Sizing By Conviction

Two traders look at the same stock, agree on the same risk budget, the same stop, the same liquidity ceiling — and take completely different sizes. The difference is not risk management, it is conviction, and if you cannot explain in one sentence why you are more or less sure than the maximum the other three ceilings allow, you are not sizing by conviction, you are guessing.

Conviction is a fraction, not a feeling

Once the risk budget, liquidity and any house caps have set a ceiling, conviction decides what fraction of that ceiling you actually take. A useful way to make it concrete is to break a thesis into its legs — the two or three things that all have to be true for the trade to work — and score how many are already confirmed versus assumed.

Say a long thesis rests on three legs: the company beats on margins, the sector re-rates, and a competitor's product delay continues. If only the margin call has independent evidence behind it (channel checks, a supplier data point) and the other two are your forecast, you have one of three legs confirmed. A simple, disciplined mapping — one leg confirmed = one-third size, two legs = two-thirds, all three = full size — turns "I feel good about this" into a number you can defend at the desk review six weeks later.

Worked example

Book equity $80m, per-trade risk budget 25bp, so $200,000. You want to short a mid-cap on a thesis that inventory is building (leg 1, confirmed via channel checks), that guidance gets cut next quarter (leg 2, your forecast), and that the multiple compresses once guidance is cut (leg 3, dependent on leg 2).

• Stop distance: 9% above entry, so the risk-budget ceiling is 200{,}000 / 0.09 \approx \2.22m$.
• Liquidity and concentration caps both sit above $3m — not binding.
• Conviction: one of three legs confirmed → take one-third.

Size taken: 2.22\text{m} \times 1/3 \approx \740,000. Not "\2.22m because that's what the math allows" and not "$500k because it feels safer" — $740k because that is what one confirmed leg out of three actually earns.

A month later channel checks corroborate the guidance cut too (leg 2 now confirmed). Conviction moves to two of three, and the position is scaled up toward $1.48m, assuming liquidity still supports it — not because the stock moved in your favor, but because the thesis got more true.

Why this discipline matters

Left unstructured, conviction sizing degrades into recency bias: traders size up names that just worked and size down names that just went against them, regardless of whether the thesis changed. Tying the fraction to specific, nameable legs of the thesis forces a written answer to "what do I actually know now that I did not know at entry" every time the size changes, in either direction.