Aave’s risk adviser proposes freezing osETH, LBTC and rETH on V3 Core, and StakeWise and Rocket Pool push back

LlamaRisk, the risk adviser funded in part by the Aave DAO, has proposed freezing osETH, LBTC and staking means depositing a proof of stake coin with a protocol that stakes it for you and issues a transferable token (such as Lido stETH or rETH) representing the deposit plus rewards, which you can use or sell while the coins stay staked.">rETH as collateral on Aave V3 Core, together with a handful of smaller reserves. The proposal, posted on the Aave governance forum on 10 November at 19:33 UTC, also covers rsETH and tBTC, which are already frozen. On 9 November the assets in scope held $1.33 billion of collateral backing $1.03 billion of debt, according to LlamaRisk’s figures.

Nothing has been voted. The post is an ARFC, a request for comment and the discussion stage of Aave governance, and it gives no date for a vote. Two issuers named in it have already asked to be taken out of the package: StakeWise, whose osETH is the second-largest position in scope at $279.4 million, replied on 11 November, and Rocket blockchain that runs exactly as written when it is called, holding and moving funds according to its code, without an intermediary able to change the outcome.">smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">Pool, the issuer of rETH, early on 12 November.

A freeze would not liquidate anyone. It stops new deposits of an asset and new borrowing against it, and sets its loan-to-value ratio (LTV) to zero, including in the E-Mode categories that allow higher leverage between correlated assets. Existing positions stay open, and the proposal says holders can repay, withdraw or move their collateral at any time.

No hack or depeg sits behind the proposal. LlamaRisk cites no incident at StakeWise or Rocket Pool. Its case is about which collateral Core, Aave’s main Ethereum market on V3, should keep, and who earns the extra yield that riskier collateral brings.

What a freeze would do to existing positions

For someone looping osETH, the freeze would end new borrowing against it, since the E-Mode LTV would drop to zero. The position itself would not be touched. The liquidation threshold would stay where it is, so the freeze alone would not bring anyone closer to liquidation; the health factor and liquidation price of an open position would be calculated as before.

The exit LlamaRisk offers is a swap inside the same E-Mode. The ETH-correlated category already accepts wstETH, weETH, cbETH and WETH, so an osETH or rETH loop could replace its collateral without repaying the WETH debt first. For rsETH, the proposal adds wstETH, weETH and cbETH to the rsETH E-Mode at 93% LTV and a 95% liquidation threshold, to allow the same move.

Borrowers using LBTC or rETH as collateral for stablecoin loans would keep their loans and could repay them on their own schedule. They could not borrow more against those assets on Core. LlamaRisk expects LBTC positions to move to Aave V4 or be repaid.

People who lend WETH or stablecoins on Core are not directly affected. The change that hits lenders is narrower: every frozen reserve that is borrowable and still carries more than $20,000 of debt would see its reserve factor raised to 99%. Borrow rates would stay the same, but the interest would go to the DAO instead of suppliers. That covers CRV, tBTC, crvUSD, FBTC, UNI, mUSD and rETH, whose reserve factor would rise from 15%. Someone earning interest by lending rETH on Core would see that income almost disappear.

What LlamaRisk proposes for Aave V3 Core collateral

LlamaRisk titled the thread “Repositioning Aave V3 Core Around Established & Liquid Collateral”. The starting point is the way Core is built: liquidity is pooled across every listed asset, so a supplier of WETH or of a stablecoin underwrites every collateral asset at the same pooled borrow rate. LlamaRisk writes that a growing share of that collateral now depends on layers outside the token itself, such as restaking services, leveraged strategies, cross-chain bridges and custodians.

The risk adviser adds that the reward for carrying those layers does not reach the people who lend. In its words, “the additional spread these assets generate accrues to the borrowers looping them rather than to the suppliers”.

After the change, Core would keep WETH, the liquid staking tokens wstETH, weETH and cbETH, WBTC and cbBTC, the stablecoins USDC, USDT, GHO and DAI, plus XAUt, AAVE and LINK. USDe, sUSDe and syrupUSDT would stay too, but remain subject to the same criteria. Everything else is meant to go to Aave V4, where liquidity sits in hubs, collateral sits in spokes with their own caps, and a collateralRisk parameter adds a premium to each borrower’s rate according to the collateral behind the loan.

