A position is carried at the price of the round that actually priced it. Not at a comparable, not at a discounted cash flow, not at a manager's view of what it is worth. If no round has priced it, the carrying value does not move.

Three things follow from that, and all three are uncomfortable at times.

• A position that has not repriced does not move. Two of Structure's five positions have been carried flat since entry because no round has repriced them. The vault would rather show a flat mark than a modelled one.
• A mark can be old, and it says so. Every mark publishes the round's date alongside its price, so a member can tell a fresh primary from a two-year-old secondary before reading anything else about the position.
• A position is never written up on a comparable. Anchor has never done it, which is why the largest and most mature vault in the House carries the second-lowest mark in it at 1.14×. It buys close to the last primary and does not carry a position above the price a real buyer paid.

There is one case where a mark moves down without a round. Frontier publishes each position's next technical milestone and the date it is due, and a missed milestone marks that position down automatically until the next round prices it. Marks move down on evidence; they only move up on a price.

The book carries at 1.38× cost, weighted by deployed capital.

A flat mark is not an error and, on its own, not a bad sign. It means no buyer has set a price since entry. What matters is how long it has been flat and why.

In Structure that is normal rather than notable: the vault holds convertible notes, capped SAFEs and discounted secondaries, and an instrument of that kind does not reprice until something converts. Two of its five positions have been carried flat since entry, and the vault publishes that fact rather than smoothing it with a model.

Perpetual publishes the round, the date and the price behind every position's carrying value for the same reason — so a stale mark is visible as stale rather than mistaken for a stable one. A mark that has not moved in two years and a mark that was set last month look identical on a summary page; they are not the same thing, and the date is what separates them.

At the other end, Reserve is marked at 1.02× and will never be marked higher. It owns no company. The only return there is the bill yield and the only exposure is the issuer of the bill.

There is one place where a priced round does not exist and a committee has to write a number: the earliest stage.

Genesis is marked at 2.74×, the strongest mark in the book. That figure is an estimate written by a valuation committee rather than a price set by a buyer, because its six positions were entered before the companies had a public price. Members are told that before they allocate, and the mark is republished every epoch with whatever round set it — where no round has, the mark says so.

Treat a committee mark as an opinion with a number attached. It is also the reason Genesis is sized the way it is: at $43.6M it is 2.2% of deployed capital, so writing it all the way back to cost would move the book's carrying mark from 1.38× to 1.34×. The most uncertain number in the book is deliberately the smallest one.

Four cadences, none of them a quarterly document.

None of this sits in a data room behind a request. What a member can verify for themselves, and what an attestation does and does not prove, is in Custody & attestation.

The headline figures are +38.4% net PnL since genesis, −7.2% maximum drawdown, a 2.1 Sharpe ratio, $412M realised and a 1.38× carrying mark. Each one is measured, and each one can be misread. How to read them.

+38.4% is not annual

It is cumulative since genesis on 11 March 2026 — under six months. It is not an annualised figure and should not be compounded forward as though it were.

1.38× is not a return

It is a carrying mark on an unrealised book. $1.81B of the book is held and marked; $412M has actually come back from four exits. The difference between those two numbers is the whole distinction between a mark and a result.

−7.2% flatters the risk

It is a drawdown on marked value. A book that marks on epoch boundaries and only off priced rounds shows a smaller drawdown than one marked against a public index, because its marks move less often. That is a property of the measurement, not evidence of a safer book.

2.1 Sharpe has few observations

Computed on marks that move on epoch boundaries rather than on daily prices, so it rests on a handful of data points and understates variance for the same reason the drawdown does. Read it as a comparison between epochs of this book, not against a liquid strategy.

Vault figures beat book figures

A member holds vaults, not the House. Momentum's own maximum drawdown is 9.4% of vault equity and Keystone's single-position concentration is 11.4% — both larger than any House-level number, and both the figure that actually applies to somebody holding them.

The honest summary: the book is young, the marks are conservative in method and unrealised in fact, and the risk figures understate the variance of the underlying asset because the underlying asset is not priced often enough to reveal it. Every one of those statements is a consequence of holding private companies, and none of them is fixed by putting the claim on chain.