What it is, what it is not

Who should stay out

Roadmap and status

Nothing is deployed. This page says what has to be true before each next thing ships, and lists every parameter still waiting on a number.

Status right now

The phases

Each phase lists what has to be true to leave it. Nothing moves on a calendar, it moves when the exit criteria are met.

1. 01Parameters signed offwhere we are

   Every number in these docs currently marked TBD gets a value, a rationale and an owner.

   Exit whenNo TBD is left in these docs

   DateTBD
2. 02Testnetpublic

   Contracts deployed to a public testnet with at least one book and a working liquidation path.

   Exit whenA liquidation has run end to end in public

   DateTBD
3. 03Auditexternal

   The contracts are audited before any book opens. Until that is done, no parameter above is live and no money is at stake. See Governance and parameters.

   Exit whenThe audit is complete

   DateTBD
4. 04Mainnet, caps lowfirst books

   One chain, a small number of books, borrow caps set well under what pool depth alone would allow. Senior vault only.

   Exit whenCaps hold through a real drawdown

   DateTBD
5. 05T3 books and the junior vaultif it holds

   Thin books open, and with them the junior tranche. This is the phase that can be cancelled: if the waterfall on Senior and junior cannot be built so that junior losses stay inside their own book, junior does not ship and T3 stays closed.

   Exit whenTranching audited separately

   May not shipSaid plainly, not quietly dropped
6. 06More chainslater

   A second chain only after the first has been through a full cycle including at least one liquidation cascade. The chains table on Oracle and listing stays empty until each one is actually live.

How a loan works

Deposit a listed meme. Borrow stables against it, up to what its league allows. Interest runs while the loan is open. Repay and get the meme back.

The four steps

One position, start to finish. Every book has two sides: lenders put stables in, you put the meme in and take stables out. Steps 1 and 2 happen in the same transaction. Step 3 runs on its own until you do step 4.

1. 01Depositmeme in

   You send a listed meme into its own book (its own wallet). It stays locked there. It is not lent out, not staked and not used to farm anything.

   You sendThe meme

   You getA borrow limit
2. 02Borrowstables out

   The book values your meme at its TWAP: the average price in its main pool over a time window, not the last trade. You can borrow up to a set share of that value, the max LTV of the ticker's league. Borrow less and the meme has room to drop. A 2.5% origination fee comes off what you receive: borrow $10,000 and you get $9,750, and owe $10,000.

   Your limitMeme value × league max LTV

   You getUSD or the chain stable, less 2.5%
3. 03Pay interestdebt grows

   Interest is added to your debt as time passes, at that book's borrow rate. Nothing is due on a schedule. The debt just grows, and the loan drifts toward liquidation even at a flat price.

   RateVariable, per book

   Due dateNone
4. 04Repaymeme back

   Pay back the stables plus interest, plus a 2.5% repayment fee on what you repay, and the meme unlocks.

   You sendStables + interest + 2.5%

   You getThe meme back

Health factor

One number tells you how close you are to liquidation. At 1 or above you are fine. Below 1, anyone can liquidate you.

What you cannot do

The risk, on this page

Leagues and rank

Your LTV is ranked on the token first, not on the influencer promoting it.

LTV is the share of your meme's value you can borrow. Every listed meme sits in a league, and the league sets two numbers for its book: the max LTV, and the liquidation threshold, the level where your loan can be liquidated. A two day old ticker with a thin pool does not get the terms of one that has survived a year with real depth.

A higher league means better terms, not a safer meme. T1 tickers dump too.

The three leagues

TBD, the ranges below are the working shape, not signed off parameters. Final values land here before the first market opens.

What moves a ticker up

• Age. Time on chain without an incident. There is no shortcut.
• Depth in its main pool. Enough liquidity that liquidating the book's full borrow cap does not crash the price.
• Real volume, held over time. Not one day of fake trades.
• LP locked, verifiably, for a meaningful term.
• Mint and freeze authority revoked. Required to be listed at all. Nothing moves up without it.

What never moves LTV

Borrow caps and pool depth

The league sets your LTV. The borrow cap sets how much the book as a whole can carry, and it is not a judgement call: it is derived from the canonical pool's depth, because the pool is where every liquidation has to be sold.

The requirement is that liquidating the book's entire outstanding debt at once should still be a trade a liquidator would take. That means the sale has to clear under the break-even slippage from the floor, with margin.

The slippage target is the only free choice here; once it is set, κ follows from the formula. The constraint it has to satisfy is that a full-book liquidation still clears, which means s_target has to sit under the break-even slippage the liquidation bonus can absorb. See the floor.

Lower leagues will get a smaller share of depth, not because thin pools deserve punishment, but because their depth is less reliable: a thin pool can halve overnight, and the cap has to survive the gap between recomputations. That ordering is settled even though the values are not.

How often it is recomputed

What happens when the cap is hit

• New borrows in that book pause. Only that book, and only borrowing.
• Everything else keeps working: repay, withdraw collateral, lend, liquidate. See Emergency procedures.
• It clears by itself as borrowers repay, or when the cap is raised through the timelock.
• A cap cut can put a book over its cap without liquidating anyone. Existing loans run on untouched; the book simply stays closed to new borrowing until debt falls back under.

Moving down

Everything that moves a ticker up can move it back down: depth leaves, volume was fake, an LP lock lapses, an authority reappears. A demotion tightens the liquidation threshold on every open loan in that book, so it can make you liquidatable without the price moving at all.

