Hasu & Hart on Oval & The Recapturing of Billions in DeFi Liquidations

Welcome to Bankless, where we explore the frontier of DeFi market structure and capital efficiency. Today on the show, we're talking about Oval, a new DeFi primitive that Heartlamber from Uma is introducing to the world of DeFi's biggest lending markets. Billions of dollars have been liquidated from protocols like Ave, Compound, and MakerDAO over the years, and these liquidations have been extremely inefficiently priced due to reasons. Reasons that we'll be discussed here on the show today. The long arc of DeFi has been one that trends towards capital efficiency, efficient price discovery, and ultimately better quality products for its users. This episode today discusses a new mechanism to further DeFi's maturity on this very important arc. Hazu also joins us on this conversation today to discuss the way that MEV share from Flashbots is a part of this conversation, as well as helping illuminate the further direction that DeFi is going forward as a whole with mechanisms like this. One more part to this conversation that I want to add here that was unsaid in the podcast. This conversation is about bringing mechanisms into DeFi that allows DeFi to leak less value to MEV bots, allowing DeFi customers and users to retain all of that value and also increasing the economic viability of the DeFi app layer on Ethereum and in crypto broadly. This is a transfer of value away from MEV arbitragers to DeFi apps. What this ultimately means is that this is a transfer of value away from ETH validators, the ultimate beneficiaries of MEV, and towards the users and customers of DeFi products on the Ethereum app layer. As long been known that the ultimate destination of the DeFi landscape, no matter what chain we're talking about, Ethereum, Cosmos, whatever, is that the value that MEV has will ultimately be retained by the source of that MEV, not by any downstream actors. The reason for this is obvious. If you're an application that creates MEV, you're also the entity that has the most optionality of directing where that MEV goes and its conditions for capture. So why wouldn't you just capture it for yourself and improve your bottom line and the quality of your product for your users? This early hypothesis has been stated by many in the MEV world, and it's the thing that the Cosmos app chain believers have been chanting about for a while now. And it's being shown to come true within the innovation being discussed today on the podcast. So enough of my table setting. Let's go ahead and get right into the episode with Hart Lamber from Uma and Hazu from Flashbots. But first, a moment to talk about some of these fantastic sponsors. Bankless Nation, I'm super excited to introduce you to Hart Lamber, the founder and CEO of UMA. Hart, welcome to the show.

Thanks, David. Thanks for having me.

Hart, we had you on two years ago, I think, to talk about bridges way back in the day. Uh it's good to have you back, my man.

Yeah, good to be back. Good to see you.

And a reoccurring guest on the podcast. Well, I guess both are reoccurring, but Hazu's been on a few more times than you, