[ARC] Aave V3 Ethereum Deployment: Assets and Configurations

As both #option 1 and #option 2 presented here include stETH I’d like to draw attention to the original blog post. BGD. Aave v3 Ethereum. New deployment vs Aave v2 Upgrade

  • It gives the possibility for the community to reassess the risk configuration of the assets currently listed on v2. With an upgrade v2→v3, everything will remain the same as currently, that is not possible. It is important to highlight that this also allows for the listing of for example fewer Aave-custom implementations for some assets, like wstETH, instead of the current stETH version listed on v2.

stETH has some issues Account's `stETH` balance getting lower on 1 or 2 wei due to rounding down integer math · Issue #442 · lidofinance/lido-dao · GitHub which cause contracts that rely on exact precision (e.g. liquiditySwap / repayWithCollateral features) to fail. From a technical perspective, I would highly recommend to consider listing wstETH instead as it’s not affected by this balance inaccuracy. Also it would allow listing wstETH with the default a/s/v implementations.

7 Likes

I would push for a more inclusive initial listing, especially as a method of convincing as many folks as possible to migrate from V2 → V3.

I would support option 2, however include more assets, specifically:

Stables: FRAX, GUSD, LUSD
Non-Stables: BAL, CVX, YFI

Rational:

  • More Stables: One of the primary uses of Aave is receiving liquidity against volatile collateral in the form of stables. Therefore, we should lean towards the inclusion of more stables.
    • FRAX is one of the largest stablecoins, with over 1.3B in circulation. Additionally, FRAX is actively incentivizing aFRAX with an FXS gauge.
    • GUSD is a stablecoin directly backed and redeemable by Gemini, an off-ramp for many crypto users. With Gemini’s new posture towards promoting its use and utility in DeFi (see their discussions with Maker), I see no reason not to include it in the initial V3 deployment. At over 300MM in circulation, it is more widely issued than the included SUSD.
    • LUSD is an immutable, censorship-resistant decentralized stablecoin. Given recent regulatory action and the subsequent reaction by the crypto community, there will be a clear demand for the borrowing of such products. Given the recency of its listing and the high amount in circulation (over 150MM, larger than the included SUSD), it is a strong candidate for initial V3 inclusion.
  • Non-Stables: There are several tokens not included in option 2 that represent strategic importance to the DAO, and should be included in an initial V3 deployment.
    • BAL: Balancer is one of Aave DAO’s closest partners, between the use of aBPT for the safety module, the use of bb-a-usd as the revenue token of veBAL, and multiple token swaps and acquisitions. Given the depth of the relationship between the 2 entities and the DAO’s desire to continue acquiring BAL, not including BAL in the initial deployment would be a misstep.
    • CVX: With GHO on the horizon, and CRV’s current status as the dominant stablecoin DEX, acquisition of CRV and CVX should be a priority for the DAO. Hence including it in a V3 should be a priority.
    • YFI: Yearn is one of the largest users of Aave, across multiple chains. Beyond their AAVE and sAAVE vaults, their various vaults across Ethereum, Arbitrum, and Optimism leverage Aave as a key DeFi building block. As a key partner, YFI should be included in the initial deployment list.

For ease of conversation, a chart of existing V2 assets + Gauntlet’s 2 options + my proposal above is included.

Asset Gauntlet 1 Gauntlet 2 oneski22 Proposal
USDC YES YES YES
ETH YES YES YES
stETH YES YES YES
WBTC YES YES YES
USDT YES YES YES
DAI YES YES YES
LINK YES YES YES
CRV - YES YES
AAVE YES YES YES
MKR - YES YES
TUSD - YES YES
FRAX - - YES
BUSD - YES YES
YFI - - YES
UNI - YES YES
ZRX - - -
MANA YES YES YES
xSUSHI - - -
1INCH - YES YES
DPI - - -
GUSD - - YES
SNX - YES YES
CVX - - YES
BAT - - -
REN - - -
sUSD - YES YES
ENJ - - -
BAL - - YES
AMPL - - -
RAI - - -
USDP - - -
UST - - -
LUSD - - YES
FEI - - -
renFIL - - -
KNC - - -

Note: Assets sorted by Aave V2 Total Supply as of Block 15726280. All numerical stats sourced from DeFi Llama as of Block 15726280.

