Portfolio Correlation: Why Five Positions Are Often Just One
Five open positions are not automatically five risks. If all five hang off the same theme, they behave on a bad day like a single position at five times the size. The number that matters is therefore not your risk per trade but your total open risk across every position.
What does correlation mean for a trading account?
Correlation describes how strongly two prices move together. For your account, though, only one part of that question is interesting: whether your positions move together on the handful of genuinely bad days.
That distinction matters. Across quiet months two instruments may well go their own way. In precisely the phases where everything gets sold, they still move in parallel. So the diversification you observe in normal conditions is not the diversification that counts.
In practice you therefore do not need a correlation coefficient. The question is enough: if the worst news for my main theme arrived today, how many of my positions would be affected?
Why is the diversification often only apparent?
Because positions differ by name but not by driver. Three names from one industry, two commodity producers selling into the same market, or five crypto assets with different tickers each share the same trigger.
On top of that sits an effect produced by the setup itself. Screening on a single criterion, say fresh breakouts, typically surfaces candidates on any given day that are all breaking out for the same underlying reason. The screen does not deliver five independent ideas, it delivers five expressions of one idea. How reliable such signals are in the first place is covered in setting stop losses correctly.
The arithmetic behind it is simple. A trader who caps risk per trade cleanly at one percent and holds five positions believes they are risking one percent. If all five move against them together, that is five percent in a day. Both numbers are correct, but only the second describes what can actually happen.
Why is the effect stronger in crypto?
Because a single theme explains a larger share of the movement there. In strong downward phases the market falls very broadly, regardless of how different the individual projects are in substance.
Two structural amplifiers come on top. First leverage: a trader using borrowed capital is not only hit simultaneously by a shared move but pushed simultaneously toward margin calls or forced liquidation. Second trading hours: a market that never closes offers no overnight gap across which a move can distribute itself.
The practical consequence: anyone trading crypto should group more coarsely than in equities and, in case of doubt, treat the entire area as a single group.
How do you measure your total open risk?
Through one number you can read off at any time: the sum of the distances between entry price and stop across all open positions, each expressed as a percentage of your account or in R.
Three things make that number usable:
- It is current rather than historical. When a stop is trailed, that position's contribution falls. Leaving the entry value in place permanently overstates your risk.
- It is uniform. Measured in R, equities, derivatives and crypto can be added together, whereas in currency terms only positions in the same account currency can. How R works is covered under R multiples in trading.
- It is visible. A figure you have to add up first will not do its job at the moment you need it, which is before the next entry.
This sum is the more honest version of what the one percent rule per trade is supposed to control. Its groundwork sits in position sizing and the one percent rule.
How should you group positions?
Coarsely and in writing. The mistake is not grouping too crudely, it is not grouping at all.
Assign each position a theme from a short fixed list. Four to six groups suffice for most accounts, by industry, region or asset class, and in crypto additionally by the broader ecosystem. What matters is that the list is set in advance rather than extended during the trading day.
When in doubt, two positions belong in the same group. Being unsure whether two instruments are connected effectively answers the question already, since separating them would need a justification. The cost of grouping too coarsely is a few missed trades, the cost of grouping too finely is a day at five times the risk.
What is a sensible limit?
That cannot be answered universally, but the structure of the answer can. You need two limits rather than one.
The first caps total open risk across all positions. It is the actual brake and answers the question of what a single very bad day is allowed to cost. The second caps open risk per theme group and stops the first limit from being filled entirely by one theme.
You derive both figures from the decline you can absorb, not from the number of signals currently on your screen. What counts as a normal decline and what does not is covered under trading drawdown.
What do you do when the limit is reached?
The consequence has to be fixed in advance, otherwise it gets renegotiated at the moment a good signal appears. Three reactions are legitimate, and all three are decisions rather than exceptions.
- Do not enter. The simplest option and in most cases the right one.
- Enter smaller, so the sum stays within the limit. That is fine as long as the size follows from the calculation rather than from a wish to be involved.
- Make room by trailing the stop on an existing position in the same group or closing it. What matters is that this decision rests on the existing position and not on the new signal.
What is not legitimate: raising the limit while a position is open. That adjustment always goes in the direction of more risk and never the other way.
Do offsetting positions count as a reduction?
Only in a limited sense, and the question deserves its own answer because it often gets used as a licence for more positions.
An offsetting position in the same theme group reduces your combined risk arithmetically but does not remove it. Two reasons: first, the relationship between the two prices is not stable and shifts precisely in volatile phases. Second, each position carries its own costs and its own stop, so two positions that cancel out on paper can still lose money together.
In practice that means not netting an offsetting position against your open risk but treating it as its own position with its own contribution to the sum. Netting it out lowers a figure rather than a risk. If you are hedging deliberately, record that in your journal so you can later assess whether the hedge was worth what it cost.
How do you check afterwards whether your grouping was right?
Through the bad days rather than the average. Take the ten worst trading days in your history and look at how many positions were simultaneously down and which groups they came from.
Two patterns are informative. If positions from different groups regularly lost together on those days, your grouping is too fine and groups should be merged. If your worst days instead come from one group only, the classification works and the open question is the level of the group limit rather than the grouping itself.
All of that requires theme and timestamp recorded per trade. A properly kept trading journal supplies both without extra effort, since the timestamp comes from the exchange connection and only the theme is set by you.
Where does this topic end and drawdown begin?
The distinction is one of timing. Correlation is the cause before the loss and can be capped in advance. Drawdown is the consequence afterwards and can only be measured and endured.
Different tools apply as a result. A rule for recovery helps against an accumulated decline, while only a limit that applies before entry helps against hidden correlation. Watching the decline alone means reliably measuring the effects of a cause you are not addressing.
One final point: a limit on total risk does not make your account safe, it makes the size of a bad day predictable. That is less than it sounds, and still the difference between a setback and damage an account does not recover from.