Ether.fi Liquid ETH - Risk Assessment
Liquid ETH Yield is an actively managed ETH yield vault built on top of eETH with a current TVL of $561.63M and 14-days yield of around 5.9%. Users deposit ETH, eETH or weETH and receive an ERC-4626 vault token that is backed 1:1 by eETH and accrues value via redemption price, instead of rebasing. Behind the scenes the vault allocates across multiple chains and a variety of DeFi protocols, including Aave v3, pendle, morpho, balancer, aura, uniswap v3, and convex, among others, where looping may be used as strategy.
The strategy is managed by Seven Seas, and the infrastructure provided by veda.tech.
Backing
LP tokens represent a pro rata claim on an eETH denominated portfolio managed by the vault. The contract specifies that LP tokens are backed 1:1 to eETH and priced using the external eETH market price, with the share price moving based on the performance of the underlying strategies such as PT eETH and LST leverage loops.
At any point in time the vault holds a mix of positions within the approved strategy set that are either directly in eETH or economically linked to eETH. As allocations between those positions change over time, the effective risk profile is driven by both the Ether.fi LRT stack and the downstream protocols used by the strategy, rather than by a static pool of idle eETH (for more see here).
Collateral breakdown
According to the technical documentation here, the vault is allowed to use the following protocols and positions via adaptors:
Pendle: YT-eETH, PT-eETH, LP-eETH
Aave v3: ETH lending
Morpho: weETH supply and borrow
Balancer: weETH/rETH, weETH/ETH pools
Aura: weETH/rETH pool
Uniswap V3: weETH/ETH 1 bps and 5 bps pools
Convex: weETH/WETH pool
However, currently the vault has exposure to the following protocols, according to their app ether.fi/app/liquid/eth:
By digging into the contract addresses of the vault:
Core vault contract address: 0xf0bb20865277abd641a307ece5ee04e79073416c (~$379M)
Subvault contract addresses:
- 0x778ac5d0ee062502fadaa2d300a51de0869f7995 (~$69M)
- 0x7c391d7856fcbc4fd3a3c3cd8787c7ebf85934af (~$15M)
- 0xba7fdd2630f82458b4369a5b84d6438352ba4531 (~$28M)
- 0xe29dee8c6805314b9abc4dbf07b7b314ea0f2e77 (~$32M)
The sum of all asset holdings is $523M, which is about 93% of the TVL shown on their app that is easily verifiable onchain, where the remaining smaller allocations could be observed on their Dune dashboard, under the label “Unlabelled Protocols”, which hinder transparency, and makes it already a red flag to recommend this strategy for onboarding.
According to deBank, we observe multi-chain exposure, being on Ethereum mainnet (78% share), Plasma (10%), and Base (3%) as the largest cases, with highest allocation being into Aave v3, which we explore more in detail as follows:
Aave v3 (~ $290M)
By considering the supply and borrowed values of the order of billions on mainnet, and the high variable yield, we can assume the presence of looping positions, which already poses additional risk leading to negative yield if supply and borrow rates change unfavorably. Additionally, the case on plasma being also unclear, by supplying weETH, borrowing USDT0 and then using another vault contract to supply USDT0. This trace of allocation also raises flags on the difficulty to follow the strategy behind.
Mainnet:
Plasma:
Base:
Assets on Reserve protocol could not be verified, as they do not appear on the Dune dashboard nor in DeBank, but we do find small traces of RSR tokens on the address 0x778ac5d0ee062502fadaa2d300a51de0869f7995, which by looking for transactions on Reserve protocols, there are two:
Which could potentially explain the mismatch. Also, there is a mismatch on the TVL shown on the app with the one shown on the Dune dashboard, suggesting that the UI may not be synchronized.
Other larger positions, include Uniswap V3, ether.fi Liquid Katana ETH (depositing $32M and bringing visible exposure to vbETH), $15M on cUSD, Fluid supplying around $11M and borrowing $2M of stETH. Current holdings also include ETH+ (around $68M), while supplying Spark around $60M and borrowing $32M of PYUSD. Thus a non-exhaustive list of new collaterals will include: USDT0, cUSD (new), stETH, weETH, rETH, ETHx, sfrxETH, ETH+, PYUSD, and RLUSD (new). Other collateral will be found once we unrolled the full lending positions. Although the current fleets already have exposure to many of those yield sources, we are unable to verify the full collateral exposure, and since this strategy seems to aggregate other yield sources, we see some similarities to FluidLite, which is already onboarded, but with a much more risky approach and less transparency.
Note:
We are not going into more details here into collateral exposure as the strategy is very complex and even integration into the risk model (required for parameters computation) would take a considerable amount of time. However, due to the complexity of the strategy, looping strategies involved, and the cross-chain borrow/lending strategies, plus queuing of withdrawals and the 0.5% platform fee, as we will describe later, with even certain portion of the backing including directional bets (for which we couldn’t verify any onchain hedging positions) we advise not to onboard this yield source even on the Higher Risk fleet, at this time. When backing is more transparent, no directional bets are present, and if/when liquidity risk is reduced, we are happy to revise our recommendation.
Fees
Liquid ETH Yield charges an annualized platform fee on the vault TVL, computed daily by the smart contract, not a performance fee on yield. The schedule currently specifies: Liquid ETH Yield – platform fee 0.5 percent per year (source), so each day 0.5 percent / 365 of TVL is accrued in the contract as platform fee.
No additional performance fee is listed for this vault at the time of this writing.
Redemptions and liquidity
Withdrawals are handled via a Withdrawal Queue. Thus, users submit a request with a number of shares, a deadline, and a discount. Off-chain solvers fill the request if they can source underlying assets within the requested price constraints.
If the LP token price falls below the target during the pending period, the solver may not fill and the request expires.
Key implications:
- Exit depends on solver participation plus underlying liquidity across Pendle, Aave, Morpho, LPs.
- No atomic “in and out in one tx” is possible, which is intentional to avoid round-trip arbitrage.
Secondary DEX liquidity in liquidETH itself is thin compared to underlying positions:
so practical liquidity is via the withdrawal mechanism and unwinding strategies.
Governance and roles
Roles for Liquid ETH Yield:
- Admin: multisig with strategy team plus ether.fi community, can act quickly in emergencies.
- Timelock: owns the Registry and gates non urgent changes, such as adding new positions. Currently read as 72h from the EtherFiTimelock smart contract.
- Strategist: can only rebalance within pre-approved adaptors and positions.
This constrains strategy changes but still allows allocations and risk profiles to evolve over time, with changes visible via timelock for new adaptors/positions.
Risk assessment
Liquid ETH Yield runs a very complex, multi chain, multi protocol strategy on a TVL of about $560M, with roughly $523M easily verifiable across the core vault and four subvaults and the rest only visible on Dune. The largest single bucket is Aave v3 at around $290M, where looping and cross protocol flows (for example weETH supplied, USDT0 borrowed and redeployed via another vault) materially increase leverage and rate sensitivity. According to DeBank, exposure is spread mainly across Ethereum mainnet (~78 percent of value), Plasma (~10 percent) and Base (~3 percent), adding operational and unwind complexity.
There are observable mismatches between the app TVL, onchain balances and Dune, and Reserve related exposure cannot be cleanly reconciled, which makes full external verification of backing difficult. Practical exits depend on the withdrawal queue and off chain solvers, with thin secondary DEX liquidity for liquidETH. Governance is concentrated in the Admin, Timelock (72h) and Strategist roles, and while the platform fee is relatively low at 0.5 percent per year on TVL, it does not mitigate the structural, leverage, cross chain and liquidity risks embedded in the current allocation.
Conclusion
Liquid ETH Yield is a large and operationally complex ETH yield vault, with around 560M TVL, roughly 523M traceable across the main vault and subvaults and a dominant allocation on Aave v3 combined with looping and cross protocol strategies across mainnet, Plasma and Base.
Together with the use of a withdrawal queue, thin secondary liquidity for liquidETH, and observable data mismatches between UI, onchain views and Dune, this profile is deemed as not suitable for lower risk fleets, while BA Labs proposes refraining from listing it in Higher-Risk ETH fleets as well due to significant risks present, compared to the Fluid Lite ETH v2 vault ARK which is the only looped/managed ETH type of ARK currently onboarded to SummerFi protocol, since Liquid ETH implies more complex, less transparent, multichain underlying strategies that are effectively backing the vault receipt token.
We’d be happy to revise the proposal if/when the risk profile of this ARKs improves on the metrics we’ve stated in the above analysis.