4 Likes

In general agree with the need to reassess the collateralization factor of many assets and get rid of the less liquid/more problematic/deprecated. My opinion is that we should leverage more isolation mode and let everyone that was able to use V2 to also use V3, with some limitations introduced by isolated assets. One important point i want to highlight is that in my opinion all the less liquid assets currently used on V2 should be demoted as non collateral to avoid risk. If we give them the possibility to still use these asset in V3 (albeit as isolated), we don’t introduce any discomfort for users, and we reduce risk on V2 as well. Here is my proposal, following @oneski22 approach

Asset Gauntlet 1 Gauntlet 2 oneski22 Proposal Emilio Proposal
USDC YES YES YES YES
ETH YES YES YES YES
stETH YES YES YES YES (wstETH)
WBTC YES YES YES YES
USDT YES YES YES YES
DAI YES YES YES YES
LINK YES YES YES YES
CRV - YES YES YES
AAVE YES YES YES YES
MKR - YES YES YES
TUSD - YES YES YES (Isolated)
FRAX - - YES YES (isolated or no collateral)
BUSD - YES YES YES (no collateral)
YFI - - YES YES (isolated?)
UNI - YES YES YES (isolated?)
ZRX - - - YES (isolated)
MANA YES YES YES YES (isolated)
xSUSHI - - - -
1INCH - YES YES YES (isolated)
DPI - - -
GUSD - - YES -
SNX - YES YES YES(isolated)
CVX - - YES YES(isolated)
BAT - - - YES(isolated)
REN - - - YES(isolated)
sUSD - YES YES YES(isolated or no collateral)
ENJ - - - -
BAL - - YES YES(isolated)
AMPL - - - -
RAI - - - -
USDP - - - -
UST - - - -
LUSD - - YES YES (isolated or no collateral)
FEI - - - -
renFIL - - - -
KNC - - - (there is a proposal to list the new KNC token, so we should evaluate how that goes)
4 Likes

Thank you @sakulstra @oneski22 @Emilio and all for your feedback. We present Gauntlet 3 below in an attempt to incorporate the community feedback.

We would like to call out the price manipulation analysis that Gauntlet has provided here. 6 assets are being discussed: BAL BAT DPI MANA REN ZRX.

Should the community decide to disable those assets as collateral, then we recommend that those assets should also not be enabled as collateral on V3 ETH.

In addition, we hear the community feedback around listing wstETH instead of stETH. There may be price manipulation potential with the wstETH mechanism, so Gauntlet is currently analyzing the risk.

Asset Gauntlet 1 Gauntlet 2 oneski22 Proposal Emilio Proposal Gauntlet 3
USDC YES YES YES YES YES
ETH YES YES YES YES YES
stETH YES YES YES YES (wstETH) YES (wstETH, pending analysis)
WBTC YES YES YES YES YES
USDT YES YES YES YES YES
DAI YES YES YES YES YES
LINK YES YES YES YES YES
CRV - YES YES YES YES
AAVE YES YES YES YES YES
MKR - YES YES YES YES
TUSD - YES YES YES (Isolated) YES (Isolated)
FRAX - - YES YES (isolated or no collateral) YES (no collateral)
BUSD - YES YES YES (no collateral) YES (no collateral)
YFI - - YES YES (isolated?) YES (isolated)
UNI - YES YES YES (isolated?) YES (isolated)
ZRX - - - YES (isolated) YES (isolated), but TBD
MANA YES YES YES YES (isolated) YES (isolated), but TBD
xSUSHI - - - - -
1INCH - YES YES YES (isolated) YES (isolated)
DPI - - - - -
GUSD - - YES - -
SNX - YES YES YES(isolated) YES(isolated)
CVX - - YES YES(isolated) YES(isolated)
BAT - - - YES(isolated) YES(isolated), but TBD
REN - - - YES(isolated) YES(isolated), but TBD
sUSD - YES YES YES(isolated or no collateral) YES(no collateral)
ENJ - - - - -
BAL - - YES YES(isolated) YES(isolated), but TBD
AMPL - - - - -
RAI - - - - -
USDP - - - - -
UST - - - - -
LUSD - - YES YES (no collateral) YES (no collateral)
FEI - - - - -
renFIL - - - - -
KNC - - - (there is a proposal to list the new KNC token, so we should evaluate how that goes) -
4 Likes

Great Breakdown!

My general opinion is also that we should include as many options as possible for users and use the e-mode as much as possible for the riskier assets! So I would be in favour of Gauntlet 3!

Thanks guys!

1 Like

Thanks for the updated suggestions and analysis performed.

In general, I am in support of Gauntlet 3.

However, where an asset is listed in 3 but not in 1 (and can be used as collateral), I believe that the supply caps of these assets should be assessed and suggested as part of this proposal. (I would expect there to be supply caps on these assets).