There is a notice period before it takes effect, and the mechanics, the timelock class and what happens to existing positions are on League demotion. If a ticker stops meeting the listing bar entirely, the book is wound down rather than demoted: see Delisting and wind-down.

Interest rates

One formula, four numbers per book. The rate is not set by anyone day to day: it is a function of how much of the book is borrowed right now.

Utilisation drives everything

Utilisation is the share of a book's stables that is currently out on loan. It is the only input to the borrow rate.

The rate curve

The borrow rate is a two-slope line with a kink at a target utilisation. Below the kink the rate rises gently, because we want the book used. Above it the rate rises steeply, because the last stables in a book are the ones lenders need to be able to withdraw, and the only lever that frees them is making the debt expensive.

What lenders earn

Interest paid by borrowers is split between the lenders in that book and the protocol. The protocol's cut is the reserve factor. Everything else goes to the lenders, spread over the whole book, including the part that is not currently lent out.

The arithmetic, with the 20% reserve factor and arbitrary inputs for the other two terms: a book at U = 0.70 and a borrow rate of 12% would pay lenders 12% × 0.70 × 0.80 = 6.72%, and route 1.68% of the book to the protocol. The utilisation and the borrow rate are not decided values.

How interest is added

There is no payment schedule. Interest compounds into your debt continuously, by a per-second index. Your debt is not stored as a number of stables, it is stored as a share of that index, so every open loan accrues without anyone touching it.

APR and APY, and which one you see

APR is the rate before compounding. APY is what you actually end up paying or earning once the per-second compounding is applied. This conversion is arithmetic, not a parameter: it holds whatever the league values turn out to be.

Per-league defaults

Four numbers per league, none of them set yet, plus the reserve factor, which is 20% in every league. The table below is the shape the sign-off has to fill in, and the rate columns are published empty rather than filled with a guess.

All values will be APR. The one property the set has to satisfy is directional: the thinner and newer the book, the lower the kink and the steeper slope 2, because a thin book cannot afford to sit pinned at full utilisation while lenders wait. That constraint is settled even though the numbers are not.

Fees

Every fee the protocol charges, in one table, with a number in every row. Where a fee is zero it says zero rather than going unmentioned.

The whole list

Nothing charges a fee that is not on this table. If you find a charge that is not listed here, it is a bug and we want to hear about it.

Where the money actually goes

There are three sources of protocol revenue: the 2.5% fee on every deposit, withdrawal, borrow and repayment; the 20% reserve factor on borrow interest; and 10% of the liquidation bonus. All three land in the book they came from, and all three serve that book's reserve before anything leaves.

Costs that are not our fees

These cost you money but none of it reaches the protocol. We list them because leaving them out is how a fee table becomes dishonest.

• Gas. Paid to the chain on every deposit, borrow, repay and withdrawal.
• Swap slippage. If you sell the borrowed stables, or buy them back to repay, the price you get is the market's, not ours.
• The liquidator's bonus. The part that does not go to the reserve goes to whoever liquidated you. It is a cost of being liquidated, not a protocol fee, and it is set out on Liquidations.
• The spread on the loan asset. Acquiring or exiting USD has its own cost wherever you do it.

Worked examples

Three positions, carried through with every number shown: one that ends well, one that gets liquidated, and one that leaves a hole. You can check the arithmetic.

1. A healthy loan, open to repaid

MEME_A, a T1 book: max LTV 60%, liquidation threshold 65%, borrow rate 6.30% APR at the book's current utilisation, reserve factor 20%.

The loan cost $6,604.69: $1,565.55 of interest, which is 1.57% of the principal over 90 days or 6.50% APY, plus $5,039.14 in origination and repayment fees. Each fee is charged once, however long the loan stays open, so on a 90-day loan the fees are most of the cost. The meme fell 12.5% while the loan was open and nothing happened, because the position was opened at 41.7% LTV rather than at the 60% ceiling. That gap is the entire reason this example ends with the meme coming back.

2. A liquidation

MEME_B, a T2 book: max LTV 50%, liquidation threshold 58%, borrow rate 15% APR, liquidation bonus 12%, close factor 50%, protocol share of the bonus 10%. Same borrower behaviour as above, but closer to the limit and on a thinner book.

What it cost each side

The liquidation worked because 12% sits inside both bounds at this threshold: the ceiling at L = 0.58 is 1/0.58 − 1 = 72%, and the floor at 5% exit slippage is about 5.8%. Health factor went up, from 0.943 to 1.24, which is the whole point. See the two bounds.

3. A bad debt event

MEME_C, a T3 book: max LTV 30%, liquidation threshold 38%, borrow rate 14% APR, bonus 12%. The borrower did nothing unusual. The token gapped.

Where the $7,749.95 lands

The loss is absorbed in the order set out on Bad debt: the book's reserve first, then junior, then senior. This book is T3, so senior was never in it.

Junior's stake in the MEME_C book falls from $120,000 to $114,650.05, a 4.46% write-down on that book's junior share price. A junior depositor whose vault had $600,000 spread over several books sees the vault's share price fall 0.89%, because only the MEME_C slice was hit. A lender who never had exposure to MEME_C loses nothing at all.

