Gauntlet is proposing the following framework and requirements for Apollo and Artemis to utilize when listing new Asset Markets:
To guide best practices to the Community, Gauntlet aims to provide a standard framework for assessing market risk when listing assets and enabling assets as collateral. Managing collateral listings is essential to the growth of the protocol. As new assets in DeFi proliferate and older assets fall out of favor, Moonwell must list and delist assets to maintain its usefulness as a protocol. Given 2 weeks of notice and strong community buy-in, Gauntlet will conduct risk assessments prior to new assets being listed.
This interplays with Open Zeppelin’s Asset listing guide and is focused specifically on market risk and how Gauntlet will support the asset listing process in Moonwell. Throughout the asset listing and collateral enablement processes, Gauntlet’s goal is to ensure that insolvency and liquidity risks are minimized and that when liquidations occur, they can be done so healthily with incentivized liquidators. In order to be
unbiased, Gauntlet will not explicitly support any asset listing but instead provides the below framework as guidance for the community.
Gauntlet will assess a given asset’s liquidity and other market characteristics to be added to the protocol. We ask that the party putting forward the proposal for asset listing include the following data:
- Market cap of the token
- Total supply
- Largest liquidity sources (exchanges including CEX and DEX)
- Volatility per our definition Gauntlet - MakerDAO Auction Report (30 day, 90 day, 1 year)
- Average daily trading volume on CEX and DEX
- Gini coefficient and Herfindahl index of token balances
Gauntlet will relay our findings to the community and make parameter recommendations for Reserve Factor and Borrow Cap. We always recommend that Collateral Factor be set to 0 on the initial listing. Our goal with this initial analysis is to ensure that liquidations will be feasible with the amount of supply and borrow expected to be added initially.
The Reserve Factor is more straightforward on the initial listing. We generally recommend 15% for stablecoins and 25% for non-stablecoin assets. These guidelines will evolve over time—namely following initial recommendations early in Q3.
In general, it is prudent to err on the conservative side for Borrow Cap, as this is a direct lever for us to ensure that Moonwell can minimize losses in an infinite mint attack (and other attacks). The community needs to decide whether a given asset should be turned on for borrowing (some assets may not make sense, i.e., rebasing assets like stETH). If the community decides to turn an asset on for borrowing, Gauntlet will conduct analysis to set the asset’s Borrow Cap.
Gauntlet’s goal with setting the Borrow Cap on initial listing is to predict the estimated supply in Moonwell in a mature state and then being conservative to ensure that there are no unforeseen technical risks that cause protocol failure or outsized insolvencies.
In particular, Gauntlet will set Borrow Cap in the following way:
For a non-stablecoin asset, the asset will be compared with similar assets already on Moonwell. Its Borrow Cap should be set as the minimum of 50% of the supply of similar assets in Moonwell or 1% of the circulating market cap of the token. By similarity, we are looking at both:
- Comparable utility of the token (i.e., pure governance, staked derivatives, etc.)
- Comparable market structure of the token (i.e., similar market caps, ADV, vol, GINI)
The idea behind using the minimum here is to establish an upper limit relative to the token’s circulating supply and ensure that the asset is compared to other similar assets on Moonwell.
For stablecoins, we recommend no Borrow Cap because stablecoins usually have high organic demand for borrow and thus a high utilization rate, leading to more protocol revenue.
Over time we will raise and potentially remove the Borrow Cap from all assets (as long
as the community is comfortable with it. Maker was a good example of keeping and
lowering the Borrow Cap to prevent governance attacks.)
Enabling Collateral can be one of the riskiest phases for a new asset coming onto Moonwell. Gauntlet’s general recommendation is a conservative, phased approach. New market risks are introduced when an asset is enabled as collateral. The community should assess the technical risks and follow OpenZeppelin’s framework. Below, we outline the relevant market risks and recommendations.
The community should take the first step in proposing that an asset is enabled as collateral by publishing a post on the Moonwell Community Forums. This forum post should contain the metrics outlined above (market cap of the token, total supply, etc.). Should there be enough community support behind enabling this asset as collateral (i.e. via a poll), Gauntlet will then provide a market risk analysis as described below.
Sufficient liquidity is required for an asset prior to enabling it as collateral. A cautious test of this is the combined slippage across all liquidity sources to measure how well a given asset can be absorbed into the market (a signal that may change upon asset listing).
We recommend that the combined (CEX + DEX) slippage be less than 5% for a sell order of either $300k or 10% of total asset supplied in Moonwell (whichever is greater) before enabling an asset as collateral. Gauntlet formulated this guideline by analyzing existing assets in Moonwell and their supplies. Then, Gauntlet modeled what the market could healthily absorb using our combined slippage model.
Here is a simplified example using FRAX:
For new asset listings, Gauntlet would use its slippage models, but to make the example simpler we can look at 1inch liquidity for FRAX. As of 2022-10-25, there is $7.6M of FRAX supplied in Moonwell Artemis and
$13.2M in Moonwell Apollo, so max($300k, 10% of supply) equates to a $2.1M order. Looking at 1inch, we see that a $2.1M order would hit around 0.18% slippage (2022-10-25; 1inch is only DEX liquidity, so slippage numbers are better when including CEX data).
- Note that this is an initial guideline to broadly understand the liquidity of a given
asset. As Gauntlet conducts analysis on new assets coming to the protocol, we may
provide more granular analysis (i.e., around the distribution of users and around
asset classes) that deviates from this guideline when we determine there is a better
understanding of the market risks.
Our approach is purposely conservative and leans on consistent parameter tuning to increase capital efficiency over a longer time frame in a safer manner. It must be the case that the community supports a given asset to be turned on as collateral. This is not an assessment that can be made purely from a market risk lens, as the non-market risk side can pose a more existential threat to Moonwell (i.e., smart contract bugs).
After the community decides they want an asset turned on as collateral, Gauntlet’s general recommendation will start conservatively at a 20-30% CF. This will be assessed on a per asset basis. The goal is to give enough time to the asset at a low collateralization ratio to make sure that mechanisms are working as intended.
After 2 weeks of collateral enablement, Gauntlet’s ongoing parameter tuning work through simulation optimization will consider increasing Collateral Factors based on the actual usage data and market conditions as long as it is safe to do so from an insolvency risk perspective.
Thus, this phased approach to collateral enablement starts with conservative Collateral Factor parameterization. After the asset is enabled as collateral, Gauntlet’s financial modeling platform will continue to ingest data on how usage and market conditions evolve. At this stage, through a more thorough understanding of risk via our simulation models. Gauntlet can optimize for capital efficiency to the protocol.
Gauntlet will not conduct simulations using fake data to assess the risk.
- Simulations do not lend well to this type of listing when there is no prior data. Gauntlet has in the past used borrower distributions from similar assets to assess risk. However, this has been a weak signal given how different each asset’s usage behavior is when actually incorporated into the lending platform. As such, Gauntlet will not conduct simulation analysis to predict user behavior ahead of an asset’s listing.
Gauntlet will not assess any of the following areas of non-market risk and instead defer to OpenZeppelin and the community on the below:
- Oracle risk
- Infinite mint attacks
- Governance attacks
- Smart contract risk
- Centralization risk
- Other technical risks
- Gauntlet will treat the asset as if it serves its underlying purpose correctly and will not assess ancillary aspects of the token design. It is up to the community to decide whether a given asset belongs in the protocol
- After the community has had time to review, a snapshot vote will be put up