LBTC is the only asset the thread gives a destination of its own: it is expected to continue on Aave through a dedicated spoke on the V4 Ethereum Core Hub. No date is given for that spoke.

The six kinds of exposure Core would leave

LlamaRisk lists the types of collateral it wants Core to move away from. Readers new to the vocabulary can start with our explainer on restaking and liquid restaking tokens.

• Restaking, where an asset’s backing is delegated to secure other services or networks, as with liquid restaking tokens.
• Leveraged wrapping, where most of an asset’s backing is itself a leveraged position in another asset, so both fall together under stress.
• Derivatives in the backing that go beyond hedging, such as covered-call writing, whose results flow into the token’s exchange rate.
• Custom non-custodial bridges, where the backing of a wrapped asset is held and released by a signer or enclave network built for that asset alone.
• Issuers whose security and operational resources are small compared with the size of their asset on Core, as measured by the Aave Risk Framework.
• Reserves with limited demand, whose exposure is too small to justify the cost of maintaining them, whatever the quality of the asset.

Nowhere does the thread say which of these criteria osETH, LBTC or rETH fails. It groups the assets by size of exposure and by the way they are used, not by criterion.

$1.33 billion in scope, and most of it sits in two ETH loops

The table below uses LlamaRisk’s figures as of 9 November. Debt is attributed to each collateral asset pro rata to each borrower’s collateral mix, so the “debt backed” column is the adviser’s own allocation, not a figure read directly on-chain. Revenue is the DAO’s annual income at the average borrow rates of the past three months, before any migration.

Two groups make up almost all of it. rsETH and osETH back $862 million of WETH debt, which LlamaRisk says is almost entirely leveraged loops in ETH-correlated E-Modes. LBTC, tBTC and rETH back $160 million, mostly in stablecoins, spread over several hundred addresses at moderate leverage.

The revenue at stake is small for a market of Core’s size: about $3.7 million a year for the DAO, gross. LlamaRisk notes that the figure ignores any capital that would replace the departing loans. The mUSD stablecoin, borrowable but not collateral, would also be frozen.

Why osETH is in the Aave freeze proposal

osETH is StakeWise’s liquid staking token. On Core it has no borrowing power outside E-Mode, and a 93% LTV inside the ETH-correlated E-Mode, where nearly all of its $279.4 million of collateral backs $256.6 million of WETH debt. LlamaRisk puts the leverage at around 14 times. Supply has reached 122.9k osETH, close to the 123,000 cap, and the ten largest positions back 66% of the debt.

That cap was itself set on LlamaRisk’s advice. In a Risk Stewards post on 15 June, the adviser recommended raising it from 110,000 to 123,000 osETH, when the reserve was 94.4% full. The top osETH suppliers then had a median health factor of 1.02, with WETH making up nearly all the debt.

A loop works like this: a user deposits osETH, borrows WETH against it, turns the WETH into more osETH and deposits again, repeating until the position is many times larger than the starting stake. The trade earns the difference between staking rewards and the WETH borrow rate. It also leaves very little room before liquidation, which is why LlamaRisk rules out the usual tool of lowering the liquidation threshold (LT) for osETH and rsETH.

Its explanation is blunt: at 12 to 14 times leverage, “an LT reduction of 2 to 5 points would make most of the collateral liquidatable at once”. osETH also shares its E-Mode with other ETH collateral, so the category threshold cannot be cut for osETH alone.

The forum thread does not discuss the token’s price. DefiLlama priced osETH at $2,199.81 at 11:37 UTC on 12 November, while ether was near $2,024 around 11:41 UTC, down 15.5% over seven days. osETH sits above ether because staking rewards build up in its exchange rate.

The follow-up steps if positions do not unwind

The proposed freeze is the first of three stages. In the second, LlamaRisk plans to work with each issuer and with the largest borrowers on moving positions to collateral that stays on Core or to Aave V4, and to report progress every month. No further step is taken against an asset while its exposure keeps falling.

If it does not fall, the ARFC pre-approves two tools, to be used through the Risk Stewards or through Direct-to-AIP proposals. For LBTC, tBTC and rETH, the liquidation threshold could be cut in steps of up to 2 percentage points, each sized so that collateral becoming liquidatable stays below $2 million or 5% of what remains. LlamaRisk calculates that these positions have room: a 10-point cut would leave less than $0.1 million of LBTC collateral liquidatable, rETH absorbs 21 points with less than $1 million, and tBTC absorbs 7 points with $0.1 million.