Checking these yourself

Every figure above is reproducible from the formulas on Interest rates and How a loan works. Two things to know if you are rebuilding them against the contracts rather than a calculator:

• Work in integers. The contracts hold balances as token base units and rates in ray, meaning 27 decimals. Reproduce the arithmetic in BigInt, not floating point, or you will disagree with the chain in the last few digits and chase it for an afternoon.
• Rounding is directional. Debt rounds up and collateral rounds down, always against the borrower and never against the book. Over a single position the difference is dust; it is deliberate.
• The TWAP is the price, everywhere. Health factor, borrow limit and the seizure in a liquidation all use the same window, read in the same block. There is no second price anywhere in the system.

Lenders

You put stables in. Borrowers put memes in and pay you interest. This page says how it works.

Where your USD goes

You deposit the loan asset: USD or the chain stable, depending on the deployment. It sits in a book and is lent to borrowers who have locked that book's meme against it. Your return is the interest they pay, minus the protocol's 20% share. Depositing and withdrawing each cost 2.5%. See Fees.

You are only exposed to the books you lend into. A book you are not in cannot cost you anything. Which books you are in is your choice at deposit, and it is the only exposure you have.

Senior and junior, in short

Two claims on the same book, in a fixed order: junior is written down first and paid more for it. The summary is below; the waterfall, the subordination cap, the rate split and the wipeout case are all on Senior and junior. TBD, ships only if we can build it right

Utilisation and getting your money out

Utilisation is the share of a book's stables that is borrowed right now. It sets the borrow rate and it decides whether you can withdraw right now.

• Low utilisation. Plenty of free stables, you can withdraw, your rate is low.
• High utilisation. Rates rise, which pays you more and pushes borrowers to repay.
• At 100%. There is nothing free to withdraw. You wait for a repayment or for new deposits. The book is not broke and nothing is bugged, but your money is stuck until then.
• Junior has two extra gates. A withdrawal cooldown, and the subordination cap: taking junior out of a book that is at its senior limit is blocked until senior leaves too. See Deposits and withdrawals.

How you lose money, in order of likelihood

Ordered by how often it happens, not by how much it costs. The two do not line up: the most likely loss is not the largest, and the largest sits near the bottom. The badge on each row says what is actually at stake.

1. 01
   Bad debt in a book you lend into Deposit

   A meme gaps down faster than liquidations can keep up, the collateral no longer covers the debt, and the gap lands on that book's lenders. Lenders in other books lose nothing.

   TriggerA price gap through the liquidation point

   HitsThe vault lending to that book: junior for T2 and T3, senior for T1

   You losePart of your deposit
2. 02
   You cannot withdraw when you want to Access

   Utilisation is pinned at 100% and you are locked in at exactly the moment you want out. Your balance is intact. You just cannot reach it until a borrower repays or new deposits arrive.

   TriggerUtilisation at 100%

   HitsEvery lender in that book

   You loseAccess, not your deposit
3. 03
   Rate collapse Yield

   Borrowers repay, utilisation drops, and you earn close to nothing while your stables sit idle. Nothing has gone wrong. The book is simply not being used.

   TriggerBorrow demand dries up

   HitsEvery lender in that book

   You loseYield, not your deposit
4. 04
   Smart contract failure Deposit

   A bug in the market, the locker that holds the memes, the price feed or the fee router. Unlikely, and by far the worst: it can take everything, and it does not care which vault you are in or which league the book was.

   TriggerA bug or an exploit

   HitsSenior and junior alike

   You losePotentially all of it
5. 05
   The stable loses its peg Deposit

   You hold the loan asset. If USD or the chain stable stops being worth $1, that is your loss, and nothing in this protocol offsets it. The book still balances, in a token that is worth less.

   TriggerLoan asset loses its peg

   HitsEvery lender, every book

   You loseValue of the deposit

Senior and junior

Two claims on the same book, in a fixed order. Junior is paid more because junior is written down first. This page is the order, the cap, the rate split, and the case where junior runs out.

Why two tranches at all

A single lender pool over memecoin books has one price for a very wide range of risk. The lender who wants 5% on stables and the lender who wants 20% and knows what they are buying both get the same blended number, and the first one is quietly funding the second one's upside. Splitting the claim lets each price their own risk.

This is closer to private credit than to Aave. Aave mutualises losses across the whole pool and backstops them with a staked module. Morpho, Euler and Silo socialise losses per market with no backstop at all. Here the loss is per book, and ordered inside the book. See What makes us different.

The loss waterfall

When a book ends up with bad debt, the shortfall is absorbed in this order, and only ever by money that was in that book.

Junior does not backstop across books

This is the question worth being precise about, because the answer changes what you are buying.

Which books each tranche may enter

The junior-to-senior cap

Senior's protection is only as thick as the junior sitting under it, so a book cannot take unlimited senior money against a thin junior layer. Every book carries a maximum ratio.

What the cap does in practice:

• New senior deposits are refused into a book already at the cap. The deposit is not queued and nothing fails silently: the book is simply closed to senior until more junior arrives.
• Junior withdrawals are blocked if taking the money out would push the book over the cap. This is the most important thing to understand before depositing junior: your exit can be gated by senior's presence, not only by utilisation.
• A breach from a loss does not force-liquidate anyone. If a write-down pushes the ratio past the cap, the book stops accepting senior and stops opening new borrows until it is back inside. Existing positions run on.

How the rate is split

The book earns one number: borrower interest, minus the reserve factor, as set out on Interest rates. That number is then split. Senior is paid its share first; junior takes whatever is left.

