Executive summary
In this proposal, we ask for community support to deploy 400 osETH from the StakeWise DAO Treasury into the new osETH-ETH / wstETH-ETH pool on Fluid, creating up to 4,000 ETH in protocol-owned liquidity for osETH at a low cost for the DAO.
If this initiative is successful, StakeWise can restore vast ETH liquidity reserves for osETH, dramatically lower the cost of liquidity for the protocol, and free up more revenue to direct towards other purposes (e.g. SWISE buyback).
Motivation
Recent Balancer V2 hack and consequent outflow of assets from Balancer has prompted the team to seek alternative venues for deploying ETH liquidity for osETH.
One such option is Fluid Protocol - a hybrid between a lending market and a DEX, where assets supplied and borrowed in the lending market can be used for routing trades between assets, earning trading fees that offset a significant portion of the borrow costs, and can be amplified akin to StakeWise Boost to enhance the total amount of liquidity available to the market.
Let’s unpack all of this.
At the foundation of Fluid is what’s called a “liquidity layer” - a single system containing all of the assets supplied to Fluid by different protocol participants.
Assets land into this liquidity layer and leave it via 5 different actions:
-
single-sided Lend: users supply a single asset to earn a supply APY; this asset can be borrowed by other users of the platform (against some collateral, like in Aave)
-
single-sided Borrow: users borrow a single asset and pay a borrow APY; they must have provided some collateral to borrow
-
two-sided Smart Collateral: a deposit of two assets that can be traded against each other (e.g. osETH and ETH), to earn a trading fee from the swaps between these assets by other users (e.g. when CoWSwap users want to sell/buy osETH for ETH), and a supply APY on each asset individually (e.g. when Fluid users borrow ETH).
Users who provide Smart Collateral risk not being able to withdraw assets in the same proportion as they deposited them (e.g. deposited 50/50 osETH-ETH, but can withdraw only in 70/30 proportions).
- two-sided Smart Debt: a borrow of one or two assets that can be traded against each other (e.g. wstETH and ETH), to earn a trading fee from the swaps between these assets by other users, and pay a borrow APY on each asset individually (e.g. wstETH borrow APY when borrowing wstETH).
Users who borrow Smart Debt risk not being able to repay assets in the same proportion as they borrowed them (e.g. borrowed 50/50 wstETH-ETH, but can repay only in 70/30 proportions).
- single-sided Smart Lending: users supply a single asset to earn a supply APY, and a trading APY from the asset being used to facilitate swaps within the platform.
StakeWise DAO can recursively deposit osETH-ETH as Smart Collateral and borrow wstETH-ETH as Smart Debt to create deep osETH-ETH liquidity via the following mechanism:
- Deposit osETH-ETH collateral on Fluid
- Borrow wstETH-ETH
- Swap ETH for wstETH on the fly within Fluid
- Swap all wstETH into ETH on another exchange
- Deposit half of ETH into StakeWise to get osETH
- Deposit osETH-ETH collateral on Fluid
- Repeat 2-6 recursively until capacity is reached
Since all assets supplied to the platform are shared among all pools and lending/borrowing markets via the unified “liquidity layer”, StakeWise DAO would effectively be adding 2 units of new ETH into Fluid’s liquidity layer for every 2 units of wstETH borrowed, with the intention of expanding the ETH supply on Fluid to make headroom for large swaps for osETH.
This design is unique and enables new primitives like earning trading fees on the borrow position, which offsets a portion of the borrow cost to make borrowing cheaper. For example, borrowing wstETH-ETH costs 2.92% per year, but when offset by the 1.13% trading fee APY earned by borrowers, the net borrow cost drops to 1.79%. These are real numbers from the Fluid dashboard as of November 26th.
This opens up the possibility for StakeWise to earn from deploying liquidity on Fluid; not pay for it e.g. supplying osETH-ETH at ca 2.3% blended rate (expected) would yield a positive carry with every loop. Pretty crazy.
While the possibility is there, this is not our base case. Since this strategy relies on extensive use of wstETH capacity on Fluid, which is already at high utilization due to other markets utilizing wstETH for similar or adjacent purposes, we can expect StakeWise DAO to roughly breakeven, or pay a marginal cost for the liquidity we plan to establish.
Still, the numbers back up the rationale for using Fluid as core liquidity venue: even in a pessimistic scenario we draw up, the cost savings are ca 50% of what we used to pay for liquidity on Balancer per month (14.65 ETH vs 6 ETH in the pessimistic scenario).
If everything goes well, however, the Treasury will roughly break even on its deployment, saving ~15 ETH per month (and making 4 ETH extra per month in the optimistic scenario!).
If our experience with running Boost and projecting performance using historical figures is of any value here, we’re likely going to see the base case holding up for most of the time, flipping to the pessimistic scenario during times of market stress. That would mean a slightly negative liquidity cost, which is a huge improvement on the former liquidity program.
Hence, we advocate for the deployment of 400 osETH from the Treasury towards this strategy.
Our numbers and projections can be found here: Fluid calculus - Google Sheets
For more information on the Fluid protocol architecture we recommend these articles:
https://cyber.fund/content/fluid-2
Specification
We suggest that the StakeWise Liquidity Committee manage the Treasury balance when deploying liquidity to maximize efficiency and avoid waiting for DAO approval on every leg of ramping it up. Once deployed, the position can be transferred back to the DAO address since it’s represented by an NFT.
Considerations
One important thing to consider is the process of exiting from this strategy should the DAO decide to cease providing liquidity on its own.
Unwinding this strategy is roughly similar to unwinding Boost positions, with turning osETH into ETH being a key bottleneck. During times of market stress, StakeWise wouldn’t be able to unwind immediately due to the validator exit queue that throttles all exits - that is, unless we monitor it and preemptively free up some amount of ETH for osETH redemption via EL partial withdrawals in the Genesis Vault. Still, even with this approach we’d go through 24-hour cooldowns on every leg of the unwind, taking approximately 70 days, which is simply an impossible amount of time for the Treasury’s funds.
The solution we propose is creating an unwind contract that would leverage flashloans to unwind the position in much newer steps. Work on the contract will begin if this proposal is approved.
In any case, we suggest that the Liquidity Committee continuously monitor ETH and wstETH rates across major lending markets, as well as the validator withdrawal queue, for signs of stress (i.e. high utilization) of these assets, and pre-emptively exit some ETH from the protocol to protect the Treasury from significant drawdowns, at the expense of marginal performance drag in the Genesis Vault. The yearly drag would be ~4 / 165 = 2.5% per year; when looked at within the period of unwinding, e.g. 1 week, this works out to 0.05% drag on the annual performance, which we find negligible in the grand scheme of things.
Vote & discussion
This is a temp check vote that will last 1 week to gauge sentiment. Should this proposal receive broad-based community support, we will proceed to the Snapshot voting stage.