Correlation and Portfolio Heat: Why Six Positions Can Be One

A trader holding six positions at 1% risk each will usually describe their risk as “1% per trade”. The honest description is that they have 6% at stake if everything goes wrong at once.

Whether that can happen depends entirely on correlation, and correlation is the variable most retail risk management ignores completely.

Portfolio heat

Portfolio heat is the sum of risk across every open position: what you lose if all stops are hit.

Positions	Risk each	Total heat
3	1%	3%
6	1%	6%
10	1%	10%
6	2%	12%

Most professional frameworks cap heat between 5% and 6%, regardless of what per-trade sizing suggests. Once you reach the cap you take no new positions until something closes. That constrains opportunity, which is the point.

Why heat understates the real risk

That table assumes positions are independent. They rarely are.

Suppose your six positions are six semiconductor names. Six tickers, six stops, 6% of heat. But these are one bet: the semiconductor cycle. Bad guidance from any of them moves all six the same day, and a sector rotation moves all six harder.

Your effective risk is closer to 6% on a single outcome, with six sets of commission and six chances to be stopped out by something unrelated to the thesis.

The reverse error exists too. A utility, a gold miner, a biotech and a regional bank carry genuinely different drivers, and 4% spread across those is materially less risky than 4% concentrated in one.

Count risk drivers, not tickers

Stop counting positions and start counting drivers. Sector is the most obvious shared driver. Factor is the next: growth versus value, large versus small cap, momentum versus quality, and positions in different sectors can still be one factor bet. Then macro sensitivity, since rate-sensitive names such as utilities, REITs and long-duration growth move together regardless of sector. And single inputs: an oil price, the dollar, one customer, one regulatory decision.

Then cap exposure per driver rather than per position:

Limit	Cap
Risk per position	1%
Risk per sector	2%
Risk per factor or macro driver	3%
Total portfolio heat	6%

Under those rules the six-semiconductor portfolio is not permitted. You hold two, at 1% each, and look elsewhere for the rest.

The correlation problem nobody solves

Correlations are not stable, and they fail in one direction: they converge towards one during broad liquidations.

In normal conditions a utility and a biotech have little relationship. In a genuine market-wide selloff both fall, because participants facing margin calls and redemptions sell what is liquid rather than what they would prefer to sell. Correlation measured on calm data understates risk in exactly the conditions where you needed it to hold.

Sizing when positions relate

Two adjustments handle most of it.

Treat correlated positions as fractions of one. If you want semiconductor exposure across three names, size each at a third of a standalone position. Same total sector risk, some idiosyncratic spread.

Cut total heat when breadth narrows. When market internals show a narrowing advance, realised correlations are usually rising. Dropping the heat cap from 6% to 3% in those conditions costs some upside and removes a large share of the tail.

Work the resulting per-position risk into share counts with the position size calculator. The arithmetic is unchanged; only the risk percentage moves.

A worked comparison

Two traders, both “risking 1% per trade”, both holding six positions.

Trader A holds six semiconductors. A sector downgrade hits and all six gap through their stops. Realised loss: roughly 8% after slippage, from a portfolio they believed was spread across six independent ideas.

Trader B holds one semiconductor, one utility, one energy, one healthcare, one industrial and one consumer staple. The same downgrade hits one position. Realised loss: about 1.3%.

Same rule, same position count, same per-trade risk. The difference is entirely structural, and it is invisible on any per-trade risk report.

What this does to expectancy

Lower variance does not raise expected return. It raises the probability that you are still trading when the edge pays off.

A strategy with +0.2R expectancy and uncontrolled correlation produces drawdowns most people abandon before the edge compounds. The same strategy with driver limits produces a shallower, slower curve that is survivable, and survivable is the only kind of edge that ever gets realised. See expectancy and sample size for why the number of trades you complete is the binding constraint.