Worked: one book, one year

Illustrative only, to show how the split behaves. A book with $400,000 senior and $100,000 junior at a 4:1 ratio, paying lenders a blended 8%, with the discount set to 25%. None of those inputs is a decided value.

Junior earns 16% instead of 8% because $8,000 a year is transferred from senior. That $8,000 is the price senior pays for a $100,000 buffer. Change the discount and every number in the last two columns moves; the mechanism does not. Whether it is a good trade depends on how often the buffer is used, which is what Risks is for.

The same book, with a loss

Assuming a $5,000 reserve in the book. Senior's first cent of loss arrives only in the fourth row.

When junior is wiped out

A junior share price of zero is not a rounding problem, it is a terminal state for that tranche in that book. The following happens automatically, in the same transaction as the write-down.

1. The junior tranche in that book is closed. No new junior deposits, because there is no positive share price to mint against. Existing junior holders in that book hold a claim worth zero, and it does not recover if the book later earns.
2. New borrows pause in that book, under the same mechanism as the pool-depth pause. Repayments, withdrawals and liquidations all keep working. See Emergency procedures.
3. Senior becomes the first-loss layer for anything further in that book, and is told so: the book is flagged in the interface and in the subgraph, and senior withdrawals are not gated while it is in this state.
4. The book is scheduled for wind-down unless a fresh junior tranche is created and funded. See Delisting and wind-down.

Deposits and withdrawals

The junior cooldown exists for one reason: without it, the first junior depositor to notice a meme wobbling exits at par and leaves the remaining junior holders carrying a loss that was, at the moment they were both in, shared. A cooldown makes that race unwinnable rather than merely unfair.

Open design questions

Published because they are not settled, and because a docs page that hides them is worse than useless to anyone deciding whether to lend.

Bad debt

Sometimes the collateral runs out before the debt does. This page is what happens next: when the loss is recognised, who absorbs it, in what order, and what your share price does the moment it lands.

What counts as bad debt

Bad debt is debt with no collateral behind it. Not a position that is merely underwater, and not a position nobody has liquidated yet: those are unhealthy positions, and they can recover. Bad debt is what is left when a liquidator has taken the last token and the debt is still not zero.

Two numbers, published separately

A book reports both, because reporting only the second is how a protocol looks solvent right up until it is not.

When the loss is recognised

Immediately, in the same transaction as the final liquidation. Not carried, not amortised, not parked in a ledger to be dealt with later.

The reason is that any delay creates a race. If a book has a known hole and the share price still says otherwise, the lenders who withdraw first get out whole and the ones who were slow carry the whole loss. Recognising it on the spot means the share price is true at every block, and everyone still in the book at the moment the collateral ran out shares the loss in proportion.

The order of absorption

Three layers, in this order, all of them inside the book that took the loss. The waterfall diagram is on Senior and junior.

The reserve is a per-book buffer funded out of that book's own revenue, as set out on Fees. It is drawn automatically, in the transaction that realises the loss, with no approval step and no discretion. Its target size is TBD, and whatever that number turns out to be, it is a buffer against small losses and not insurance against large ones.

What it does to your share price

Lender deposits are held as shares in a book, priced against the book's total assets. Writing down bad debt reduces total assets without changing the share count, so the price per share falls. Nothing is taken from your wallet; each share is simply worth less.

A 4.46% write-down on a junior tranche means 1,000 shares that were worth $1,000 are worth $955.40. The shares are still yours and they keep earning; they are earning on a smaller base.

Worked, end to end

The full arithmetic is example 3 on Worked examples. Every figure below is illustrative: the parameters it uses are not signed off.

After a book takes bad debt

1. All buyback routing from that book stops. Every cent of that book's revenue rebuilds the reserve until it is back at target. A book carrying realised bad debt never contributes to a buyback. See The $LOAN Flywheel.
2. The book is flagged in the interface and on chain, with the realised figure and the date.
3. New borrows pause if junior was wiped out. If junior survived, the book keeps running and the pause rule on Emergency procedures applies as normal.
4. The event is written up. What happened, what the parameters were, what we changed. Published whether or not anyone asks.

Oracle and listing

Price comes from a TWAP of the canonical pool, not the last tick. If the pool is empty, new borrows pause. A ticker only gets a book if it passes every listing check.

Why TWAP, not last tick

The last tick is the price of the last trade in the pool. On a memecoin it is cheap to move. If collateral were priced off it, anyone could pump the price, borrow against the fake value and walk, or dump it and liquidate someone else for the bonus.

So every book prices its meme with a TWAP, a time weighted average price: the average price in the pool over a set window. To move a TWAP you have to hold the price away from the market for the whole window, which costs real money and gives everyone else time to trade against you.

Oracle specification

"We use a TWAP" is a decision, not a parameter set. On memecoins the oracle is the single most likely point of failure, so every number that defines it belongs here.

Window and cadence

Cardinality is how many historical observations the pool stores. A pool that keeps too few cannot serve a window as long as we need, and the oracle read reverts rather than silently returning a shorter window. Cardinality is checked at listing and rechecked on every league change; if it is insufficient, the protocol grows it itself and pays the gas, once, at listing.

Staleness

A TWAP read from an accumulator cannot be stale in the usual sense, but it can be built on observations that are all very old, which is the same problem wearing a different hat. A pool nobody has traded in for hours reports a confident average of a price that no longer exists.