Regarding wstETH vs stETH:

Other lending protocol (won’t name) have both stETH and wstETH. Can this not be an option as well?
It would be interesting to get @Emilio’s thoughts here.

Hi @Pauljlei,

Please see below from Llama:

Asset Gauntlet 1 Gauntlet 2 oneski22 Proposal Emilio Proposal Gauntlet 3 Llama Proposal
USDC YES YES YES YES YES YES
ETH YES YES YES YES YES YES
stETH YES YES YES YES (wstETH) YES (wstETH, pending analysis) YES (wstETH)
WBTC YES YES YES YES YES YES
USDT YES YES YES YES YES YES
DAI YES YES YES YES YES YES
LINK YES YES YES YES YES YES
CRV - YES YES YES YES YES
AAVE YES YES YES YES YES YES
MKR - YES YES YES YES YES
TUSD - YES YES YES (Isolated) YES (Isolated) YES (Isolated)
FRAX - - YES YES (isolated or no collateral) YES (no collateral) YES (no collateral)
BUSD - YES YES YES (no collateral) YES (no collateral) YES (no collateral)
YFI - - YES YES (isolated?) YES (isolated) YES
UNI - YES YES YES (isolated?) YES (isolated) YES
ZRX - - - YES (isolated) YES (isolated), but TBD YES (isolated)
MANA YES YES YES YES (isolated) YES (isolated), but TBD YES (isolated)
xSUSHI - - - - - -
1INCH - YES YES YES (isolated) YES (isolated) YES
DPI - - - - - YES
GUSD - - YES - - -
SNX - YES YES YES(isolated) YES(isolated) YES
CVX - - YES YES(isolated) YES(isolated) YES
BAT - - - YES(isolated) YES(isolated), but TBD YES(isolated)
REN - - - YES(isolated) YES(isolated), but TBD YES(isolated)
sUSD - YES YES YES(isolated or no collateral) YES(no collateral) YES
ENJ - - - - - -
BAL - - YES YES(isolated) YES(isolated), but TBD YES
AMPL - - - - - -
RAI - - - - - -
USDP - - - - - -
UST - - - - - -
LUSD - - YES YES (no collateral) YES (no collateral) YES
FEI - - - - - -
renFIL - - - - - -
KNC - - - (there is a proposal to list the new KNC token, so we should evaluate how that goes) - -
ENS - - - - - YES

Llama is supportive of retaining DPI as it is the market leading index product on the market. It has a reliable composite Chainlink oracle and liquidators are able to interact with Index Coop’s Exchange Issuance contract to mint DPI which eliminates the need to rely of DPI liquidity. Llama has worked with Fireblocks to attain approval for listing DPI on the Aave ARC Liquidity Pool and is currently working towards compiling a proposal for the community to vote on.

If a new proposal is needed to list wstETH, then please flag that here and Llama can work with Lido to create this proposal. This way stETH can remain on v2 and wstETH goes through the normal asset listing governance process.

BAL, CRV and CVX are all significant assets for Llama’s future plans and consequently, we seek to maximise the utility of these assets on the v3 deployment.

Is it Gauntlet’s plans to present SupplyCap and BorrowCap for various assets in a separate post ?

Llama would like to present the eMode categories for the community discussion as a separate post. The eModes for BAL, CRV, CVX and ETH are of particular interest. We also seek to incorporate feedback from various partners around utilising the wstETH/ETH recursive loop to generate yield on other assets, especially those more correlated with ETH price movements.

Separate to the assets listing aspects to the new v3 deployment, we should consider the Interest Rate curve and the possibility of re-baselining the revenue generation model. The ETH, BAL, CRV and CVX Reserve Interest Rate can be revised to improve profitability and user experience. The ETH Reserve Interest Rate should be revised to improve profitability as the v2 Interest Rate curve changes are overly optimised for users. In time, Llama will be presenting a proposal to tweak the v2 ETH Reserve Interest Rates.

3 Likes

hello boys and girls

just like to point out that the list you’re operating with is missing ENS token for some reason - even though less utilized tokens (cough, renFIL, cough) are on it

1 Like

weren’t those v2-specific features? are those present in v3 (or others that rely on exact precision)?

Not v2 specific, the same issue is present on v3 as well.

I would advocate for omitting DPI due to a low market cap (~$32mm.)

We are supportive of the third version of this proposed list, a combined iteration of Gauntlets’, Emilio, and Oneski22. @sakulstra thanks for the technical clarity here - wstETH seems best.

