With the recent passing of QIP 166: Add MAI-DAI Arrakis vaults as collateral for MAI, Qi Dao approved the addition of a MAI-DAI UniSwap V3 LP managed by the team at Arrakis as collateral for our stablecoin, MAI.
The main incentive behind the addition of the LP is to provide increased MAI liquidity on Polygon, a chain which currently accounts for around 37% of total MAI backing across 23 different collateral types.
As Qi Dao provides zero interest loans, it can be difficult to incentivize MAI liquidity while also reducing protocol emissions using traditional methods. By allowing its users to mint MAI through a stablecoin LP such as MAI-DAI with a 102% collateral to debt ratio, users can earn trading fees at a projected 1% annual percentage rate with up to 50x leverage without much risk of impermanent loss or massive price fluctuations that can often arise from other LPs, even those composed of correlated assets.
The motivation behind the LP stems from a similar tactic deployed by MakerDAO with its own DAI-USDC LP also managed by Arrakis (formerly Gelator Network). The rationale behind creating such an LP is as follows:
Increase liquidity without increasing protocol emissions
Tighten the MAI peg with a highly liquid, low fee pool with one of the largest decentralized stablecoin protocols in DeFi.
Improve MAI liquidity in order to increase the frequency of vault debt ceilings without impact on MAI price.
Increase volume in order to get Chainlink oracles approved for MAI on chains outside of Polygon.
Generate sustainable yield for users while creating an interest bearing, but stable form of collateral to back MAI.
UniSwap provides extremely capitally efficient LPs, but due to the nature of its NFT positions, it does not reinvest trading fees back into the LP. Arrakis’ solution creates an ERC20 version of UniSwap positions which does, thereby allowing LP providers to auto-compound gains from LPs, while also increasing such fees by concentrating liquidity amongst a smaller segment of the trading range.
As can be seen in the historical DAI-USDC annual percentage yield chart above, courtesy of APY Vision, the net APY (yield — impermanent loss), has remained consistent over a 90 day period while generating an unprecedented amount of volume thanks to UniSwap’s high capital efficiency as seen below in UniSwap’s own analytics page for DAI.
Currently, the most liquid and highest volume pool on Polygon remains the QuickSwap MAI-USDC LP, accounting for $13.5m in liquidity with a seven day trading volume of $132k. The protocol incentivizes this LP on the Qi Dao farms page with a 5.5% return via Qi emissions. This means that we are currently emitting roughly $740k per year or nearly 6.75m Qi tokens to maintain this pool.
In comparing the capital efficiency of the DAI-USDC pool with a seven day volume of 33% total TVL, the main MAI stable pool only currently offers less than 1% volume vs TVL. As a result of QIP 170: Migrate Polygon Curve Incentives To UniSwap, the DAO has voted to move the incentives that were assigned to the Curve MAI-3Pool LP, towards incentivizing a MAI-USDC LP on UniSwap, also managed by Arrakis. By comparison, this LP currently has $1.94m TVL, with a seven day trading volume of $1.18m, or 60% volume to TVL.
By comparison, let’s take a look at the trading volume on the DAI-USDC LP courtesy of Dune Analytics.
Arrakis LPs are minted and redeemed from the pool contract with the underlying assets, meaning that pool liquidity should be viewed as a function of the liquidity of each of the constituent assets and not necessarily of the pool as a whole. This means that the ERC20 generated from depositing into the Arrakis LP is not tradable on an exchange or anywhere on open market, unless of course, you consider trading your MAI-DAI LP vault NFT. This might be an important factor to consider for liquidations as only the price dynamics of the individual assets (in our case MAI and DAI) can be leveraged, and not the pool itself.
For the initial pool, the Qi Dao team has chosen an initial trading range of 0.975–1.001. As the pool gains TVL and begins generating trading volume, the Arrakis team will manage the ratios to concentrate liquidity amongst the most common trading range.
DAI Price Volatility And The USDC Factor
Historically, DAI has experienced a relatively stable peg since the introduction of its price stability module (PSM) which has helped tighten the dollar peg. From MakerDAO’s documentation:
A PSM creates the same danger as low liquidation ratio stablecoin vault types: the Maker Protocol will take on a large amount of stablecoin collateral at times where DAI demand outstrips DAI supply. This is usually cited to be a risk due to the centralization of other stablecoins and the possibility of regulatory action targeting Maker specifically. The risk of regulatory action may be slightly greater with a PSM than the alternative because the stablecoin collateral created through a PSM is effectively owned by the Maker Protocol, rather than being owned by a user using it as collateral for a loan of DAI.