On a stale read, in this order:

1. New borrows in that book pause immediately. Nobody can take on debt against a price we do not trust.
2. Liquidations pause too. This is the uncomfortable half of the rule: a stale price is as likely to liquidate someone wrongly as to miss a liquidation, and we would rather carry the risk than seize a borrower's meme on a price we have just declared untrustworthy.
3. Repayments and withdrawals keep working. Both shrink the book's risk, and neither needs a reliable price to be safe.
4. The book un-pauses on its own as soon as a fresh observation lands. No human step, no multisig transaction.

Deviation circuit breaker

A single read that jumps further than a plausible market move is treated as an attack until proven otherwise.

When the canonical pool dies or moves

The canonical pool is named at listing and is the only price source for that book. Memecoin liquidity migrates: a new pool opens on a different AMM, the old one drains, and the book is left pricing collateral off a pool with nothing in it. The procedure is deliberately slow, because a fast pool-swap is an attack vector wearing a helpful face.

1. Detection. Pool liquidity below the book's minimum for a sustained period, not at a single instant. The check is the same one used at listing.
2. New borrows pause. Automatic, no human input. The book keeps accepting repayments, withdrawals and liquidations while the price is still trustworthy.
3. A replacement is proposed through the normal parameter process on Governance and parameters, with the new pool address, its depth, and the price difference between old and new published before anything changes.
4. The timelock runs in full. An oracle source change is in the slowest parameter class. Nobody can repoint a book's price feed quickly, including us, including in an emergency.
5. If no acceptable pool exists, the book is wound down rather than repointed at a bad one. See Delisting and wind-down.

If the pool is drained mid-liquidation

A liquidator seizes meme at the TWAP and then has to sell it. If the pool empties between those two moments the liquidator eats the difference, not the book. The protocol never promises an exit price.

• The seizure itself is priced at the TWAP read in the liquidating transaction, so the book's accounting is settled before any selling happens.
• A liquidator using the callback flow on the liquidator guide finds out inside the same transaction and can simply revert, which costs gas and nothing else.
• What the book loses in this scenario is not money, it is a liquidator: the position stays open, and the book is now in the no-bonus-clears state.

What we do not use, and why

Listing checklist

A ticker gets a book only if every line is true. Not most of them.

Listed means the meme can be used as collateral. It does not mean the meme is safe to hold.

Authorities, and what we check

Chains

Memes from Robinhood Chain, Base, BSC and Solana can live in the LOAN MEME universe. A market only exists on a chain where we have actually deployed, and this table only lists chains that are live.

Liquidations

When your health factor drops below 1, anyone can repay part of your debt and take that much of your meme, plus a bonus. If the meme gaps down too fast for that, the loss stays in that book.

What a liquidation looks like

Three sides are involved: you, the liquidator, and the lenders in that book. Anyone can be the liquidator. No permission, no queue.

1. 01HF drops below 1price falls

   The TWAP updates and your meme is worth less, or interest has pushed your debt up. Your health factor crosses below 1, and from that moment anyone can liquidate the loan.

   Measured onThe TWAP, not the last tick

   WarningNone
2. 02Debt repaidstables in

   A liquidator repays part of your debt in stables. Those stables go back into the book, to its lenders.

   WhoAnyone

   Per callUp to the close factor
3. 03Meme takenmeme out

   In return, the liquidator takes your meme: the value they repaid, plus the bonus. That meme does not come back to you.

   They takeRepaid value + bonus

   Bonus12%, 10 to 15% by book, see the bounds
4. 04Loan recheckedstop or repeat

   If your health factor is back at 1 or above, it stops there. If not, the loan can be liquidated again, straight away.

   HF 1 or aboveIt stops

   HF still below 1Liquidated again

The bonus, and the two bounds it sits between

The bonus is a fixed 12% of the repaid debt, set between 10 and 15% per book. That is one number, and the thing it has to satisfy is not one condition but two, only one of which holds still. It is worth showing the arithmetic rather than asserting that the number is fine.

Write C for the collateral tokens, P for the oracle price, L for the liquidation threshold, D for the debt, R for the amount repaid in one call, and b for the bonus.

The ceiling: a liquidation must leave the book healthier

That is the entire job. If seizing collateral makes the health factor worse, every liquidation drags the position further under and the mechanism is working against itself.

The league thresholds on Leagues and rank are well under 0.80, so the ceiling is unlikely to be what binds. It becomes the binding constraint the moment anyone proposes a high-LTV league, and it is the reason a stablecoin-style 0.90 threshold can never be bolted onto a 15% bonus.

The floor: somebody has to want the trade

The liquidator repays R, sells the seized meme at slippage s, and pays g in gas and financing. Below a certain bonus the trade loses money and no rational bot takes it.

What the chart says

• A fixed 10% stops attracting liquidators once exit slippage passes about 8.6%. The 12% default holds to about 10.3%, and a fixed 15% to about 12.6%.
• Above roughly 14.5% slippage no bonus satisfies both bounds at a 0.85 threshold. The floor has risen above the ceiling and the window has closed. The only remaining exits are a smaller close factor, splitting the liquidation across several blocks, or bad debt.
• A constant sits correctly between a moving floor and a fixed ceiling at exactly one depth. Everywhere else it is either paying more than it needs to, or not enough to get the job done.

The bonus is set at 12% of the repaid debt, with a range of 10 to 15% by book. This chart is the reason it stays a parameter worth revisiting rather than a settled choice.

