Protocol mechanics, parameters and current status.

Every swap trades against a pool of capital that someone committed. Whoever committed it receives a share of the fee on every trade that uses it. Resident is that capital, and that share is what it earns.

Liquidity does not have to cover every price. A position can be committed to a single band, say between $9 and $11. Inside that band it takes a much larger share of the trades, because the capital is concentrated where the trading happens. Outside it, the position earns nothing and holds whatever the pool left it with.

The protocol does three things: select pools worth being in, size a range the price will hold inside, and close a position once it no longer earns its keep.

This is not a position on price. It is rent, paid for making a market other people trade in.

Every time someone buys or sells $RES, a fee is charged. That fee goes into the vault, and the vault is what buys the positions. There is no raise, no VC round, no treasury sitting on the side.

Capacity is therefore bounded by the token's own trading activity. Lower volume means less capital deployed and smaller distributions. The two are coupled by design, and nothing decouples them.

Seven conditions, evaluated continuously. All seven must hold; a pool that fails any one is not a candidate at any size.

The pool can be priced in dollars. Pools are quoted in USDG or in WETH, and every threshold below is a dollar figure, so a WETH pool is converted at the rate its own market against USDG is trading at. A quote the protocol cannot price is held rather than measured in the wrong unit, because eight ether an hour of volume looks like nothing against a $25,000 threshold and a hundred ether of depth looks like nothing against a $400,000 cap. The fee the pool charges reaches the position. The protocol reads the fee being charged now rather than the tier the pool was created with, so a pool that has quietly stopped paying its providers fails here. At least $25,000 has traded in it in the last hour, because a quiet pool pays nothing. No more than $400,000 of liquidity sits near the current price, because the more crowded a pool is, the smaller the share. The price is still within 40% of its 24-hour high, so the protocol is not providing liquidity into a fall. The pool is at least 20 minutes old, which excludes the first minutes of a launch. And the providers already in it are making money rather than losing it.

These are Uniswap v4 pools. Many of the best of them charge a dynamic fee that rises when the market is busy, and a dynamic fee is implemented with a hook. A rule that avoided hooks would therefore avoid the pools worth being in. What matters is not whether a pool has custom code attached, but whether the fee it charges still arrives, which is read from the pool directly.

A pool that stops qualifying remains on the board for 24 hours rather than disappearing, so activity that has already passed stays visible.

A narrow range earns more per trade but the price escapes it sooner. A wide range earns less but holds on longer. There is no setting that is good at both, so the width is worked out per pool from how much that pool actually bounces around.

Halving the width roughly doubles the earnings and increases time spent out of range. Resident targets a width that holds the price inside approximately 91% of the time. Widening to 99.8% coverage would surrender about half the income for the remaining nine points, so it is not taken.

Ranges are bounded at 1% and 60% either side of the price.

When the price moves while capital sits in a range, the position ends up holding more of whichever asset fell and less of the one that rose. That is a worse outcome than holding both and doing nothing, and the cost scales with how far the price travels.

A pool paying 3% a day into a book that moves 20% a day loses money while collecting fees. A board ranked on fee income recommends it every time.

Resident prices both sides before opening: expected fee income against the expected cost of price movement, at that pool's own volatility. A position is opened only when fees exceed that cost.

A position out of range earns nothing. It is not moved immediately, since prices cross an edge and return routinely, but a sustained move re-centres it on the new price.

A position closes when net earnings fall 35% below the capital committed, where net means fees collected less the cost of price movement. A pool that stays unprofitable is dropped rather than re-centred, because what has changed is the pool and not the position.

Positions that move against the protocol are held and re-centred rather than sold into thin liquidity, which usually deepens the loss. The trade-off is that capital can remain committed to a weak market for an extended period.

Choosing a position and taking one are different problems. A process reads the chains on a fixed interval, prices every pool it watches, decides, and writes down what it is about to do before it does it. If it stops halfway through, it restarts knowing there is a transaction it cannot account for, and it checks the chain rather than assuming either way. That is what prevents the same position being opened twice.