For the rsETH and osETH loops, the last resort is the WETH borrow rate on Core. Its base variable rate could rise by up to 100 basis points, in 50-point steps no more than once a month, enough to make the loops lose money so that they unwind on their own. A matching rise in the WETH reserve factor would keep the WETH supply rate unchanged, and a Merkl campaign would rebate the increase to every WETH borrower who holds neither rsETH nor osETH. LlamaRisk says each step would be announced on the forum first and removed once the loops are gone.

StakeWise and Rocket Pool ask to be judged on their own

StakeWise’s founder, Dmitri, replied on 11 November at 14:02 UTC. He wrote that StakeWise has run liquid staking since January 2021 with no hack and no validator slashing, and that osETH has been collateral on Aave for two and a half years. He said StakeWise built its leverage product with its own immutable contracts, in which borrowed ETH is returned by validators instead of being swapped on decentralized exchanges, unlike approaches that rely on third parties such as DeFi Saver or Mellow.

His request: “we respectfully ask LlamaRisk and the Aave DAO to move the proposed osETH freeze to a separate vote”. He added that he is confident osETH can reach one of the highest scores under Aave’s Risk Framework.

Rocket Pool’s general manager, Darren, posted at 03:08 UTC on 12 November. He said Rocket Pool supports focusing Core on established collateral, but that rETH belongs on that side of the line: “rETH meets none of the structural criteria.” He described rETH as a plain liquid staking token backed by staked ETH, with no restaking, leveraged wrapping, derivatives or bridge.

His other arguments are about demand and revenue. rETH backs around $82 million across 317 borrowing positions, he wrote, with roughly nine times as many tokens supplied as cbETH, and has sat at its 35,000 supply cap since 6 November. Its loans are mostly stablecoin borrowing at moderate leverage, which he says earns the DAO roughly twice the revenue per dollar of debt of the looped positions. He also pointed to the Saturn 2 upgrade, which adds a direct redemption path, and to a Security Council that can pause key functions. Our guide to liquid staking tokens such as stETH and rETH explains how these tokens differ.

LlamaRisk’s own figures do not contradict Rocket Pool on how rETH is used: its table shows rETH mostly backing USDC and USDT loans, not E-Mode loops. The thread had not received a reply from LlamaRisk to either post at 11:50 UTC on 12 November.

Earlier freezes on Core: rsETH in April, tBTC on 31 October

The proposal continues a clean-up that started earlier this year. A Low Adoption Asset Deprecation ARFC wound down 71 reserves across eleven deployments and six whole markets, about $98 million of supply. An Oracle Deprecation ARFC replaced the price feeds of long-tail assets with fixed values. Those rounds gave eBTC, BTC.b, tETH and the governance tokens a three-month exemption, which this proposal ends.

rsETH was frozen in April, after a suspected exploit involving Kelp’s token. On 18 April, the Aave Guardian froze rsETH and wrsETH markets on every deployment where they were listed, starting at 18:52 UTC. Since then rsETH supply on Core has fallen 45%, from 525.8k to 290.7k tokens, according to LlamaRisk; the remaining debt sits in eleven positions above $0.2 million, and the three largest hold 69% of it.

tBTC has been frozen on Core since 31 October, after Threshold paused its bridge, as we reported when the tBTC bridge came back but Aave’s freeze held. Its supply has fallen to 1,599 tBTC, the lowest in a year, across 309 borrowing positions. On other V3 markets, rsETH and wrsETH on Arbitrum, Base and Linea, and rETH and tBTC on Arbitrum, are already frozen. BTC.b stays available on Avalanche, where it is the main bitcoin asset.

What is still open

No date, form or quorum for a vote appears in the thread. Its only wording on timing is that, after approval, the initial measures will be submitted as an AIP, Aave’s on-chain proposal format. Whether LlamaRisk splits osETH or rETH into separate votes, as both issuers ask, is not known.

Nor is it clear how quickly the loops would unwind. The ten largest osETH positions hold two thirds of the debt, so a few decisions by large borrowers could change the picture quickly, in either direction.

LlamaRisk discloses that it is funded in part by the Aave DAO and says it received no payment from the protocols it reviewed. If the ARFC passes, its next deliverable is the first monthly report on how much of the $1.33 billion has moved.