Fixed bonus, or reverse auction

Both models end with the same debt repaid. They differ in who decides what the liquidation costs: the parameter file, or the bidders.

An auction turns the bonus into a discovered price. It opens at zero and rises until a bidder's profit turns positive, which is precisely the floor from the chart above. Everything between that clearing point and the fixed number is value the fixed model gives away by construction.

The cost of the auction is latency. The position stays underwater while the ramp runs, which is a real risk on a meme that gaps 40% inside a minute: 87 seconds of price discovery can be 87 seconds during which the collateral stops covering the debt at all. Price discovery against time to close is the trade, and it is the decision this protocol has not yet made.

What gets published, either way

Either way, what happens when no bonus clears gets documented rather than left in a footnote. That path ends in bad debt, and it has its own page: Bad debt.

When no liquidator turns up

This is the failure mode that actually costs lenders money, and it is not hypothetical. A meme that has just fallen 70% has wide spreads and thin depth exactly when the protocol needs someone to absorb a large sale.

1. The position sits. It is liquidatable, nobody liquidates, and interest keeps accruing on top. The health factor keeps falling.
2. The shortfall is published as unrealised on the book's book’s status page and in the subgraph. Nothing has hit any share price yet.
3. New borrows in that book pause under the pool-depth rule, so the hole cannot get bigger while nobody can close it. See Emergency procedures.
4. It resolves one of three ways. The price recovers and the position comes back over 1. Depth returns and a liquidator clears it at a loss smaller than the wait cost. Or the book is wound down and the shortfall is realised against the waterfall.

How not to get liquidated

• Borrow well under the max. The gap between what you borrowed and the liquidation threshold is your whole buffer.
• Watch the debt, not just the price. Interest moves your HF on a flat chart.
• Repay early. Partial repayments work and count straight away.
• Watch the league. A demotion tightens the threshold on your open loan. See Leagues.
• Do not borrow to buy more of the meme you posted. When it drops, you lose on the extra meme and get liquidated sooner on the loan.

Liquidator guide

For the people running the bots. How to find a liquidatable position, what to call, what you get paid, and the exact point at which it stops being worth doing.

Why this page exists

Liquidators are not a nice-to-have. A book stays solvent only if someone is watching it and acting within minutes of a health factor crossing 1. If nobody finds it profitable to do that, positions sit underwater, the gap widens, and the loss lands on lenders. Everything on this page is written to make the job easy enough that it gets done.

Finding liquidatable positions

Three ways, in rough order of how fast they are.

1. Read the book directly

Cheapest per call, and the only source that cannot be stale.

// Health factor, 1e18 = exactly 1.0
function healthFactor(address borrower) external view returns (uint256);

// True when healthFactor < 1e18
function isLiquidatable(address borrower) external view returns (bool);

// The most you may repay on this position right now, in loan-asset base units.
// Already has the close factor applied.
function maxRepay(address borrower) external view returns (uint256);

// What you would get for a given repay, before your own swap costs
function previewLiquidate(address borrower, uint256 repayAmount)
    external view returns (
        uint256 collateralOut,   // meme base units you receive
        uint256 bonusValue,      // the bonus portion, in loan-asset units
        uint256 protocolCut      // the slice routed to the book reserve
    );

A multicall over the borrower set of one book costs a single RPC round trip. The borrower set comes from Borrow events, or from the subgraph below.

2. Watch the events

Every state change that can move a health factor emits. Watching these is enough to keep a local index without re-reading the whole book.

event Borrow   (address indexed book, address indexed borrower, uint256 assets, uint256 shares);
event Repay    (address indexed book, address indexed borrower, uint256 assets, uint256 shares);
event Deposit  (address indexed book, address indexed borrower, uint256 collateral);
event Withdraw (address indexed book, address indexed borrower, uint256 collateral);
event PriceUpdate (address indexed book, uint256 twap, uint256 observedAt);
event Liquidate(address indexed book, address indexed borrower, address indexed liquidator,
                uint256 repaid, uint256 seized, uint256 bonusValue);
event BadDebt  (address indexed book, address indexed borrower, uint256 amount);
event BorrowsPaused (address indexed book, uint8 reason);

PriceUpdate is the one that matters. Between two price updates no position in that book can become liquidatable on price, only on interest, and interest is predictable. Recomputing the whole book on each PriceUpdate is the simple correct strategy.

3. Query the subgraph

Easiest to start with, one block behind at best. Endpoint TBD.

{
  positions(
    first: 100
    orderBy: healthFactor
    orderDirection: asc
    where: { healthFactor_lt: "1000000000000000000", debtShares_gt: "0" }
  ) {
    borrower
    book { id  ticker  league  liquidationThreshold  liquidationBonus }
    collateral
    debt
    healthFactor
  }
}

The call

function liquidate(
    address borrower,
    uint256 repayAmount,     // or type(uint256).max for the full close factor
    address collateralTo,    // where the seized meme is sent
    bytes calldata data      // empty: you pay up front. non-empty: callback, see below.
) external returns (uint256 repaid, uint256 seized);

With empty data, the book pulls repayAmount of the loan asset from you first and sends the meme after, so you need the stables in hand and an allowance set.