Each interval runs the same rules in the same order. Close anything that has stopped earning; collect fees that are worth collecting; re-centre anything that has drifted off the price; move money off a pool that still works but has been overtaken by a better one; put whatever is idle to work in the best pool that qualifies; and only then consider moving money to another chain, which is the one decision that cannot be undone within the interval.

That fourth rule is the one most desks do not have, and its absence does not look like anything. Every position is above its floor, nothing has stopped earning, and the money is sitting in the pool that was the best one when it was deployed rather than the one that is best now. A pool paying many times what a position currently earns is a reason to move, and the move has to clear three times what the round trip costs before it is taken.

Fees are collected once they reach $100, or after 15 minutes, whichever comes first. They are never collected when the amount would not cover several times the cost of collecting it, so a pool that has gone quiet stops being swept rather than being drained a few dollars at a time.

The process cannot change anything about the vault. It can move money between venues; it cannot change which venues exist, raise the payout cap, or allow a new bridge. It refuses to start if the key it has been given is the one that could.

Nothing is remembered that was not written down first.

Realized profit splits at a fixed ratio enforced by the contract: 15% to $RES holders, 85% retained to fund new positions.

The holder share is distributed in USDG, pro-rata by balance. No staking, no claiming, no signature. A distribution runs at most every 15 minutes, and only once every holder is owed at least $5 and gas is under 2% of what goes out. Paying a holder six cents costs more than six cents to send, so a cadence fixed to the clock would spend the distribution on distributing it. Nothing is lost by waiting: what is not paid stays in the ledger and the next distribution is larger. The practical effect is that the interval widens with the holder count rather than the gas bill doing so.

The balance owed only increases until it is paid. Losses are absorbed by the retained share; profit already credited to holders is never reversed. Accrual pauses until the loss is recovered, so distributions go quiet in the interim.

The retained 85% is working capital, not a deferred holder claim. It funds positions, absorbs losses, and holders have no claim on it.

Almost every liquidity dashboard shows fees earned. Fees are the flattering half of the number: a position can be collecting handsomely while the money underneath it loses more than the fees bring in, and a page that shows fees alone will show that position as a winner for as long as it keeps losing.

Resident reports net over the same window, meaning fees less what the price move cost the capital committed, and reports it when it is negative. Both halves are shown alongside it, so the arithmetic is visible rather than asserted.

Positions are kept in a ledger that nothing leaves. What a closed position made is arithmetic on amounts that went in and came out, with the transactions listed; what an open one is worth is marked and reported separately, never added to the closed figure. Adding the two produces something that moves with the market and reads like a bank balance.

Where a figure has not been measured, the page says so instead of showing a zero. A pool with no comparable market elsewhere has no reference price, and a position without enough history has no six-hour net.

Resident is pre-deployment. The vault contract is complete and covered by a test suite; it has not been externally audited and is not deployed. No vault holds assets, and the positions page displays example data, labelled as such.

The process described above runs, and it signs nothing. It reads chains, ranks pools, and records every position it would have opened. That record is worth having on its own, because it can be checked against what those pools actually paid before any capital is at risk. Two things stand between it and taking a position: a signing service, which does not belong in this code, and the venue-specific component that knows how to open a position on each venue.

Fee projections start at full capture: the formula credits a position with every unit of a pool's reported flow at its full share, and a pool's flow includes trades at prices the position does not cover. That figure is an upper bound and has never been what measurement returned. The protocol measures the gap instead of assuming it, taking one sample per sweep of what the model predicted against what arrived, and prices each pool at what it has actually been paying, its own history first, the desk's average second. Until a pool has been traded there is no measurement, and it is priced at a discount to the ceiling rather than at the ceiling.

Every dollar of realized profit splits the same way, and the contract is what enforces it.

Holder balances are reconstructed from the token's complete transfer history, maintained locally rather than read from a third-party indexer, and verified against chain state before any value moves. Pools, protocol contracts and desk wallets are excluded from the eligible supply.

All capital, including fee inflow, cash and inventory, is custodied by a single on-chain vault contract that is publicly auditable in real time. An autonomous keeper operates against that contract: it surveys venues, routes orders and triggers distributions, while the contract constrains every action it can take. The keeper holds nothing itself. Every balance sits at the vault address.