Methodology
Position sizing: the arithmetic behind the weights
Selection decides the move. Size decides how much of it reaches the portfolio — and a ceiling set once does not stay where it was put.
Last updated: 29 September 2026 · By The Acutic Research Team
Most writing about single stocks is about which ones. Far less is about how much — which is odd, because the second question governs how much of the first one ever reaches the portfolio. The link between them is not a theory. It is one multiplication, and it is worth doing explicitly.
The multiplication
The portfolio-level effect of a move in one position is that position’s weight multiplied by the move. Nothing more. A -30% decline in a name at 5% of the portfolio costs -1.5% of the total; the identical decline at 20% costs -6.0%.
The relationship is linear, which makes it easy to state and easy to underestimate. The more useful way to read it is backwards: rather than asking what a decline would cost, ask what decline it would take to cost an amount that would actually matter.
| Position weight | Decline costing 5% of the portfolio | Effect of a -30% move |
|---|---|---|
| 5% | -100.0% | -1.5% |
| 10% | -50.0% | -3.0% |
| 15% | -33.3% | -4.5% |
| 20% | -25.0% | -6.0% |
Read the middle column as a ceiling on damage: at 5% the position would have to go to zero and still not cost 5% of the whole.
That inversion is the whole argument for caring about size. At 5%, a position going to zero — the worst case that exists — still costs less than 5% of the portfolio. At 20%, a -25.0% move gets there, and declines of that size are ordinary events for individual companies. The same research, the same conviction, the same instrument: what changes the outcome is the weight.
Three ways the question gets answered
There is no standard size, and it is worth being blunt about that, because a great deal of writing presents one as though there were. What does exist is a small number of distinct approaches, each of which holds something different constant.
- Equal weighting fixes every position at the same share of the portfolio. It holds exposure constant and makes no claim that some ideas are better than others — which is either its honesty or its limitation, depending on the view taken.
- Conviction weighting varies size with the strength of the case. It holds judgement as the input, which means its results depend entirely on whether that judgement is calibrated — and calibration is measurable only after the fact, by comparing stated confidence against outcomes.
- Volatility scaling sizes inversely to an instrument’s variability, so a more volatile name occupies a smaller share. It holds risk contribution roughly constant rather than capital, which is a different thing from equal weighting and is often confused with it.
The vocabulary around these — a starter position, a full position, a ceiling — is widely used. The numbers attached to those words are not standardised, and this article does not propose any. The arithmetic above works with whatever figures are actually in use.
A ceiling erodes on its own
The part that surprises people is that a size limit does not stay where it was put. Weight is a ratio, and a position that does well grows its own numerator.
| Position multiplies by | New weight (started at 10%) |
|---|---|
| ×1.5 | 14.3% |
| ×2 | 18.2% |
| ×3 | 25.0% |
| ×4 | 30.8% |
Assumes the rest of the portfolio is unchanged — a simplification that isolates the drift. Computed as grown ÷ (rest + grown).
A position that starts at 10% and triples, with everything else unchanged, arrives at 25.0% of the portfolio — without a single additional euro going into it. The uncomfortable implication is that drift is driven by the positions that worked. Concentration is very often the residue of success rather than a decision anyone made, which is exactly why it goes unnoticed: nothing ever happened that would prompt a second look.
What that concentration then means for a portfolio is the subject of portfolio concentration risk, which takes the question from the other end.
Why this is a measurement problem
A weight is not a matter of opinion. It is computable at any moment from current values, which makes the practical question narrow: is it being computed, and is there anything to compare the answer against?
The second half matters more than the first. A number on its own is not a signal; 25.0% means nothing until it sits next to the size that was intended. This is the argument for writing sizing down at the point the position is opened, when the reasoning is available and before any outcome exists to rationalise it — the same case made in portfolio rules you actually keep and, for the reasoning itself, in the investment decision journal.
Acutic computes current weights against the sizes recorded for each position and surfaces the gap as a factual observation — the number, the intended number, and the difference, with no view about what to do about it. What the analysis is and how it is produced is set out on the methodology page; the workspace itself is described under product.
Frequently asked questions
What does position sizing actually decide?
The arithmetic is unglamorous: the portfolio-level effect of any move is the position’s weight multiplied by that move. A 30% decline in a 5% position costs 1.5% of the portfolio; the same decline in a 20% position costs 6%. Selection determines the move; sizing determines how much of it reaches the total.
Is there a standard position size?
No. The vocabulary is common — a starter position, a full position, a ceiling — but the numbers attached to those words vary widely between investors and no figure is authoritative. Anyone presenting a specific percentage as the established rule is describing their own preference, not a standard. The arithmetic on this page works with whatever weights apply.
Why does a position exceed its ceiling without anything being done?
Because weight is a ratio and the numerator moves. A position that begins at 10% and triples, with the rest of the portfolio unchanged, becomes 25% of the total. Nothing was added; the denominator simply grew more slowly. This is why a ceiling set once tends to be exceeded quietly rather than visibly.
What approaches to sizing exist?
Broadly: equal weighting, which fixes every position at the same share; conviction weighting, which varies size with the strength of the case; and volatility scaling, which sizes inversely to an instrument’s variability so each contributes similar risk. Each embeds a different assumption about what should be held constant — the share, the judgement, or the risk contribution.
How is drift noticed in practice?
By measuring it. A weight is a fact that can be computed at any time from current values, so the practical question is only whether it is computed regularly. A written record of intended sizes makes the comparison possible; without one there is nothing to compare against.
Further reading: how often to check a portfolio, behavioural mistakes and the system fix for each and how to benchmark a portfolio. Create free account.
Acutic provides investment research and educational analysis under MAR Art. 20 / § 85 WpHG. Acutic does not provide investment advice (Anlageberatung per § 1 Abs. 1a S. 2 Nr. 1a KWG / Art. 4(1)(4) MiFID II), portfolio management, or any other licensed investment service. No content on this platform constitutes a personal recommendation.