The callback flow, so you do not need a flash loan

Pass non-empty data and the order reverses: the book sends you the seized meme, calls you back, and only checks it has been repaid when you return. Sell inside the callback and repay from the proceeds.

interface ILoanMemeLiquidateCallback {
    function onLoanMemeLiquidate(
        uint256 repaid,   // what you must have transferred back by the time you return
        uint256 seized,   // meme base units already sitting in your contract
        bytes calldata data
    ) external;
}

This removes the flash-loan fee from your cost model entirely. A flash loan still works if you prefer it, from anywhere; the protocol does not offer one and does not charge for one. See Fees.

How much you can repay

The close factor caps a single call. It is not a fixed fraction: a position that is only slightly underwater gets a partial liquidation, and one that is badly underwater can be closed in full, because leaving a dust position behind helps nobody.

Both the fractions and the health factor they switch at are TBD. What is settled is the shape: a partial close near 1, a full close well below it, and a hard cap once the collateral runs out. maxRepay() always returns the live answer, so read it rather than reimplementing this table.

The profitability model

You repay R, receive collateral worth R(1 + b) at the TWAP, sell it at slippage s, and pay g in gas and any borrowing cost.

Assuming g at 0.5% of the repay. Above the break-even column the trade loses money and a rational bot does not take it, which is exactly when a book most needs one to.

Gas, roughly

Nothing is deployed, so nothing has been measured. These are published as soon as there is a testnet to measure them on.

What we will and will not do about MEV

A fixed bonus makes liquidation a race for transaction ordering rather than a contest on price, so most of the bonus on a competitive book ends up with the block builder rather than with you. We are not going to pretend otherwise.

• Liquidation is permissionless and stays that way. No allowlist, no keeper set, no private mempool arrangement with a favoured party.
• There is no protocol-run bot. If we ran one we would be competing with the people we need to show up.
• The incentive model itself is open. The case for replacing the fixed bonus with a reverse auction, which would move the competition from ordering back to price, is set out in full on Liquidations. It is not decided.

Reference implementation

A minimal working bot, MIT licensed: event listener, position index, profitability filter, callback contract and a simulation harness. It is a starting point, not a competitive edge, and it is deliberately not optimised.

Delisting and wind-down

Books get closed. This page is what causes it, what happens to your open loan or your deposit while it is happening, and how long you get.

What closes a book

Two of these are automatic and the rest are decisions. All of them are announced before anything changes.

The wind-down procedure

A book is never switched off. It is deprecated, and then it drains. The sequence is fixed, and at no point in it are you prevented from repaying and taking your meme back.

1. 01Announcednotice starts

   The book is marked deprecated in the interface and on chain. The reason is published at the same time, not later.

   Notice periodTBD

   Borrow capSet to zero immediately
2. 02Frozenno new anything

   New borrows and new collateral deposits stop. Lender deposits stop. Everything that shrinks the book stays open.

   Still openRepay, withdraw, liquidate

   ClosedBorrow, deposit
3. 03Rates riserepay pressure

   The book's rate curve is switched to a wind-down curve with a much steeper slope, so that carrying an open loan gets expensive on purpose. This is the only push there is: nothing force-closes a healthy position.

   Applies fromEnd of the notice period

   Liquidation thresholdUnchanged
4. 04Drainslenders out

   As borrowers repay, stables free up and lenders withdraw. Utilisation still gates withdrawals: a deprecated book with loans still open is not a book you can exit instantly.

   WithdrawalsFirst come, as liquidity frees

   QueueNone. There is no priority list.
5. 05Closedor left open

   Once debt reaches zero and lenders have withdrawn, the book is done. If loans remain open indefinitely, the book simply stays in this state: it keeps accepting repayments and liquidations forever, because a contract that stops accepting repayment would trap collateral. Nothing is ever self-destructed, and the reserve stays with the book until the last loan closes.

League demotion, which is not a delisting

A demotion moves a book down a league and tightens its terms. The book stays open. It is a much more common event than a delisting and it is the one most likely to affect an open position.

The mechanics of a demotion, including why there is no grandfathering, are on Governance and parameters.

Appeals and relisting

If you are a lender in a book being wound down

• Your deposit is not locked, but it is not liquid either. You can withdraw whatever share of the book is not currently lent out, the same as any other day. The difference is that no new deposits are arriving to free up more.
• Your yield usually goes up first. The wind-down rate curve is steeper, so while borrowers are still repaying you are earning more on a shrinking base.
• The last lenders out wait the longest. If a borrower never repays and is never liquidatable, their loan can outlast your patience. There is no mechanism that ends it.
• Junior still sits in front of senior throughout, and the waterfall on Bad debt is unchanged by a wind-down.

Governance and parameters

Who can change what, how long it takes, and what you can do about it in the meantime. There is no DAO and we are not going to pretend there is one.

Who holds what

Parameter classes and their delays

The rule is simple: the more a change can hurt someone with an open position, the longer it takes. Anything that can make an existing loan liquidatable sits in the slowest class. The classes are settled; the delays attached to them are not.

The asymmetry in the fast class is deliberate: a change can be made instantly only if it can make the protocol safer and cannot make an existing position worse. Lowering a borrow cap stops new risk and does nothing to an open loan. Raising one takes the standard delay.

League demotion, specifically

This is the change most likely to affect you personally, because it tightens the liquidation threshold on a loan you already have open. It is treated as a slow change with an extra step.

