Blog
Correlation Is Hidden Risk: Why Five Tech Stocks Are One Bet
Risk one percent on each of five stocks and the arithmetic says five percent. Should the five fall together, the whole five percent arrives in a single afternoon.
Risk per trade is easy to control and easy to see. Risk across trades is neither, and it is the place where careful people get hurt while following every rule they have written down for themselves, one trade at a time, and discovering at the end of a bad week that the rules never covered the thing that actually happened.
The missing rule is a short one. Positions that move together are, for practical purposes, one position.
Everyone in our house catches the same cold within about four days of each other. Five people, five separate immune systems, one household .. and the illusion of independence holds right up until somebody brings something home from school!
The arithmetic of “diversified”
Risk 1% on each of five stocks and the total open risk, if every stop is hit, comes to 5%. That figure goes by the name portfolio heat. Most frameworks set their limit near there. 5% sits at the upper end of what is usually permitted.
Now make all five large-cap technology names. On an ordinary day they wobble more or less independently of one another, and the 5% looks like an overstatement of anything that could plausibly happen. Then the sector sells off. They fall together, the stops trigger together, and the 5% arrives inside an afternoon. The independence you counted on existed only in weather that never called for it.
So there were never five positions. There was one position in large-cap technology at five times the intended size, and the five tickers were the thing that kept you from noticing it, spread as they were across five separate order confirmations, each of which looked modest enough on its own.
Worth adding: correlation rises at precisely the moment you would prefer it to fall. During a broad selloff, stocks that looked only loosely related in calm conditions begin to move as one, because the selling is a fact about the market rather than a judgement on any individual company. Diversification measured in calm weather overstates what you have in a panic.
Where correlation hides
Sector is obvious: five semiconductor names, four banks, three homebuilders.
Factor is less obvious. Growth stocks spread across unrelated industries share a sensitivity to interest rates, and they will fall together on a jump in rates whatever it is they happen to manufacture. High-dividend stocks share a sensitivity of their own. Small caps as a group move with risk appetite.
Thesis is the least obvious, and the hardest to spot in your own book. Six positions taken because rates are peaking are one bet on rates, however different the companies look when you line them up side by side on the screen and admire the variety of it. Should the thesis be wrong, all six are wrong on the same morning.
An illustration
Consider a portfolio holding a chip designer, a cloud software company, a payments processor, an online retailer and a consumer electronics maker. Five industries on paper. One factor in the market’s eyes, since every one of the five is priced largely on growth expectations, and growth expectations are the first thing to go when rates rise.
Then a rate shock arrives and all five gap down on the same morning. Planned risk was 1% each. The realised loss comes closer to 6% in a single session, and every stop did precisely what it had been designed to do, in the order it was meant to do it, at the price it was meant to do it at. The design was sound. The count was wrong.
Counting exposure properly
- Cap total open risk at a fixed figure, counting what every position loses to its stop. Between 5% and 6% of equity is common. At the cap, nothing new goes on.
- Group the positions by what actually drives them, whether that is a sector, a factor or a thesis, and then size each group as though it were a single position, with one allocation shared out between its members.
- Before adding a name, ask what would have to happen for the positions you already hold to fall hard, and then ask whether the new one falls in that same scenario. If it does, you are buying the same bet under a new name.
- Measure correlation in bad weeks. Pull the holdings up on the worst days of the past year. Look at what moved together.
The correlation and portfolio construction guide works through the arithmetic, including the reason correlations converge toward one in a crisis, and the risk management guide covers portfolio heat alongside single-trade sizing.
Frequently asked questions
What is portfolio heat?
The sum of the risk to the stop across every open position. Five positions each risking 1% carry 5% of heat, and if those positions are correlated, the whole 5% can arrive in a single session. Most risk frameworks cap total heat somewhere between 5% and 6% of equity, for exactly that reason.
How do I check for hidden correlation in my positions?
Group your holdings by what actually drives them, whether that is a sector, a shared factor such as rate sensitivity, or simply a common thesis you have expressed several ways. Give each group a single risk allocation between them. Then look at how those holdings behaved on the worst market days of the past year, because that is the correlation that will matter when it matters.