[TEMP CHECK] Integrate Oval for the BAL & SNX Ethereum V3 Markets

[ACI]

[TEMP CHECK] Integrate Oval for the BAL & SNX Ethereum V3 Markets

Title: [TEMP CHECK] Integrate Oval for the BAL & SNX Ethereum V3 Markets

Author: UMA

Date: 2024-01-31

Summary

This proposal seeks to integrate Oval developed by UMA into the BAL and SNX Ethereum V3 markets. This pilot program will enable Aave DAO to capture MEV associated with oracle updates from liquidations that occur in the BAL & SNX markets. This proposal is purposefully small in its initial scope, targeting only two isolated pools (BAL & SNX), to allow Aave to experiment with MEV capture with a bounded, low-risk pilot.

Motivation

Aave incentivizes third parties to liquidate Aave debt positions that have become distressed through a liquidation bonus. These liquidations generate enough profit that searchers compete to send their liquidating transactions immediately after the Chainlink price feed update. The MEV from Aave liquidations has benefitted searchers, block builders and validators, but not Aave protocol, which could have generated up to $62.2MM (Jan 2021 to Jan 2024) via Oval.

Oval allows Aave to capture the majority of this MEV as revenue. Oval is a MEV capture tool for Oracle Extractable Value built on Chainlink Data Feeds and Flashbot’s MEV-Share infrastructure. Oval requires no smart contract code changes to integrate.

Specification

1. What is Oval?
2. Oval Revenue
3. Chainlink & Oval
4. Implementation
5. How Oval Works?
6. Pilot Program Updates
7. Support
8. Security
9. Other
10. Glossary

What is Oval?

Oval enables OEV capture by wrapping Chainlink price updates as they are pushed onchain so that OEV cannot be extracted without the consent of the protocol.

With Oval, when searchers perform a liquidation on Aave an order flow auction is triggered using Flashbots’ MEV Share—a trusted tool that allows MEV searchers to compete for MEV extraction offchain.

To extract the OEV created by an update, searchers must submit a winning bid, most of which is paid to the protocol as revenue. The winner of the auction gets the right to backrun the oracle update transaction—performing the liquidation by adding their transaction to the block immediately after the price update.

If no bid is received by the end of the auction window, the Chainlink price is automatically released and Aave liquidations function as normal.

The diagram below showcases, at a high level, this flow.

This approach utilizes the solid foundation of Chainlink price feeds and MEV-Share to extract OEV from liquidations. Oval is applicable to all markets within Aave that employ Chainlink price feeds.

As a comparison, the diagram below shows the current Aave liquidation process.

Oval is fully compatible with the liquidation fee the Aave protocol currently takes on all liquidation bonuses.

Oval Revenue

As mentioned earlier, Oval enables Aave DAO to capture up to 90% of the liquidation bonus. The liquidation bonus depends on the asset used as collateral. These bonuses range between 5-10% and can be found at the bottom of the dashboard.

The data below offers an in-depth analysis of the potential Oval recapture across the BAL & SNX markets, from January 2021 to January 2024.

Market	Liquidations	Total collateral Liquidated	Potential Oval Revenue
BAL	112	$961,403.07	$76,139.67
SNX	156	$681,286.40	$45,986.83

Source: Aave OEV (Jan 2024) with methodology

Given the fundamental differences between isolated and non-isolated markets, we recommend the integration of Oval initially within isolated markets for this pilot program. This ensures that Oval’s operational scope is strictly confined to these BAL & SNX markets, thereby eliminating any impact on other markets within the Aave ecosystem.

Despite the relatively modest potential for Oval recapture in these isolated markets, this small scope was deliberately chosen to enable the Aave DAO to gain confidence in the integration’s performance. The successful integration of Oval within the BAL & SNX markets will serve as a compelling indicator of its applicability and efficacy for additional markets.

Aave DAO could have profited over $62M if Oval had been integrated into the entire protocol over the past 3 years.

Chainlink & Oval

Oval is not an oracle, and it does not replace Chainlink.

• Using Oval means still using Chainlink price feeds. Oval wraps incoming prices from Chainlink, shielding them from MEV searchers.
• The Oval contract will use the Chainlink price feed and initiate an auction for the OEV searchers to bid for access to liquidations.
• No additional pricing logic is inserted.
• If no one participates in the auction, the price update will be released after 36 seconds (3 blocks).