1. The demotion is announced the moment it is queued, in the interface and on chain.
2. A notice period runs before the timelock even starts. Length TBD, so the total warning is the notice plus the slow-class delay.
3. The new threshold applies to every open loan in that book when it takes effect, not only to new ones. There is no grandfathering, because a book with two liquidation thresholds in it is a book nobody can reason about.
4. During the notice period the borrow cap is lowered immediately under the fast class, so the book does not take on more exposure at terms that are about to change.

Upgradeability

The change log

Every executed parameter change gets a row here: what changed, from what to what, when it was queued, when it executed, and why. Including the changes that make the protocol look bad.

Emergency procedures

What pauses a book, what keeps working while it is paused, who can press the button, and what they cannot do with it.

The pause rule

A pause stops new borrows in one book. Nothing else, and never across the whole protocol. These are the exact conditions, and all but the last are automatic: no human decides, and no human can stop it happening.

What still works while a book is paused

A pause is not a freeze. Every action that reduces risk stays open, in every pause condition above, without exception.

Pause, freeze, deprecate

Three different things that get used interchangeably elsewhere. Here they mean exactly this:

The guardian, and its limits

One address exists that can act without a timelock, because some failures are faster than a timelock. It is deliberately the weakest emergency power that is still worth having, and its limits are enforced by the contract rather than by policy.

Guardian address: TBD, published before any book opens.

The $LOAN Flywheel

The meme is the culture. The protocol is the engine. Protocol revenue buys back the meme. You do not need to hold it to lend or borrow.

What the token does

It does

The Flywheel

1. 01Fees build upstables in

   Fees collect in stables inside each book: the 2.5% fees on deposits, withdrawals, borrows and repayments, the 20% slice of borrow interest, and 10% of the liquidation bonus. There is no listing fee and there never will be. Every fee is on Fees.

   FromFees, interest and liquidations

   Held inStables, per book
2. 02The reserve comes firstreserve

   The book's reserve is served before anything else. A book with bad debt sends nothing to buybacks.

   Paid firstThe book's reserve

   Bad debtNothing to buybacks
3. 03The rest is movedto buyback

   What is left goes, on a schedule, to the buyback contract.

   Goes toThe buyback contract

   ScheduleTBD
4. 04It buys LOAN MEMEopen market

   The buyback contract buys the token on the open market. No private deals.

   BuysLOAN MEME

   WhereOpen market only
5. 05Every step is publishedon chain

   Each step above goes out with its transaction hashes, so anyone can check it on chain.

Risks

Everything that can cost you money here, in one place, including the ones we cannot do anything about and the ones that are our fault.

Protocol and contract risk

Oracle risk

Liquidation and MEV risk

Market and asset risk

Chain and infrastructure risk

Risks that are about you, not us

• Phishing and fake sites. The realistic way to lose money here is not an exploit of our contracts, it is a convincing copy of this website, or a token carrying the same ticker at a different address. Nothing is deployed, so there is no genuine address to compare against yet, which makes every claim about one false by definition.
• Approvals. A token approval you granted to something else can drain the meme before it ever reaches a book.
• Key loss. There is no recovery, no support, and no admin who can move a position to a new address.
• Borrowing to buy more of the meme you posted. When it drops you lose on the extra meme and get liquidated sooner on the loan. It is the most common way people lose everything here and it is entirely self-inflicted.
• Tax. A liquidation may be a taxable disposal where you live. We do not report anything to anyone and we cannot advise you.

Risks we do not carry

Listed because knowing what is not a risk is part of understanding the shape of the thing.

• Contagion between books. A book going bad costs its own lenders and nobody else. No book can be liquidated to cover another.
• Rehypothecation. Deposited memes sit in the book's locker. They are not lent out, staked, or used to farm anything.
• A run on the protocol. There is no shared pool to run on. A book at full utilisation gates its own withdrawals and that has no effect on any other book.
• $LOAN being liquidated. The token is not collateral in any book, so a fall in it cannot trigger anything in the lending markets.

What makes us different

How we compare with Aave, Morpho, Euler and Silo, in short. Including where they are better.

At a glance

What is actually ours

Three things. The rest is a variation on something another protocol shipped first.

• Memecoins as collateral. Big lenders will not accept a two-week-old meme. We accept that it can go to zero and build around it: low LTVs, a slow average price, borrow caps tied to pool depth, and one book per token so a collapse stays contained.
• Ordered losses inside each book. Junior lenders take the first loss and earn more for it. Senior lenders earn less and are hit last. Almost no DeFi lender does this. See Senior and junior.
• Leagues instead of case-by-case rules. Each token is scored on on-chain data and placed in T1, T2 or T3, which sets its terms. Cruder than a risk team, but it works on a meme that launched yesterday. See Leagues and rank.

Where we are weaker

• The price is our biggest risk. A thin pool can be pushed, and the average lags in a crash. No professional feed exists for new memes. See Oracle and listing.
• No track record. Nothing is deployed. Aave has survived years of attacks.
• A small safety net. Each book's reserve only covers small losses. There is no big backstop behind it.
• Liquidations are not settled. A fixed bonus can be too low to attract liquidators or cost the borrower more than needed. An auction is still an option. See Liquidations.
• Senior and junior is new. There are few precedents to learn from, and it ships only if it can be built safely.
• Less flexible. Aave lets you borrow against several assets at once. Here every meme stands alone, which is the price of containing losses.
• A trusted team. A multisig controls the settings. The timelock slows it down, it does not stop it.

Is it for you?