Gauntlet would like to kickstart the conversation around an important strategic consideration for the wETH market: how would wETH supply be encouraged?
On Compound V2 markets, there is natural demand to supply wETH, given that the wETH can be used as collateral. However, this natural demand as collateral does not exist on the Comet market, as the wETH supplied can not be used as collateral. Of course, wETH supply is a limiting factor to the growth of this market.
In the case of the wETH Comet market, the yield of supplying wETH will be significantly more important to attract wETH supply than it is on the V2 market.
Although Gauntlet does not make strategic recommendations to the DAO, we wanted to kickstart the conversation, as the decisions the DAO makes here may impact our recommendations on the interest rate configuration for wETH.
Several options for the community are:
Option A: Configuring the wETH interest rate curve differently on Comet than on V2.
Benefits: the interest rate curve can be configured to provide a higher yield to wETH suppliers on V3 to encourage supply while aiming to prevent excessive utilization.
Tradeoffs: there is currently a scarcity of data on the elasticity of users with regard to interest rate curves.
Option B: Pause supplying of ETH on V2 to encourage supply on V3.
Benefits: encourages migration to both the Comet USDC and wETH markets.
Tradeoffs: there are nuanced risk tradeoffs here. For example, pausing supply of ETH on V2 prohibits borrowers from adding more ETH collateral to their positions in times of market stress.
Option C: Distribute COMP incentives on Comet (for the supply side of wETH)
Benefits: encourages supply to wETH market.
Tradeoffs: cost to the protocol. In addition, there is currently a scarcity of data on the elasticity of users with regard to interest rate curves. Any future COMP incentive reduction should be made gradually to not shock the market.
We look forward to hearing the community’s input. Of course, the community may choose one or a combination of the above options.
All three approaches are suitable (and possibly necessary), but as a subjective preference, I would order them as A (at launch), C (at launch, or immediately thereaftrer), B (after V3 wETH is stable). My reasoning is as follows:
There is no downside, cost, or risk for cWETH to use a custom interest rate model that is different than the legacy interest rate model. In fact, given the starting assets and parameters, the use-case for borrowing Ether from this market (likely to efficiently apply leverage to staked Ether) will create very different rate expectations than the existing Compound protocol. There is likely significant borrowing demand for Ether (say, 2-4% APR) below the staked Ether yields (currently about 6-8% APR), and these users will likely be borrowing for extended periods of time. On the supply side, users (individuals, institutions, or other applications) may tolerate a lower yield given the liquidity of Compound markets (say, 1-3%). This market should likely have an interest rate model that targets high levels of utilization, with a ceiling on borrowing costs in line with staked Ether yields.
In regards to including WETH in the distribution of COMP, you’re right that it makes more sense to include the suppliers of Ether (and not borrowers) in this market, as borrowing will be the more “obvious” use-case, whereas USDC is the opposite (massive supply of USDC seeking yield).
Only after this market is running, at scale, would it make sense for the community to debate whether to disable functionality in the V2 protocol (though my view is, it might make sense to keep it as-is in perpetuity, just incentivize migration in economic ways).
So, I agree that if we add liquid staking ETH as collateral in comet, it is likely to get increased demand of borrowing ETH(mainly from dapps using liquid staking ETH strategy such as instadapp lite, rocketpool, and coinbase)
For user groups(wstETH, cbETH, and ETH), there is a strong incentive(for borrowing ETH) to deposit their wstETH and cbETH on comet but it has not much merit for native ETH holders. So, I think that it is good idea to assign some COMP distribution for ETH deposit side.
Improved IRM
With high utilization, it is important to keep low ETH borrowing rate level for liquid staking eth strategy players.(making maximum profit)
Currently, I’m not sure it should be using the same IRM for both wstETH and cbETH. Because cbETH can be considered as centralized asset in the community.
Lido is really glad to see the discussion of Compound III moving forward. I wanted to hop in to mention that we have been in touch with Gauntlet to make any data needed available. We would be glad to offer our analytic team’s time and effort to the Compound community as well if there is a need/opportunity in support of this discussion.
We also have a great deal of experience in helping protocols that integrate stAssets manage supply and borrow demand. We will be happy to consult!
Gauntlet has conducted an initial market risk analysis on the parameterization of the wETH Comet market.
stETH
Our systems observe ample liquidity in the stETH/wETH market. The slippage/liquidation curve suggests that a $130mm market sell of stETH through 1inch only incurs 1.5% slippage, implying a liquid market. We also ran simulations hypothesizing the amount of insolvencies and liquidations given various CFs and stETH/ETH deviations for accounts that have historically supplied stETH and borrowed wETH - and found that only at high deviations is there a likelihood for insolvencies.