Implementation

Deploying Oval requires two simple steps:

1. Deploy Aave’s own instance of the Oval smart contracts for the BAL and SNX markets (UMA is happy to assist with this).
2. Update the BAL and SNX market’s oracle address from the Chainlink price feed to the Oval smart contract that points to the same Chainlink price feed.

These steps are further detailed in our docs. UMA is committed to providing ongoing support to ensure a successful implementation of Oval.

How Oval Works

Integrating Oval requires no code revisions to Aave smart contracts. Currently, Aave smart contracts receive price data directly from Chainlink price feeds. To integrate Oval, Aave would instead receive Chainlink price data through Oval smart contracts. It’s crucial to note that Oval cannot modify Chainlink price data in any way. The only function of Oval smart contracts is to temporarily lock the latest Chainlink price from Aave so it can be auctioned off to MEV searchers. The steps below outline the basic Oval flow and how it integrates with Chainlink and MEV-share. Please see the docs page here for more info on this flow:

1. Chainlink price update is released to the public Ethereum mempool.
2. Searchers see this update and submit liquidation bundles to the Oval Node to backrun the update and perform a liquidation.
3. The Oval Node creates a transaction that unlocks the newest price (unlockLatestValue). This unlock transaction is added to the searcher bundle along with refund instructions and forwarded to MEV-share. This is done for each searcher bundle sent to the Oval node.
4. MEV-share forwards all bundles to all connected builders; this includes all the main block builders. The builders run a standard Flashbots auction, selecting the winning bundle based on the builder payment. Because of the refund instructions, the builder is required to refund most of this payment to the protocol (Aave).
   1. The refund will be directed to a multi-sig controlled by Aave DAO stakeholders & UMA.
   2. Oval’s revenue distribution: For the pilot program, we propose utilizing the revenue to fund Aave OEV research (i.e. to evaluate the efficiency of Oval on the BAL & SNX Markets).
5. The winning bundle is selected by the builder and is sent within the proposed block to the proposer.
6. The proposer appends the block to the Ethereum chain. The relevant parts of this block execute the following actions, in this order:
   1. Oval unlock transaction
   2. Winning searcher bundle (may include Chainlink price update)

Due to possible inclusion delays within the Ethereum MEV/PBS supply chain, in some situations there can be a small delay between steps 4 and 6. For example, if the proposer for a given slot does not use mev-boost (90% of proposers use mev-boost). In this case, the searcher bundle would not land in the same block as the Chainlink update and might execute in a subsequent block. Oval bounds the delay that integration must wait before receiving the latest price to 36 seconds (3 blocks). This means that—in the worst case scenario—a Chainlink update will be delayed 36 seconds (3 blocks). For more information on how this could happen and the mechanism that controls when to release the price see here and here.

Pilot Program Updates

We will create a dune dashboard that will track Oval revenue recapture and provide quarterly updates on Oval’s performance within the BAL and SNX markets. These updates will be shared in the “UMA [Oval Updates Thread]” on the Aave forum.

Support

Oval documentation contains all the information to perform an integration. In addition to these docs, Aave DAO will benefit from the dedicated support of the UMA team to facilitate this integration process.

To ensure seamless collaboration and efficient problem-solving, we propose establishing a direct communication channel between the UMA and Aave teams.

Security

• To ensure the security of this system, Oval has completed an audit by Open Zeppelin (public report available soon) and has an additional security bounty of up to $1M for any vulnerabilities that are identified.
• UMA has worked closely with Flashbots to develop Oval. Flashbots has updated the MEV-Share design to add support for Oval, and is committed to helping support this pilot.

Other

• The integration requires no modifications to Aave’s UI and does not impact any normal user actions of creating new positions, increasing debt or repaying loans.

Glossary

Maximal Extractable Value (MEV): the maximum value that can be extracted from block production in excess of the standard block reward and gas fees by including, excluding, and changing the order of transactions in a block.

Oracle Extractable Value (OEV): a subset of MEV created by oracles posting new price data onchain.

Oracle Value Aggregation Layer (Oval): an OEV capture product developed by UMA

MEV-Share: an orderflow auction infrastructure built by Flashbots that enables MEV capture for private transactions.