While DPI had demand and fit well into the meta-governance play, it feels best to not include it in V3 due to the recent mango hack and threats circulating social media.

If this perspective changes, we welcome a proposal to re-list and enable borrowing.

The list IMO which should be regarded this way is:

DeFiPulse Index ($32m)
Ren ($111m)
xSUSHI ($100m)

This is being discussed in parallel at: [ARC] Price Manipulation Implications on Aave: October 2022

Hey,
Gregory from Lido here :desert_island:
Thank you for taking a close look at wstETH, we are also thinking of price manipulation potential around its mechanism. The Lido team is in touch with Gauntlet, we will be back with more details if we find anything unexpected on our side.

1 Like

The DeFi Pulse Index, operated by Index Coop, has been available as collateral on Aave v2 markets for over a year. Snapshot vote for reference.

When referencing the recent Mango Markets incident, I assume this means oracle manipulation. TVL should not be the sole parameter for determining the risk of price manipulation. For DPI, in particular, the permissionless mint/redeem capabilities (as mentioned by Llama) allow for arbitrage opportunities when any of the underlying assets see significant price deviation.

Index Coop previously dealt with an attack similar to the recent Mango incident. At the start of the year, an attacker attempted to manipulate an oracle for the BED Index through Rari markets. This attack was unsuccessful because all Index Coop products can be freely minted making it easy for arbers to act quickly. This thread gives a detailed account of this failed attack. Moreover, our Flashmint functionality makes this arbitrage opportunity even easier, using Aave to instantly create a flashloan into the position. We’ve seen significant volume in our Flashminting contracts up until now and expect any significant deviation to see an immediate arb.

As Aave and the defi ecosystem matures, Index Coop believes structured products with inherent risk management are critical. We’ve seen larger DPI holders use DPI as collateral on Aave and expect this to continue as we strive to get DPI into the hands of more traditional investors. We hope to see DPI listed amongst this first batch of assets and bring increased buy pressure for the largest tokens in DeFi including AAVE. While DPI is one of the lower market cap assets, it is far less risky for Aave than most assets being proposed. As one of the largest voters in AAVE governance we looks forward to voting on these new assets in V3 and are happy to answer any questions.

3 Likes

Thank you, everyone, for your thoughts. We propose Gauntlet 4 below. To keep the discussion moving forward, we are targeting a Snapshot vote next week for Gauntlet 4 with the options YAE, NAY, and ABSTAIN.

Following this Snapshot vote, we look forward to providing parameter recommendations for the standard V3 market.

We’d note that we include USDT in isolation mode, similar to V3 Avalanche and V3 Polygon.

Asset Gauntlet 1 Gauntlet 2 oneski22 Proposal Emilio Proposal Gauntlet 3 Llama Proposal Gauntlet 4
USDC YES YES YES YES YES YES YES
ETH YES YES YES YES YES YES YES
stETH YES YES YES YES (wstETH) YES (wstETH, pending analysis) YES (wstETH) YES (wstETH)
WBTC YES YES YES YES YES YES YES
USDT YES YES YES YES YES YES YES (Isolated)
DAI YES YES YES YES YES YES YES
LINK YES YES YES YES YES YES YES
CRV - YES YES YES YES YES YES
AAVE YES YES YES YES YES YES YES
MKR - YES YES YES YES YES YES
TUSD - YES YES YES (Isolated) YES (Isolated) YES (Isolated) YES (Isolated)
FRAX - - YES YES (isolated or no collateral) YES (no collateral) YES (no collateral) YES (no collateral)
BUSD - YES YES YES (no collateral) YES (no collateral) YES (no collateral) YES (no collateral)
YFI - - YES YES (isolated?) YES (isolated) YES YES (isolated)
UNI - YES YES YES (isolated?) YES (isolated) YES YES (isolated)
ZRX - - - YES (isolated) YES (isolated), but TBD YES (isolated) YES (isolated)
MANA YES YES YES YES (isolated) YES (isolated), but TBD YES (isolated) YES (isolated)
xSUSHI - - - - - - -
1INCH - YES YES YES (isolated) YES (isolated) YES YES (isolated)
DPI - - - - - YES YES (isolated)
GUSD - - YES - - - -
SNX - YES YES YES(isolated) YES(isolated) YES YES(isolated)
CVX - - YES YES(isolated) YES(isolated) YES YES(isolated)
BAT - - - YES(isolated) YES(isolated), but TBD YES(isolated) YES(isolated)
REN - - - YES(isolated) YES(isolated), but TBD YES(isolated) -
sUSD - YES YES YES(isolated or no collateral) YES(no collateral) YES YES(no collateral)
ENJ - - - - - - -
BAL - - YES YES(isolated) YES(isolated), but TBD YES YES(isolated)
AMPL - - - - - - -
RAI - - - - - - -
USDP - - - - - - -
UST - - - - - - -
LUSD - - YES YES (no collateral) YES (no collateral) YES YES (no collateral)
FEI - - - - - - -
renFIL - - - - - - -
KNC - - - (there is a proposal to list the new KNC token, so we should evaluate how that goes) - - -
ENS - - - - - YES Yes (isolated)
2 Likes

