Allocator view. Everything below is re-derivable by you from the public bundle: no account, no NDA, no call with us. The manager grants nothing and reveals nothing. You are not being asked to believe anyone, including MIZAN. Fictional holder · several fields not yet issuable.
Meridian Systematic · Book I
Every strategy this holder has ever submitted appears above, because the registry is append-only and each entry was anchored to Bitcoin on the day it was made. A holder cannot show you only their winners. This is the denominator allocators have not been able to obtain from a manager, database or administrator.
| Condition | Gate requires | Result | Proven |
|---|---|---|---|
| Sharpe, after costs · book | ≥ 1.20 | 1.44 | in-circuit |
| Max drawdown, intrabar-marked | ≤ 25.0% | 14.9% | in-circuit |
| Out-of-sample Sharpe | ≥ 0.50 | 1.02 | in-circuit |
| Strategies tried before this book (N) | declared → proven | 8,410 | Merkle leaf count |
| Breadth · committed instrument count | ≥ 20 | 1,284 | in-circuit |
| Research velocity · models turned over / yr | ≥ 10% | 18.2% | chained ledger |
| Alpha models in book | ≥ 5 | 11 | each committed |
A gate can only ever judge a record. This section judges the behaviour that followed it: the committed specification re-checked against attested positions, so a holder who passes and then trades something else is visible rather than assumed honest.
Every tearsheet in the industry uses its own conventions, so cross-manager comparison has rarely been honest. These percentiles are computed against books judged by the same hash-pinned gate, on the same cost treatment and the same intrabar drawdown marking. The comparison set is public and append-only; anyone can recompute the distribution.
Correlation to your own book, proven. Commit your portfolio’s return series: it never leaves your machine and we never see it. The holder proves, in zero knowledge, that this book’s correlation to yours sits below a threshold you choose. You learn whether it diversifies what you already own; the holder learns nothing about your portfolio; we learn nothing about either.
Removed · you no longer take these on trust
- Reconstructing whether the record was computed honestly
- Asking how many strategies were tried, and believing the answer
- Re-running their numbers from files they chose to send you
- Wondering whether costs, the out-of-sample split or the start date were set after the fact
- Normalising their conventions against another manager’s
- Taking research velocity from a questionnaire
Unchanged · still entirely your judgement
- Capacity, crowding, and what happens after a real allocation
- Whether they can manage a book at your size
- Operational DD, custody, valuation policy, admin
- Counterparty relationships and financing
- Key person, team stability, incentives
- Whether you want this exposure at all
This is a screen, not a decision. It removes one category of doubt completely and leaves every other category exactly where it was: the only honest claim a verification layer can make.
Download the 231 KB bundle and re-check every number on your own machine in ~84 ms. No network, no account, no trust in us.
Prove this book’s correlation to your portfolio below a threshold you set, without either side revealing anything.
Send the return series of any manager you are screening. We return their deflated Sharpe at N = 1, 10, 50, 200. Expect it to be unkind.
One line in your diligence template: “track-record verification may be satisfied by a VTR-1 credential.” The spec is CC BY.
Crowding, capacity, the ability to manage the book at size, operational integrity, custody, valuation policy and counterparty competence are outside every proof system and outside this document, permanently. It does not predict returns and is not advice. It proves that the record is real, complete, and honestly measured, and then it stops. Anyone selling a credential that claims more is selling the thing this standard exists to refuse.