Liquidation Bonus: a financial incentive for participants to execute liquidations.

Useful Links

• Aave Liquidations
• Oval documentation
• Flashbots MEV-Share
• Aave OEV Dashboard
• Oval Launch Post

Disclaimer

This proposal is powered by Skywards. The Aave Chan Initiative is not directly affiliated with UMA Team and did not receive compensation for creation this proposal.

Next Steps

1. If consensus is reached on this [TEMP CHECK], escalate this proposal to the Snapshot stage.
2. If the Snapshot outcome is YAE, this proposal will be escalated to ARFC stage
3. Publication of a standard ARFC, collect community & service providers feedback before escalating proposal to ARFC snapshot stage
4. If the ARFC snapshot outcome is YAE, publish an AIP vote for final confirmation and enforcement of the proposal

Copyright

Copyright and related rights waived via CC0.

[hal2001]

Hey all, I’m the founder of UMA and my team wrote this proposal. I wanted to underscore a couple points:

In recent days the idea of oracle MEV (or OEV) capture has gotten some good traction. There seems to be a growing consensus that this is a big problem worth solving. This proposal uses Oval to capture OEV in two smaller isolated v3 markets totalling ~$10m in TVL. Given Aave’s ~$10B in overall TVL, this represents just ~0.1% of Aave’s total TVL.

This pilot program is small on purpose: it’s an easy way for Aave to experiment with OEV capture without affecting the broader protocol.

By our calculations, Aave has lost >$60m in OEV in since the launch of v2. This number is likely >>$100m if you include Aave v1. This is real revenue that Aave deserves to keep. Separately, Chainlink is not currently paid for the oracle service they provide—this is unsustainable. Oval, and OEV capture, is an elegant way to both return value to Aave itself, and to pay Chainlink for the useful oracle service they provide.

Obviously this tiny pilot program will not capture $60m in revenue for Aave, but, if approved, it provides a path to both capture OEV for Aave’s benefit and to pay for Chainlink for their oracle services.

I’m looking forward to hearing the Aave community’s feedback.

[Kene_StableLab]

This proposal goes forward to identify an opportunity for the Aave DAO to extract additional value from its markets.

@ACI

The MEV from Aave liquidations has benefitted searchers, block builders and validators, but not Aave protocol, which could have generated up to $62.2MM (Jan 2021 to Jan 2024) via Oval.

The potential revenue which could have been generated is a clear indication of the need to take steps to begin to capture this value. This initial pilot for this proposal is a great first step towards ensuring that Aave benefits most from its protocol-owned MEV.

[Michigan_Blockchain]

Thanks to the Oval team for presenting this proposal and highlighting an old issue in the DeFi lending protocol sphere. Regardless of the specific design of the OEV capturing mechanism, from the DAO’s perspective, there’s no downside to embracing this concept, especially considering its potential to unlock significant revenue amidst rising expenditures.

Since UMA’s announcement of Oval about two weeks ago, we’ve observed notable liquidations across major assets on v3 Ethereum. So it’s good to have proposers initiating a pilot with less mainstream assets as a cautious approach to evaluate Oval’s reliability.

We also echo concerns raised by other community members regarding the added layer of centralization introduced by Oval. Despite the ease of implementing the OEV capturing mechanism it offers, exploring alternative solutions from native Oracle Providers or third-party integrations like Oval, and assessing whether the revenue generated justifies the operational costs, remains a priority for the Aave DAO. It’s encouraging to see the Oval team’s commitment to transparency and ongoing dialogue with the community.

As a delegate, we believe an OEV solution that enhances user experience, increases DAO revenue and maintains trust-minimization and efficiency would definitely be worth pursuing in the future.

[UMA]

Thank you @Michigan_Blockchain for your comment and support.

The most effective way to capture OEV is to run an auction for liquidation opportunities. There are no fully decentralized means of doing this today, so Oval chose to build it’s auction mechanism upon a battle tested, industry leading MEV infrastructure product, Flashbots MEV-Share. This auction is started by a trusted auctioneer that Aave can change at any time. Importantly, the worst thing the auctioneer role can do is steal the OEV that is currently being lost. The auctioneer cannot manipulate the Chainlink price, nor can the auctioneer prevent liquidations or Chainlink price updates from occurring.

