Open research instrumentArithmetic, not a forecast
More sleeves stops working, and you can see where.
A book of equally weighted sleeves, each worth s, sharing a
pairwise correlation rho, is worth
s × sqrt(N / (1 + (N-1) × rho)). Take that to the limit and the
ceiling is s / sqrt(rho), which does not contain N at all. Past a point, breadth
is not the lever and correlation is. This is the most important arithmetic in this project
and it decides whether its own objective is reachable.
What this is. Closed-form arithmetic under strong assumptions. It shows the SHAPE of the constraint breadth operates under. It is not a forecast of any book, and no output here is a performance claim.
The assumptions, which are strong
What it assumes
- every sleeve carries the same Sharpe ratio
- every sleeve is weighted equally
- every pair of sleeves shares one correlation
- correlations are stable, which is the assumption that fails first in a crisis
Real books have none of these. The lab is for the shape of the constraint, not for
forecasting a book.
Where the numbers come from
The presets read this project's own published claims: the research-curve average
pairwise correlation, the current and planned sleeve counts, and the declared forward
Sharpe objective. 0.5 is used as the per-sleeve Sharpe in every preset. It is an illustrative round number, not a measured sleeve figure: no sleeve here has an established forward Sharpe.
The case it refuses
1 + (N-1)*rho must be strictly positive. At zero the equally weighted book has zero variance and an infinite Sharpe, which is refused rather than displayed. A tool that displays an infinite Sharpe for an impossible
correlation is doing the opposite of its job.
Why it matters here
This project's live contract permits an average pairwise correlation up to 0.15. Load
that preset and read the ceiling: it is the honest answer to whether the declared
objective is reachable by adding sleeves, and it is not the answer anyone wants.
Sandbox contract. every input is stored in the URL, so a scenario is shareable and reproducible. no write path to any ledger, artifact, broker or published record exists. outputs are arithmetic under stated assumptions, never a forecast.
Rendered from
breadth_lab_contract.json. Related:
the Selection Risk Lab and
the arithmetic of not fooling yourself.