A vault is a strategy with a boundary. Four things define one, and all four are published before a member can allocate to it.

Mandate

What the vault may buy, at what stage and through which instrument. Narrow by design — Perpetual has reviewed twenty-two companies for six slots.

Risk tier

One of R1 to R5. Exactly one per vault, stated on every surface the vault appears on. See Risk tiers.

Lock-up

How long capital is committed. From same-epoch exit in Reserve to 84 months in Genesis. The term is the strategy, not a penalty.

Capacity

The most the vault will hold. A stated ceiling, not a target — the book runs at 77.3% of it.

A vault holds its positions directly. It does not lend to another vault, does not net against another vault's book, and cannot be gated to fund another vault's redemption. Reserve is the only vault that funds the others, and it does so by holding $132.7M in cash and tokenised bills rather than by drawing on anyone.

Every vault publishes its own mandate and its own risk statement. Those sit on the vault's page at /vaults/<name>. This document is the comparison across all of them.

Sorted by risk tier ascending, then by deployed capital descending inside a tier — the same order as the vault index and the same order the prev/next navigation runs in.

Deployed capital sums to $1.94B. Reserve shows zero positions because it owns no private company — it is the cash sleeve.

Four axes separate the mandates. A vault is usually distinctive on one of them and ordinary on the rest.

Stage

Genesis writes seed and Series A cheques before a company has a public price. Anchor requires more than $200M of annualised revenue, growth above 40% and an auditor the House will name. Same book, opposite ends of it.

Instrument

Structure buys the paper rather than the share — convertible notes, capped SAFEs, discounted secondaries — at an average entry discount to the last primary of 17.4%. Momentum buys employee stock in the secondary market inside ninety days of a confirmed up round. Both hold the same class of asset through a different door.

Concentration

Keystone holds exactly ten positions; adding an eleventh requires selling one first, named in public on the epoch boundary. Perpetual trims a position when it passes 30% of the vault and tops it up when it falls under 8%, and that is the only trading it does.

Duration

Momentum's median holding period is 14 months. Perpetual's weighted holding period is 4.1 years and it has no scheduled liquidity at all. Genesis commits for 84 months because that is how long the strategy takes.

Two of the 10 are structural rather than directional, and they are the reason the other eight can behave the way they do.

• Reserve owns no company. It is 82% cash and 18% tokenised bills, holding $132.7M. Every allocation window draws from it first, and every exit settles out of it in the same epoch, at par. It exists so that the other nine vaults never have to sell a position to fund a redemption or a follow-on.
• Compound takes no new capital. It is funded entirely by realised proceeds from the other nine, written into mid-stage companies at a maximum $20.0M per name. Its size is therefore a lagging read on how much the rest of the book has actually exited.

Capacity is a per-vault parameter, not a House total. The ten stated capacities sum to $2510.0M and $1.94B is deployed against them — 77.3%.

Because it is per vault, one vault can be closed while others stay open. Keystone is the fullest at 84.2% of $340.0M; Genesis is the emptiest at 54.5% of $80.0M. A vault that reaches its capacity stops accepting new capital at the next window and stays closed until governance raises the ceiling, which moves on an epoch boundary like every other parameter.

A company can sit in more than one vault. That is not double counting: each vault holds its own position, sized against its own book and marked on its own terms. A late-stage name can be a core holding in Anchor and one of the ten conviction slots in Keystone at the same time, for different reasons and at different sizes.

Active companies held by three or more vaults. The full list of holdings is on the portfolio.

The relation is built from each vault's own holdings list rather than typed twice, so a company can never claim a vault the vault does not claim back. If a position is not in the vault's published holdings, the vault does not hold it.

The vault is the decision. Read the tier first and the mandate second.

• You may need the money inside a year. Reserve, and only Reserve. It is the one vault that settles in the epoch it is called, at par, with no queue and no gate. It is also marked at 1.02× and will never be marked higher — the return is the bill yield.
• You want private exposure with a plausible date. Anchor. Six companies past product risk and inside two years of a listing, on a 24-month lock-up. It carries the second-lowest mark in the book at 1.14× because it buys close to the last primary and has never written a position up on a modelled comparable.
• You want the exposure with the top taken off. Shield. It sells a public technology index against all four of its holdings, which costs roughly 4.2% a year in carry and is why it is the only equity strategy in the House rated R2. It will always mark below Anchor. That is the trade, and it is stated before allocation rather than after a good quarter.
• You are content to be told nothing for years. Perpetual or Keystone. Perpetual compounds and never fully realises; Keystone holds ten names on a 48-month lock-up and a single write-down moves it — a total loss on its largest holding would take 11.4% off vault equity in one epoch.
• You want variance and can afford to lose it. Genesis or Frontier. Genesis writes six positions and expects two of them to return nothing; that is the base case, not the failure case. Frontier underwrites companies that fail on physics and permits rather than on product-market fit, on a 60-month lock-up with no secondary market of any depth behind it.