[0xkeyrock.eth]

We’re supportive of this proposal as it clearly indicates a massive opportunity for the DAO that needs to be tested with isolated markets before considering any expansion.

If this proposal progresses, we would like to hear the opinion of risk providers as well on potential research that can be conducted around this subject

[lbsblockchain]

We’re broadly supportive of this initiative and have engaged deeply with the UMA team on this, who have been very prompt in answering our queries. Overall, this is a worthy experiment to run and given that MEV (OEV) is a Pandora’s Box that has been opened, the Aave protocol should benefit from it. We are encouraged by experts like Hasu showing their support to UMA for Oval and also want to commend UMA on how they have proactively reached out to a variety of Aave delegate platforms.

The main concerns we have are below:

• The centralisation vector with Oval, although this has been mitigated by Oval being built on top of MEV-Share which is extensively used across the ecosystem.
• The security trade-offs of block confirmations being potentially delayed - this will be key to test out during this initial experiment.

It would be great to hear from @Gauntlet and @ChaosLabs on this as well - their opinions will be invaluable.

[EzR3aL]

I hope i can join todays call from UMA to talk about Oval.
My main concerns have been adressed by other delegates.

Still I want to highlight them:

• Centralisation risk
• Auction system (e.g. API3 has an open auction system on a zk rollup for full transparency)
• block delay (we have seen what seconds did in the past, what will happen in a black swan event with all liquidations?)

Aave has always been top notch regarding security thats why I at least don’t feel good trying something new, even if it is isolated. At least for now.
Has Oval been integrated somewhere else yet? If yes which protocols are using it and whats their experience?

Overall I agree that OEV/MEV is a bigger problem which results in lost revenue for everyone.

[UMA]

As you’ve mentioned, Oval is built on tried and tested Flashbots MEV-Share infrastructure.

Regarding the block delay, we will break that down further below:

• In order to capture OEV, the Oval auction must be run through Flashbots MEV-Share, which enforces the refund parameters.
• Approximately 90% of all Ethereum blocks are built by builders connected to Flashbots MEV-Share. This means that 90% of all Oval auctions should be mined in 1 block, with NO DELAY to price updates compared to not using Oval.
• However, if the Oval auction occurs in a block that is not created by a Flashbots-connected builder (approximately 10% of all blocks), the price update is delayed by an additional block so that the OEV can be captured.
• This second block has a 90% chance of being included in the MEV-Share infrastructure, meaning there is a ~99% probability of the Oval auction being captured in 2 blocks.
• This process can be repeated with protocols choosing to extend the delay to whatever window fits their use.
• If an Oval auction passes the delay, then the price becomes freely available and no OEV will be captured.

90% of the time you capture OEV in 1 block; 99% in 2 blocks; 99.9% in 3 blocks. If Aave doesn’t want to delay more than 1 or 2 blocks, that’s amendable, but Aave may not capture all OEV.

In practice, when market prices are volatile, liquidations will be available and searchers will bid to liquidate them the same as they do today. In these conditions, the price will unlock to the Aave market in the same block as the Chainlink price update. In flat market conditions, there may be no liquidations and, therefore, no searchers bidding to unlock the newest price update. In this case, the price will be delayed by a maximum of 3 blocks. This would have no effect when the market is flat.

We have been in conversation with both Gauntlet and Chaos Labs, and we welcome them to comment now or at the ARFC stage, per their SOP.

[UMA]

Aave should take caution with introducing new technology, and that is why we are presenting this isolated pilot program. We will be taking a data-driven approach here, leveraging the expertise of Aave’s risk providers to ensure that Aave DAO has the relevant information to make an informed decision on this integration and future integrations.

It’s important to point out that Aave is not shy of upgrades and innovation. The recent Aave v3 governance is a good example of this. GHO is another example. OEV capture is another “upgrade” that Aave can tap into with a single contract address change, low dev effort and comes with strong security assurances. We have built on top of MEV-Share and Oval has undergone an audit by OpenZeppelin.

API3 has built an interesting solution, and we have great respect for their team and design. However, API3’s OEV capture solution requires utilizing API3 price feeds. Oval utilizes the same Chainlink data feeds that Aave uses today; we do not need Aave to swap their price feeds in order to receive OEV revenue.

We are currently in talks with numerous other protocols regarding Oval integration.