While MAI does not have such a mechanism in place, the large volumes present in the MAI-USDC LPs primarily, have helped tighten the peg much closer to the 0.99–1.01 range expected. Such pools have been much more successful in maintaining the peg than previous arbitrage mechanisms, such as the now deprecated Anchor, ever could.
There has been discussion on DAI’s reliance on USDC as collateral as it can be seen as a centralized risk factor. Currently, DAI’s is 53.1% collateralized by USDC so the fear is that Circle could blacklist the pool address causing DAI to depeg, though with DAI currently accounting for nearly 8% of the total USDC supply. it is highly unlikely that Circle would risk taking such a hit on its market cap and the fallout that would result from the negative press given by the DeFi community. In the case of an asset freeze by Circle, this may cause a cascading depeg on DAI which could directly impact the MAI peg as well.
Qi Dao has been historically apprehensive in using a centralized asset like USDC as collateral. As part of our risk assessment when onboarding new collaterals, we take into consideration whether an asset is centralized both in permissions and in the number of wallets holding that token, as well as how concentrated ownership is within those wallets (not including certain contracts, such as would be the case for governance tokens like CRV or BAL).
We do currently have some level of exposure to DAI as collateral as the DAO has voted several different DAI derivatives to be used in vaults, including Compounding Aave Market DAI, Tetu xxDai, and StakeDAO’s am3CRV on Polygon, Beefy Aave Dai on Optimism, StakeDAO am3CRV on Avalanche, and Yearn DAI on Ethereum. While these vaults do open MAI to some of DAI’s price action, note that they are all yield bearing versions of DAI whose value grows in size through various strategies thereby offering additional protection to possible DAI price fluctuations to the downside.
Cascading Liquidation Risks
As mentioned in the section “DAI Price Volatility And The USDC Factor” above, the first risk of cascading liquidations comes from DAI’s reliance on USDC as collateral in the event that Circle decides to freeze the USDC holdings of MakerDAO. The above reference PSM does help stabilize the price if such an event were to happen, but adding additional reliance on DAI to maintain MAI’s peg stability would mean that MAI could also suffer from similar downward price action as liquidity providers withdraw the underlying tokens and swap back to DAI to increase market demand. Such price action could be exacerbated in the event that the pool grows to account for a majority of MAI backing. At the time of this writing, the current market cap of MAI is roughly $50m.
Conversely, if MAI were to depeg, this would give arbitragers a prime swapping opportunity, as well as create additional market demand as people swap other stables, mainly USDC and DAI, to repay any outstanding loans.
There is an additional risk to consider. If MAI-DAI were to become the primary source of MAI backing and stable swaps across multiple chains, we would have to consider what percentage of the underlying MAI is actually backed by trusted collaterals (see the Qi Dao Risk Matrix for a full list). This could easily be mitigated by the Qi Dao team by placing a cap on the percentage of MAI able to be minted from the LP, and with a max borrow of $50m per vault, it may be a worthwhile discussion for the community to have in further assessing whether this would be a likely risk.
Oracle Price Manipulation
The most important risk factors to consider with the creation of MAI-DAI LPs are therefore front running risks created by both the LP management contract, and oracle manipulation.
Since Arrakis is actively managing the LP, it is possible that a bad actor could front run any fee generation and therefore any price manipulation. ChainSecurity has an excellent article on Curve LP oracle manipulation, much of which will be relevant here. It is unlikely that the Arrakis team would allow for this to happen as the management fee is 2.5% of any fees generated, and with the pool trading with 0.5% swap fees, it would cause more damage to their business model than would it would be worth.
Due to MAI’s relatively low market cap compared to DAI (with DAI having 111 times the market cap of MAI), the Qi Dao team has been working extensively over the past several weeks to add additional layers of security to the LP oracle price feed to ensure that price manipulation of the LP is mitigated. We expect a technical update from the team with regards to its additional safeguards and implementation to be presented to the community at the public launch of the MAI-DAI LP vaults.
While reliance on DAI as MAI collateral does not come without risk, particularly decentralization risks from DAI’s reliance on USDC backing, it is the opinion of this committee that the benefits of using a MAI-DAI LP as MAI backing far outweigh the risks, particular as it would increase the capital efficiency of MAI, improve its peg further, and help increase MAI’s market penetration through additional volume and liquidity.