Because of this, it can prudent to initialize the market at $100mm supply cap, 90% CF (collateral factor), 93% LCF (liquidation collateral factor), and 5% LF (liquidation factor). Combining capital-efficient borrowing parameters with a stringent supply cap will be valuable to understanding and studying the pool behavior.
We recommend a lower supply cap for cbETH but with the consistent CF/LCF/LF parameters as wstETH.
cbETH is less liquid on DEXes compared to wstETH (1.5% slippage for $1.3mm sell of cbETH, versus $130mm sell of wstETH). cbETH is also not very liquid on CEX - where a $1mm sell incurs roughly 2% slippage. In addition, cbETH is only redeemable to locked ETH on Coinbase, whereas wstETH is redeemable to stETH, which is more easily transferable and more liquid. There exists a 2-3% discount on cbETH, perhaps due to this intransferability.
However unlikely, we may imagine a situation in which insolvency news on the CEX leads to the cbETH market rate plummeting, creating risk for Compound. Although this existential risk is neither a market risk nor a quantifiable risk, a conservative supply cap can mitigate the potential loss posed to Compound.
By approving this proposal, you agree that any services provided by Gauntlet shall be governed by the terms of service available at gauntlet.network/tos.
Thanks, good question (and sorry, I’ve been away). Personally I think a USDC and ETH market both make sense on Ethereum mainnet. Beyond that, I would think the focus should be on USDC and/or native token base markets on other chains, starting with popular L2s/natively bridged chains.
Thanks Paul and Gauntlet team for these analyses. We’ve applied the parameters suggested to the deployment and initial proposal script. There was a last-minute change to the cbETH price oracle, which is going to be posted against ETH instead of USD from here on out, so we have some price feed wrapper contracts to handle that, which ought still be audited by OpenZeppelin before deploying and launching this market.
Hey Frank, we are still awaiting audit of the price feed changes, and overall feedback on the deployment from OpenZeppelin. At the same time, we discovered an issue with the deployment tooling that will end up deploying a different Configurator/Rewards contract than cUSDCv3, which we want to reuse, so we are making a change to support that. I’m not sure which of these things will finish first, and we are also in prime holiday season, but we should be very close! I would guess a few more weeks.
Note that these numbers will differ slightly from the APR numbers because of compounding interest in blocks.
Methodology and Rationale
Some core assumptions to this recommendation are that 1.) stETH has an annualized yield of approximately 5%, 2.) the current utilization statistics of the Compound II ETH market are based on the fact that there is no strong use case for borrowing ETH and not that the interest rates are too aggressive, 3.) the Compound III USDC Comet provides reasonable data points, but the WETH Comet will have a significantly different use case, 4.) users will be elastic with regards to interest rate changes around the boundary of stETH annualized yield.
Since the WETH borrowed cannot be used as collateral, there is less market risk concerning WETH liquidations, and thus there is less need to make the max interest rate especially high.
By approving this proposal, you agree that any services provided by Gauntlet shall be governed by the terms of service available at gauntlet.network/tos.
This might be a silly question, but per the USDC IR Curves, how is the USDC supply interest rate higher than the USDC borrow interest rate at 100% utilization?
If my calculations are correct, with the curves you are proposing the implicit reserve factor at 90% utilization would be 70% (and 53% at half utilization).
This instance of Compound III is very promising and could have a real organic traction without any rewards, with small reserve factors (see the metrics of stEth/Eth on Aave, Morpho-Aave, Euler…). It don’t understand why it should depend so much on COMP rewards.
Is the goal of this to profit from COMP distribution to increase the reserve of the Compound DAO? In this case, it should at least be explicit, in order for the community to discuss about it, especially if the reserve factor is so high.
The logic surrounding borrowing costs makes sense, @pauljlei.
I agree with @MathisGD that the implicit reserve factor could be significantly lower (e.g. interest rates across utilization rates could be higher); one of the original concerns with the market would be that native ETH suppliers would have below-average incentives to participate (vs, staking their Ether natively via proof of stake).
Are there other models that maintain positive implicit reserve factors, simply by adjusting supplyPerSecondInterestRateSlopeLow?