[Gavi]

Hello Aave,

Standard Crypto had the chance to share some of our questions and feedback directly with the Oval team. We wanted to share our thinking here and the reasons for our support.

We’ve dug through the details of the proposal and found this to be a clear source of value, with a mechanism that is simple to understand and which asks for minimal changes to the Aave protocol. We recommend walking through the mainnet fork integration tests for those interested in understanding the practical effects on the liquidation flow.

Our primary questions to the Oval team were themed around the following:

1. How should the Aave community evaluate the success of this pilot when later considering proposals to expand Oval into larger markets? We imagine this to involve some combination of revenue figures, a count of happy-path liquidations, a retrospective from one or more of Aave’s risk service providers, and a history of no observed exploits or bugs.
   We figure it may help the community for the authors of this proposal to explicitly highlight how they plan to evaluate the pilot. It’s also worth highlighting that the SNX and BAL markets have historically seen ~1-3 liquidation events per month collectively (source), so we should expect to see tangible data within a few months.

2. As already discussed in comments above, we were interested in understanding the implications of the potential added 3 block delay on liquidation events in the unhappy path – both understanding how this should affect Aave’s collective thinking around risk modeling for the protocol, and what could go awry that might prevent instantaneous liquidations. Notably, we think it’s worth discussing any censorship concerns or any uptime requirements that the happy-path scenario introduces with respect to the Oval Node, which we understand would initially be the sole entity to handle requests by liquidation searchers whenever chainlink feeds are updated.

We were thoroughly pleased with the team’s responses to our questions, and invite them to include their thoughts here for community reference.

We are excited by this opportunity for Aave holders to see added protocol revenue by gaining control of the already-existing markets on Aave’s fixed liquidation bonuses. The potential revenue opportunity makes this something well worth exploring, as long as all community concerns are addressed. This pilot feels like the appropriate choice of scope, and would allow for feedback from Aave’s risk service providers.

Thanks to the Oval team

[Kene_StableLab]

An important aspect of this proposal is that this is a trial run which will only be tested out on the BAL and SNX Markets for Aave V3 Ethereum, there is significant time and opportunity to deprecate this in the event of an unsatisfactory outcome.

[UMA]

UMA hosted a call for Aave delegates and invited Chainlink to attend to discuss this proposal in an open format to discuss the pros and cons. Thanks to those who attended the call. The recording can be found here, and the timestamps are below:

Attendees: 0.00
UMA Intro: 1:42
Proposal TLDR: 2:25
Q&A: 3:10
Next Steps: 27:45
Closing: 29:00

[UMA]

Hey Gavi and Standard Crypto. The team put a lot of time and thought into the design of Oval to simplify integrations for protocols such as Aave. We’re glad to see that came across and we encourage others to walk through the mainnet fork integration tests and check out our demo. We appreciate your comments and support of the proposal.

@Gavi

How should the Aave community evaluate the success of this pilot when later considering proposals to expand Oval into larger markets?

For a retrospective evaluation of the pilot program, these are some of the key points you can expect to find in the report:

• BAL & SNX Market
  • Liquidations occurred (with and without Oval)
  • Revenue earned
• Retrospective analysis of Oval from one of the risk service providers [Gauntlet or Chaos Labs]
• Revenue losses without Oval integration
• Next Steps

We will ensure that Aave DAO has the relevant information to make an informed decision for the next steps of Oval.

We will work with one of the risk service providers like Gauntlet or Chaos Labs who can evaluate the impact of Oval on these 2 smaller markets and share their research with the Aave DAO to help in their decision-making to expand to further markets. We want Aave DAO and the risk providers to have enough time to see a fair amount of liquidations prior to performing the research, so I expect the pilot to last a minimum of 3 months and will adjust if needed.

@Gavi

As already discussed in comments above, we were interested in understanding the implications of the potential added 3 block delay on liquidation events in the unhappy path – both understanding how this should affect Aave’s collective thinking around risk modeling for the protocol, and what could go awry that might prevent instantaneous liquidations.

As mentioned in our reply to LBS, we have shared our rationale for having a 3-block delay for MEV Share searcher inclusion since this achieves a 99.99% inclusion rate.

