Automated yield vaults on Base
Every reinvestment has a cost. A keeper that compounds every block burns more gas than the position earns. So the vault prices the fold against what the fold is worth — in the same unit, in the same block — and refuses a crease that would not pay for itself.
—
—
—
Every fold the vault makes is written down: when it was made, what it claimed, what it cost, and how long it had been since the last one. This is that log, for the position set on the Fold tab, over one year at the gas price Base is charging right now.
| # | at | gap | gross | cost | fee | folded |
|---|
—
The gate is economic, not a role. Anybody may call fold(); what decides
whether it goes through is arithmetic nobody can set.
cost = foldGas × block.basefee + Base L1 data fee (predeploy 0x42…0F)
hurdle = max( cost × (1 + minGainBps), minGainWei )
fold() = revert unless pendingYield ≥ hurdle
and block.timestamp ≥ lastFold + minFoldGap
Nothing in that is frozen into the source. The hurdle is read off the block,
so a vault configured once keeps making sense when the chain gets expensive.
foldGas is the one claim about the world, and it is measured: a fold against
Moonwell Flagship ETH on Base uses 1,447,034 gas, because claiming the yield walks
every Morpho market the vault is allocated across. The first number written into the repo
was 420,000, and the fork test refused it.
The multiple matters more than it looks. At break-even a fold spends a wei to claim a wei; at a gain half again its cost — which sounds prudent, and was the first number written here — two thirds of every harvest goes to the chain. Twenty times puts the gas at five per cent of what is claimed, at worst. And the hurdle alone does not bound how often a vault folds: a position large enough clears a fixed floor every minute, inside the six hours the last fold is still vesting over. Hence the gap.
A fold's net gain is
added to lockedProfit and released linearly over six hours. Without it,
arriving in the block before a fold and leaving in the block after is free money taken
from everybody who was already there.
The keeper's reimbursement is a share of the performance fee — never of the principal, never of the gross yield. A payout that can exceed the fee is a faucet, and a bot will find it the same afternoon.
Whatever slice the yield source keeps on the way back in comes out of the gain being deferred, not out of the money that was already there.
| PLEAT | One billion, minted once. No owner, no setter, no pause, no blacklist, no fee, no second mint. | — |
|---|---|---|
| FoldVault | ERC-4626 over WETH. The cost gate, the fee, the drip and the crease log. | — |
| ERC4626Strategy | Puts the principal into Moonwell Flagship ETH and draws the line between principal and the yield above it. | — |
| Reserve | Where the fee lands. Anybody can sweep it into PLEAT and burn what it buys. | — |
The protocol charges its fee in WETH, on yield, inside the vault. It is never charged on a transfer of this token. What the token does is sit at the end of that fee: the Reserve spends what arrives on PLEAT and burns what it buys, and anybody can make it happen.
A fold claims yield and takes 10% of it, in WETH.
20% of that fee reimburses whoever paid the gas. The rest reaches the Reserve.
Anybody calls sweep(). The Reserve buys PLEAT on the pool and burns it.
The swap is quoted at spot, and refused if spot has drifted from the pool's own time-weighted tick. Capped per call and per block.
A tick is not a price anybody can read, so the opening position was chosen in dollars and then converted. 200,000,000 PLEAT — 20% of supply — goes in single sided: entirely on its own side of the price, with no WETH beside it. At ETH = $2,292:
| FDV | PLEAT sold | WETH out | |
|---|---|---|---|
| pool opens | $150,456 | — | — |
| float's low edge | $153,496 | — | — |
| ×2 | $306,991 | 82.4M | 7.81 $17,895 |
| ×5 | $767,479 | 155.6M | 23.30 $53,400 |
| ×10 | $1,534,957 | 192.4M | 40.75 $93,414 |
| range ceiling | $1,832,706 | 200.0M | 46.27 $106,078 |