You size every position at 1% risk. You never have more than five trades open. On paper, your worst-case day is a clean, bounded number. Then one afternoon four of those five trades stop out within minutes of each other, and your "5% max risk" day turns into a 5% loss that felt like one giant trade — because it was. That's correlation risk, and it's one of the most common ways disciplined position sizing quietly stops working.
What Is Correlation Risk?
Correlation risk is the danger that positions you treat as separate bets actually move together, so your real exposure to any single event is far larger than your position sizing implies. It has nothing to do with how many trades you have open and everything to do with how independent those trades actually are.
Two trades are correlated when the same underlying driver — a sector, a currency, an index, a single piece of news — can move both of them in the same direction at the same time. Long Apple and long Microsoft. Long EUR/USD and short USD/JPY. Long gold and long silver. Long three different tech growth stocks. Each pair looks like diversification on a position list. Each pair is, in a real risk sense, closer to one trade sized twice.
Position sizing controls how much you risk per trade. Correlation controls how much of that risk actually overlaps. Ignore the second one and the first one is just a comforting number.
Why It's So Easy to Miss
Correlation risk hides because it doesn't show up anywhere in a normal trade log. Your journal shows five entries, five stops, five position sizes — each individually well within your risk rules. Nothing about the list itself signals that four of those five are effectively the same trade wearing different tickers.
- It only becomes visible when several correlated positions move against you at once — which is exactly when you can least afford the surprise
- It changes over time: two assets can be loosely correlated for months and then move in lockstep during a specific news event or regime shift
- It compounds across asset classes traders don't think to compare — a stock index position and a currency position can share the same macro driver
- Standard position size calculators size one trade at a time and have no concept of what else is already open
A Concrete Example
Say your rule is 1% account risk per trade, max five open positions, so your theoretical worst case is 5%. Now say all five of today's trades are long: two mega-cap tech names, one semiconductor stock, one Nasdaq index ETF, and one long call option on a tech ETF.
A single piece of bad news in the tech sector doesn't hit one of those five positions — it hits all five, in the same direction, in the same minutes. Your stops don't fail independently; they fail together. The realistic loss on that news event isn't "up to 5% if things go badly across five unrelated setups" — it's "close to 5%, in one move, because there was only ever one real bet." The position sizing rule did exactly what it was designed to do. It just wasn't sizing the thing that actually mattered.
How to Measure Correlation Between Your Positions
You don't need a statistics degree to catch most correlation risk — you need a habit of asking one question before adding a new position: "if the thesis behind my other open trades is wrong, does this one lose too?"
| Check | What It Catches |
|---|---|
| Same sector or industry | Two or more positions that rise and fall with one sector's news flow |
| Same index membership | Individual stocks that are large, index-moving components of an ETF you also hold |
| Shared macro driver | A currency pair and a commodity or index that both react to the same rate or risk-sentiment move |
| Same catalyst or earnings date | Positions that all face the same scheduled news event in the same session |
| Directional bias across the book | Every open position pointing the same way (all long, all short) with no hedge |
For traders who want a number rather than a judgment call, a rolling correlation coefficient between two instruments' daily returns (available in most charting platforms) quantifies the relationship: values above roughly 0.7 or below roughly -0.7 mean the two move together closely enough to be treated as one position for sizing purposes.
Portfolio Heat: The Metric That Actually Reflects Your Risk
This is why experienced risk managers track portfolio heat — total open risk across all positions — rather than just per-trade risk in isolation. Portfolio heat on its own still undercounts correlated risk unless you go one step further and group correlated positions before summing their risk, treating a cluster of correlated trades as a single combined position for the purpose of your max-risk rule.
A simple adjustment: if two open positions have a correlation above roughly 0.7, sum their individual risk and cap that combined number at your normal single-trade maximum — not at double it.
Practical Rules to Cap Correlation Risk
- 1Before adding a new position, scan open trades for shared sector, shared index, or shared macro driver
- 2Treat a cluster of correlated positions as one position for total risk purposes, not as separate 1% bets
- 3Cap how many positions you'll hold in the same sector or asset class at once, independent of your per-trade risk rule
- 4Watch for correlations that tighten during high-volatility events — assets that normally diverge often move together hardest exactly when it hurts most
- 5Log the sector and primary driver as a tag on every trade so your journal can surface clustering after the fact, not just per-trade stats
Correlation Risk Isn't Just About Stocks
The same trap catches forex and crypto traders just as often. Currency pairs sharing a base or quote currency move together by construction — long EUR/USD and long GBP/USD are both, in large part, a bet against the dollar. Crypto is worse: outside of a handful of idiosyncratic tokens, most altcoins are highly correlated to Bitcoin, so a portfolio of ten "different" coins can behave like one leveraged Bitcoin position during a broad market move.
Why This Belongs in Your Trading Journal
Correlation risk is nearly invisible trade-by-trade and completely obvious in hindsight — which makes it a perfect candidate for journal review rather than in-the-moment judgment. If you tag every trade with sector and primary driver, a few months of data will show you exactly which clusters have hit you together before, and how often "diversified" days turned into single concentrated losses.
TradeEdge lets you tag every trade by sector, instrument, and setup, so patterns like correlated losing days show up clearly in your stats instead of hiding inside a list of individually reasonable trades.
Position sizing answers "how much am I risking on this trade?" Correlation review answers the question that actually determines your worst realistic day: "how much am I risking on this idea, across every trade that shares it?" Traders who only ask the first question are usually more exposed than they think — right up until the day every position moves at once and shows them the real number.