From our research, we expect that the max delay would only occur when there are flat market prices, meaning no liquidations are possible or in the unlikely event that Oval Node or Flashbots’ MEV-Share is down. When market prices are volatile, liquidations will be available and searchers will bid to liquidate them the same as they do today. In these conditions, the price will unlock to the Aave market in the same block as the Chainlink price update.

Aave’s risk service providers take a conservative approach with their methodology and we expect them to do the same with Oval, therefore considering the max block delay in their analysis. On that note, both risk providers have had an opportunity to look at the proposal and we hope to hear their opinion at the ARFC stage, as per the SOPs. As mentioned in our Oval evaluation report, we also hope to leverage Aave’s risk service providers to evaluate the impact of Oval on liquidations from the pilot program.

[bgdlabs]

Thanks @UMA for the TEMP CHECK submission, and props on the innovation being brought to the space.

Even if the solution seems reasonable, from a high-level perspective, we think that most probably the community should not take an integration decision like Oval’s lightly, and do introspection about alternatives. The rationale is the following:

• First and foremost, liquidations is arguably one of the most important mechanism of Aave and similar protocols, and given the stakes, stability is key. Yes, it can be improved, but it should be deeply evaluated from all perspectives: technical, risk, strategy.
• Auctions-based systems introduce without doubt additional centralisation. We agree that the suggested architecture seems to not add major trust compromises, but by nature, these type of designs add some dependencies.
• Having a high liquidation bonus is a pessimistic approach to security on liquidations. The idea is the following: it is perfectly fine to overpay liquidators if that means that Aave will be really attractive on market distressed scenarios. The goal of the liquidation bonus is not maximising profit for the DAO, is about protecting all users.
• A system with deeper integration directly with the price provider of the protocol (Chainlink) seems a more promising direction to explore, reducing any external dependencies, a strategy what we usually defend on critical components. We welcome comments and ideas from representatives of Chainlink on the topic.
• This type of decisions should be holistic. No matter if the technical integration looks first sight trivial or if starting with only a reduced number of assets, the operational cost could be relatively high.
• Innovation on top of Aave’s liquidation system could be a good area to improve together with AAVE token dynamics. If that’s the case, we don’t think that should be done by a party with no history with the DAO.

In summary, from the technical side, we think it is probably better to hold on the integration of a system like Oval, and study deeper if it is good long term for Aave, and potential alternatives.
We will try to come up with some idea to consider by the community.

[hal2001]

@bgdlabs, thanks vm for your comments.

Zooming way out, I think it’s important to consider the opportunity vs the risks here, and to remind forum readers that this is only a TEMP CHECK proposal.

The opportunity: Aave can capture >>$25m/year in additional revenue, more than doubling the protocol’s current revenue. This should benefit $AAVE meaningfully. I also believe that Oval can actually make it safer for Aave to onboard more collateral types, allowing Aave to expand its product offering and capture even more value (more on this below). Simply put: the opportunity is big!

The risks: As you’ve said, liquidations absolutely MUST happen. Any changes to the liquidation process need to be considered extremely carefully. However I genuinely believe that we have robust answers for all risks and all edge cases (more on this below too).

From my perspective, the opportunity vastly outweighs the risks, provided that Oval is carefully researched and implemented. We purposefully proposed a very limited pilot program with this exact goal in mind: this proposal is for two isolated markets that represent 0.1% of Aave’s TVL. This phased approach allows Aave to fully understand the risks and gain valuable experience with OEV capture in a risk bounded way, which, if successful, will allow the protocol to safely expand to capture the full opportunity.

I also want to remind everyone that this proposal is currently at the TEMP CHECK stage. Passing this stage does not mean Aave will choose to implement Oval; it simply means that the DAO believes this proposal is worth investigating more deeply. At the ARFC stage Aave can and should go far deeper on any concerns, and we will have both Gauntlet and Chaos Labs closely analyze all risks as they are required (and excited) to do.

Some more detailed responses on your specific concerns:

First and foremost, liquidations is arguably one of the most important mechanism of Aave and similar protocols, and given the stakes, stability is key. Yes, it can be improved, but it should be deeply evaluated from all perspectives: technical, risk, strategy.

Fully agree. Our proposal has carefully considered this and we are openly engaging the Aave community on these topics. My hope is that Aave will find this proposal compelling enough to pass this TEMP CHECK stage. If that happens we’ll be going wayyy deeper on all concerns at the ARFC stage where Gauntlet and Chaos will be engaged to provide their analysis.

