How to build a vault

A vault is a wrapper. You deposit, you hold a token, the token tracks what the strategy is worth, and you redeem when you want. Writing that contract takes a week. Deciding what goes inside it, and who it is for, takes longer and matters more.

Two kinds of vault

The word covers two different machines, and you should know which one you are building. A wrapper vault takes one strategy and turns it into one token. Midas does it with private credit, BlackRock with BUIDL, Ondo with OUSG. An allocator vault does the opposite. It pools a single deposit asset and spreads it across many on-chain venues, with a curator rebalancing as rates move. Morpho runs roughly two hundred such vaults across its lending markets, and the large yield-bearing dollars, Ethena's sUSDe and Sky's sUSDS, sit in the same family.

This is worth saying plainly. Aggregating liquidity across venues is the whole point of the allocator type: one token, one risk frame, many markets underneath. The wrapper type does not do that. It gives you clean access to a single strategy and leaves the routing to the fund. Most of the decisions below apply to both, but you are building one or the other, so pick first.

The seven layers

Strip a vault down and you find seven layers. Each one is a choice. A weak choice anywhere shows up later as a frozen redemption, a NAV nobody trusts, or a key that should not have existed. Most of this rides on one token standard, ERC-4626, with ERC-7540 adding the queued redemptions further down. Click through them.

Decision one: who can hold it

This is the first fork, and it sets everything downstream. A capital-safe strategy can cross into full retail under an approved prospectus. Midas did exactly that with tokenized T-bills, an EU first for non-accredited buyers. The moment you add leverage, credit, or short options, you carry real principal risk, and the offer stays restricted to qualified investors. Superstate's USTB is a qualified-purchaser fund for that reason, while the retail-friendly wrappers are the plain Treasury ones like Ondo's USDY or Franklin Templeton's BENJI. Notice that even Midas keeps its 15 percent private-credit product on the restricted side of that line. Whatever you decide has to be enforced on-chain, usually an allowlist on the token so only cleared wallets can hold or receive it. The policy is the easy half. The transfer restriction is the half that actually gates.

Decision two: what the token is worth

A token is only as good as its price. The risk is that a strategy manager reports a number and everyone takes it on faith. The fix is a chain of parties: the manager publishes NAV, the issuer checks it for consistency, an independent verifier signs off, and the result is attested on-chain. Skip the middle steps and you are trusting one party with the one number that matters.

Decision three: how you get your money back

Redemption is where most vaults quietly break. A strategy that earns yield is rarely sitting in cash, so a vault cannot honor every exit on demand. The working pattern is a liquid buffer for small redemptions, plus a refill cycle for the rest. Larger requests are queued, the manager sets aside the funds, and you are paid at the NAV of your request. Ondo holds liquid tokens like Hashnote's USYC and Superstate's USTB inside OUSG for exactly this, a ready buffer so small exits clear at once.

Decision four: what it composes with

A wrapper that only wraps is dull. The reason to put a strategy on-chain is reuse. The same token can post as collateral, back a loan, or settle a trade, without giving up the yield underneath. That optionality is most of the upside, and it only exists if the token is clean and the venues accept it. BlackRock's BUIDL is already used this way across on-chain markets, and credit vaults on Morpho and Maple lean on tokenized Treasuries as their risk-free leg.

Decision five: fees, and where the risk sits

Two numbers tell you who a vault is really for. The wrapper fee is usually thin. Midas charges around 0.4 percent on assets with no performance fee, because the wrapper is plumbing. The strategy keeps its own economics inside the strategy. Read the two together and you see where the work is priced, and who absorbs the loss when the strategy turns. On a leveraged or credit strategy, the parameters that matter most are the ones nobody markets: exposure caps, leverage limits, and the de-risking rule that fires before the buffer is gone.

The token is the easy part

None of this is about the code. A vault is a set of promises: who can hold it, where the assets sit, what it is worth, when you can leave, and who holds the keys. Get those right and the token almost builds itself. Get them wrong and no amount of yield saves it.