Regarding some technical aspects affecting the listing of assets in the future Aave v3 Ethereum:

  • In order to have unified behavior across all Aave v3 instances, we recommend using USD as the reference currency of the Aave oracle, changing this way from the ETH reference on Aave v2 Ethereum.
    Given the presence on the Ethereum v2 pool (and plans for the upcoming v3) of stETH, we will be introducing an oracle adapter to be used by wstETH as a price feed, to avoid any problem with async updates from Chainlink price feeds.
  • We recommend using the BTC/USD feed for WBTC, as we consider it more representative than an alternative WBTC/USD, based exclusively on the secondary market value of WBTC.
  • For the listing of wstETH, it will be necessary to evaluate the availability of a conversion feed between stETH and wstETH, as the pricing should be through the chain ETH/USD → stETH/ETH → wstETH/stETH.
  • For if not taken into consideration, it is essential to highlight that DPI is a Chainlink calculated price feed, meaning that its value derives from the aggregated value of its underlying assets.

We will be synchronizing with Chainlink to verify that all the requirements in terms of feeds are met before any deployment of Aave v3 Ethereum by the community.

3 Likes

We have published the Snapshot vote below and encourage all in the community to participate.

https://snapshot.org/#/aave.eth/proposal/0x288caef0d79e5883884324b90daa3c5550135ea0c78738e7ca2363243340c2da

1 Like

Hi @Pauljlei - we have expressed this privately but feel concerned about the inclusion of DPI / ENS.

Both seem to be proposed as a result of external partnerships, which are now no longer relevant.

Happy to be more explicit if needed.

Looking at these assets, on V2 - only 43 addresses borrowed DPI at a total of $1,971,056.

ENS had 77 unique addresses for a total of $8,732,512.


xSUSHI, omitted from the list, was borrowed by 167 users for a total of $186,286,217

Or look at ENJ, 292 unique addresses for a total $24,448,819.

I understand there are important risk metrics involved as well, but at face value, some of the other assets would be a better fit - looking at demand and users.

This deliberate choice of long-tail assets feels a divergence of two different strategies. This is a critical time for the DAO to choose: either we list a bunch of assets, boosting a UX and freedom.

Or - we focus on the ones with verifiable, proven demand.

This analysis needs to include the recency of demand but it is important question to consider for developing a new market - which type of assets do we focus on?

We say this as ENS delegates too.

2 Likes

Chaos Labs would like to express our support of this proposal. As we commented before, Aave v3 offers improved risk mitigation mechanisms (such as Isolation Mode, Supply/Borrow Caps) and we believe it is of the community’s best interest for users to migrate from v2 to v3. We believe this initial list of assets is a significant first step that can help achieve this goal. In that spirit, we also recommend adding ENJ as an isolated asset, as it is currently the only asset available to be used as collateral on v2 that is not on the proposed list.

We look forward to the deployment of Ethereum V3 and partnering with other contributors to enhance the ongoing safety and growth of the protocol via parameter recommendations and new asset listings.

2 Likes

Why enabling stETH/wsthETH as collateral at deployment? I understand that’s the current situation with Aave V2, but I wonder if it was the right choice at that time.
Allowing borrowing any asset against stETH in Aave, it’s a significant risk to the protocol solvency which Aave cannot control.
Firstly, although there is a tentative timeline for withdrawals enabled on main-net, Aave cannot know how long it will take and what will be impact of it on stETH/ETH liquid markets.
Secondly, the Lido withdrawal contract is not even deployed yet. Before allowing stETH as non-isolated collateral, I think Aave should wait for multiple audits/formal verification on the final implementation and that the contract is deployed and immutable.

I propose stETH/wsETH to be enabled in isolated mode in Aave V3, pending Lido full withdrawal contract implementation and withdrawals enabled on main-net.

What the rationale for not including GUSD? I don’t see any risks with GUSD being a stable. It cannot be used as collateral anyway, so what is the reason?