We agree that a highly important consideration is properly incentivizing supply. Our original supply interest rate (even without COMP incentives) is in line with historic data for ETH supply rates. We acknowledge that since ETH cannot be used as a collateral in this Comet, the protocol requires a higher supply interest rate to motivate users to supply. However, given the behavior of these linear curves, having interest rate values that are closer together (both at the base value and at the kink) leads to additional utilization points where there will be a negative reserve rate.
Our recommendation above (which we shall refer to as Option 1) constructs the interest rate curves such that there are fewer points in which there are negative reserve rates. There are some tradeoffs here:
With fewer areas in which there are negative reserve rates, there is less risk of “running out” of reserves.
However, the tradeoff is that if you want to minimize areas where there is a negative reserve rate, this means that the supply rate has to be lower. When the supply rate is lower, this means that there is less incentive for users to supply wETH. As such, growth in the market may be bootstrapped slower.
To @rleshner’s point about maintaining a positive reserve rate, one of our most important considerations was minimizing the behavior of negative reserve rate. However, if we assume that the utilization of the pool will mainly be around the CF and LCF (90%, 93%), we can build the curves below where suppliers have more incentive, but there will be more regimes with negative rates (thus increasing the chances that reserves “run out” in extended regimes of negative rates).
Open questions: what is the total amount of ETH reserves that the Comet plans to launch with? That will help give us bounds for how negative the reserve rate can be. If we seed the Comet with more ETH reserves, then we can be more aggressive with allowing negative reserve rates. With this additional data point, we can further fine-tune these recommendations.
Largest negative reserve rate for Option 2: -0.69423076% at utilization: 36.54%. This means that a reserve of $100k ETH (in USD) will run out in approximately 140 days.
Following the conversations on this thread, and discussions on recent community calls and in Discord, the parameters to deploy cWETHv3 have been finalized on the pull request:
The initial risk parameters follow Gauntlet’s recommendations:
wstETH: $100mm supply cap (64,500 tokens), 90% CF, 93% LCF, and 5% liquidation fee
cbETH: $10mm supply cap (7,100 tokens), 90% CF, 93% LCF, and 5% liquidation fee
The interest rate models follow Gauntlet’s first recommendation, with an adjustment to increase the supply rate slope. After the launch of the market, Gauntlet will monitor and consider recommendations to the interest rate model based on preliminary utilization & growth.
The price feeds of the deployment use the following inputs:
The base asset, WETH, uses a constant price feed of unity in terms of ETH with 8 decimals.
For wstETH, we deployed a custom wrapper around the ChainLink stETH/ETH feed to convert using the wstETH/stETH exchange rate.
For cbETH, Coinbase has launched a new cbETH/ETH price feed together with ChainLink, which we wrap with a scaling price feed to convert the decimals to the constant 8 decimals that Comet expects.
OpenZeppelin has completed an additional audit of modifications to the price input contracts, since the original post in this thread.
To initialize the market, the deployment process deviates from the script used to initialize the USDC market, since the WETH market shares contracts in common with the USDC market, including the Configurator and Rewards contract.
The initialization proposal will take the following actions:
Migrate the COMP distribution, 38.70/day, from Compound v2 ETH to the new WETH market
Set the CometFactory for the new Comet instance in the existing Configurator
Configure the Comet instance in the Configurator
Deploy an instance of the newly configured factory and upgrades the Comet instance to use that implementation
Configure the existing Rewards contract for the newly deployed Comet instance
Seed the market with 500 WETH of reserves, by wrapping Ether held by the Timelock, then
Transfer the Timelock’s WETH to the deployment
Transfer 25,000 COMP from the Comptroller to the Rewards contract to refresh it’s supply
One thing I’m curious on is Gauntlet’s denomination in $USD with their initial recommendation. On November 22nd of last year, ETH was $1100 USD. Today, ETH is closer to $1500 USD. With liquidity for wstETH and cbETH both being mainly in ETH, this means there’s deeper $USD liquidity for these assets as well.
If the whole market is denominated in ETH, could Gaunlet denominate their recommended supply caps in ETH as well? cc @pauljlei
From November 22nd to now, liquidity for newer assets like cbETH has improved meaningfully as well (greater than ETH’s increase in price). At the time of their initial recommendation, there was 1.5% slippage for $1.3mm sell of cbETH. Today, there is roughly 1.5% of slippage on a $7.3mm swap.
Your intuition is right, the governance cycle is too long, the market changes cause the data to change too much, and the previous data is not enough to support the conclusion of the parameters.
The result is that the coefficients introduced cannot match the market, and then they will modify the parameters many times to do some repetitive work.