Auctions-based systems introduce without doubt additional centralisation. We agree that the suggested architecture seems to not add major trust compromises, but by nature, these type of designs add some dependencies.

I really appreciate your review of Oval’s trust assumptions and acknowledgement that Oval does not add any major compromises. Today, auctions are occurring with every Aave liquidation—they are simply occurring at the block builder level in the existing MEV supply chain. ETH validators are earning these auction proceeds and Aave is simply being taxed. My belief is that moving this auction into Oval + MEV-Share doesn’t change these centralization vectors. Again, Gauntlet and Chaos Labs are well positioned to opine on this at the ARFC stage.

Having a high liquidation bonus is a pessimistic approach to security on liquidations. The idea is the following: it is perfectly fine to overpay liquidators if that means that Aave will be really attractive on market distressed scenarios. The goal of the liquidation bonus is not maximising profit for the DAO, is about protecting all users.

I fully agree that the goal should not be maximizing DAO profits: it should be about efficiently protecting all users. Today Aave is overpaying liquidators on every liquidation. This does not give users any additional protection—this overpayment is simply getting lost to ETH validators in the searcher → block builder auction. Oval (via Flashbot’s MEV-Share) simply redirects these auction proceeds back to the protocol.

Moreover, you could imagine Oval being used as a tool to improve Aave’s security and expand its collateral types. Since Oval can efficiently set a market clearing price for all collateral sales, you could imagine Aave choosing to set liquidation bonuses to higher levels, and then refunding some or all of the collected revenue back to the borrower that was liquidated. This would make Aave safer. An example: assume $FROG is a high risk asset. Aave could onboard $FROG with a 30% liquidation bonus and send any captured OEV back to the liquidated borrower so their liquidation was “efficient”. (We are NOT proposing this change in this proposal, but it highlights how powerful this tool could be for Aave).

A system with deeper integration directly with the price provider of the protocol (Chainlink) seems a more promising direction to explore, reducing any external dependencies, a strategy what we usually defend on critical components. We welcome comments and ideas from representatives of Chainlink on the topic.

Aave is losing OEV today. To use a Chainlink solution, Aave would need to wait for a potential Chainlink solution to be built. Oval is a mechanism that can be turned on today, uses Chainlink price feeds, and is built on top of battle tested Flashbots infrastructure.

My view is that Aave should begin experimenting with OEV capture as soon as possible. Chainlink and others will be positioned to build a better mechanism for Aave if they can observe Oval capturing OEV in real Aave markets. We welcome this. We also welcome feedback from Chainlink here in these forums or elsewhere. (Earlier this week we hosted a call for Aave delegates and invited Chainlink to attend—we want feedback from all stakeholders!)

This type of decisions should be holistic. No matter if the technical integration looks first sight trivial or if starting with only a reduced number of assets, the operational cost could be relatively high.

We agree that this proposal should be reviewed thoroughly from all angles. After a simple integration the ongoing operational effort for Aave to run Oval will be zero. Importantly, there is also no vendor lock-in after introducing Oval. Aave can easily remove Oval by simply changing the market’s oracle pointer back to the Chainlink price oracle.

Innovation on top of Aave’s liquidation system could be a good area to improve together with AAVE token dynamics. If that’s the case, we don’t think that should be done by a party with no history with the DAO.

One of the most valuable ethoses of DeFi is open innovation. The UMA team has been building in DeFi for over 5 years (since before DeFi was a term!) and has a long track record of providing value to DeFi protocols. I do not think our proposal should be eliminated based on Oval not being a current Aave partner—we have a long history of being a good actor in DeFi.

That said, I do feel a bit silly for not previewing this proposal with you (@bgdlabs) more thoroughly before posting it. You have an important and well respected position in the Aave community, and we want your feedback. I’d love to host a call (public or private) to go deeper on your concerns and also express our genuine excitement for what Oval can do for Aave.

