gets misdiagnosed as a strategy problem and "fixed" by replacing a system that was working.

The Clusters Most Traders Miss

Everyone knows EURUSD and GBPUSD move together. The expensive correlations are the ones that do not announce themselves in the ticker.

The shared currency leg

Every major pair is a bet on two currencies, and traders reliably notice only one of them. Long EURUSD, short USDCHF and long AUDUSD look like three different views. They are one view: short dollar, expressed three times. The tell is simple — if a single dollar-driven release can resolve all three, they are one position.

The risk-on / risk-off bloc

Equity indices, AUD and NZD, high-beta crypto and commodity currencies frequently move as a single sentiment complex. Long S&P, long AUDJPY and long bitcoin is not a diversified book — it is one leveraged bet that this week is calm.

The bitcoin beta

Most altcoins are a bitcoin position with amplified volatility. Holding four of them is holding one bitcoin position at four times the size, with worse liquidity and wider spreads on the way out. Our comparison of crypto vs forex journaling covers why this is easy to miss in a crypto journal specifically.

The setup correlation

This one is invisible in any correlation matrix, because it does not live in the instruments. If all three positions came from the same pattern, read from the same session, or entered within twenty minutes of each other, they share your judgment as a driver. When the read is wrong, it is wrong three times. Instrument correlation might be 0.3; decision correlation is 1.0.

The state correlation

The most expensive version. Positions opened in the same emotional state — after a loss, late in a frustrating session, chasing a move you missed — fail together because they were produced by the same compromised process. The market did not cluster them. You did. This is the mechanism behind most overtrading damage: it is rarely one catastrophic trade, it is five correlated mediocre ones.

Correlation Rises Exactly When It Matters

If correlation were a stable property, this would be a solved problem: measure once, size accordingly, move on. It is not stable. Correlation drifts with market regime, and it has a strong tendency to increase during stress — the well-documented pattern where relationships that held for months converge as everything reprices against a single factor at once.

For a trader this has a precise and unwelcome meaning. The diversification in your book is measured on ordinary days and consumed on extraordinary ones. On the quiet Tuesday when you do not need protection, your three positions behave semi-independently. On the release that produces your worst day of the quarter, they behave as one. Diversification evaporates on exactly the day it was supposed to pay for itself.

The practical consequence is that sizing to average correlation systematically understates the risk of your worst outcomes. Size for the correlation you get on the bad day. The same reasoning applies to weekend gap risk, where every open position is resolved by the same news vacuum simultaneously.

What This Does to a Prop Firm Account

On a personal account, correlation risk costs money. On a funded account, it ends the account.

A typical evaluation permits roughly 5% daily loss and 10% total. A trader risking 1% per position with three correlated positions open is one adverse headline away from −3% before lunch, with most of the session still ahead. Add the near-universal reflex to trade back a bad morning and the daily limit is not a distant boundary — it is one decision away.

Firms design those limits knowing this. The daily drawdown rule fails traders far more often than the profit target does, and correlated baskets are the most common mechanism. The full breakdown is in our guide to the prop firm daily drawdown, but the correlation-specific rule is short: on a funded account, the risk budget belongs to the driver, never to the ticker.

How to Size a Correlated Basket

The fix is not "never take related trades." Related trades are often where the strongest signals live — a genuine dollar view should be expressible. The fix is to stop letting the number of tickets determine the size of the bet.

• →Budget by driver, not by trade. Decide that the dollar gets 1% today. If you want three dollar-driven positions, they risk about 0.33% each. Same conviction, same exposure, three chances for one of them to be the good expression of it.
• →Treat anything above roughly 0.7 correlation as the same instrument. Not as a similar instrument — the same one. At that level you are choosing which ticker to use, not how many positions to hold.
• →Cap simultaneous themes at two or three. Beyond that you are adding exposure without adding independence, and you cannot monitor the book properly anyway.
• →Size the basket so the all-stops-hit case is survivable and boring. If every position in the cluster hitting its stop produces a number that would change how you trade tomorrow, the basket is too large regardless of what the correlation coefficient says today.
• →Check correlation before adding, not after. The second position in a cluster is where the decision is made. By the third, you are managing a mistake rather than making a choice.

The One Question That Replaces All of This

You do not need a correlation matrix at the moment of entry. You need one question, asked before the second position of any group: is there a single event that resolves all of these the same way?

If a Fed decision, a bitcoin move, a risk-off session or one earnings release settles every open position at once, you are holding one trade. Size it as one trade. If no single event connects them, you have genuine diversification and can size each one on its own merits.

That question takes five seconds and catches the overwhelming majority of correlation damage in a retail account. The traders who blow evaluations rarely do it by risking 5% on one position — they almost never take a position that size. They do it by risking 1% five times on what was, in every way that mattered, the same idea.