[UMA]

Aave DAO, delegates, and stakeholders,

Thank you for considering this proposal to improve Aave. We appreciate all the open discussion in the above comments and in the delegate call (see link above for recording). We would like to provide a short summary of the pros and cons of this proposal:

Pros:

• Oval could have generated up to $62.2MM of revenue for Aave between January 2021 to January 2024.
• Oval requires no code changes to Aave and no ongoing operational effort for Aave.
• Oval works with the Chainlink price feeds Aave already uses and can not manipulate that data.
• This proposal is for a pilot program that would only apply to the BAL and SNX markets which account for 0.1% of Aave’s TVL. The findings of this pilot program can help guide any type of OEV recapture Aave decides to use in the future.

Cons:

• The MEV auction that Oval runs does introduce some trust-minimized centralization. The worst outcome of this centralization is that UMA or Flashbots acts dishonestly to steal the MEV that should be distributed to Aave. In this worst case outcome, Aave would lose the MEV that it currently does not capture—in other words, it’s no worse off than it is today. If this were to happen, Aave can turn off Oval.
• The Oval auction mechanism introduces a lock window that can cause a minor delay in Aave accessing the most recent Chainlink price under unlikely circumstances. If a Chainlink price is updated in an Ethereum block built by a Flashbot’s MEV-Share supported builder (90% of all blocks), there will be no delay. If delayed, the subsequent block would also have a 90% chance of unlocking the latest Chainlink price. This would continue on until the lock window delay is exceeded and the price automatically unlocks. We have initially proposed a lock window of 36 seconds (3 blocks) to capture 99.9% of liquidations. We are open to revising the lock window length and look forward to Gauntlet and Chaos’s analysis of this delay at the ARFC stage.

Based on the data above and the above comments from Aave delegates, OEV recapture on liquidations is a huge opportunity that warrants Aave’s current consideration and analysis. That is why we think this proposal warrants being voted through the TEMP CHECK phase into ARFC. Of course, any protocol change that affects liquidations should be carefully and thoroughly considered. Gauntlet and Chaos will conduct a thorough analysis of this proposal, provide relevant data, and advise on any potential risks to Aave during ARFC. At that point, Aave delegates and stakeholders will be best equipped to suggest revisions and ultimately make a decision on whether to integrate an Oval pilot program.

[aleks-larsen]

Blockchain Capital is supportive of this proposal. A couple of points that influence our thinking:

1. We were initially attracted to Aave in part because of its emphasis on security. While we strongly believe that this should continue to be a cornerstone of the DAO’s decision making process, it’s important to continue to strike a balance with innovation. The approach suggested by the UMA team to test with isolated and smaller markets reflects consistency with such a balance. We also appreciate the UMA team’s commitment to work with risk service providers to evaluate the impact of Oval and share learnings with the DAO and would like to see the final proposal include a plan for making a third party analysis available.

2. We’ve worked with the UMA team since backing them in their seed round in 2018 and know them to be a high integrity, high competence team with a long track record of shipping complex products in this space. They are the type of team that we believe Aave would benefit from more collaboration with.

3. The fee potential if Oval works is substantial and improves outcomes for AAVE DAO members. An increase in fees of such a scale could provide funding for improved security measures, protocol maintenance and other new innovations that may contribute to the growth and success of Aave.

Thanks to the UMA team for a comprehensive proposal and for continuing to address the concerns of community members!

[WintermuteGovernance]

Great proposal @UMA, super interesting!

Consistent with the nature of a TEMP CHECK we are in favour of this proposal and believe the DAO should consider further exploring the Oval integration for BAL and SNX markets. It makes a lot of sense from Aave’s POV to be auctioning off their liquidations and capturing the fee that already goes to validators.

In saying that, for us to be in favour of the integration at the AFRC phase, we’d like to see a more thorough analysis of the effect of longer auction periods on potential bad debt specifically for longer tail assets with lower liquidity and under high volatility scenarios. Such results should also better inform the length of the max lockout period. Right now the 3 block justification is purely focused on capturing 99.99% of OEV and not necessarily ensuring liquidations are done in a timely manner.

For example, is capturing 99% of OEV under a 2 block max lockout window worth it if it means the protocol is safer from bad debt? Or does such an argument not hold up because liquidator behaviour is expected to be approximately the same under Oval as it is currently during such a scenario?

[UMA]

Thank you, Aleks and Blockchain Capital, for your support and comments.

We want to emphasize one of Alek’s points, which is that we are committed to working with Aave’s risk service providers both during and after the governance process to ensure that Aave DAO has the relevant analysis to make an informed decision.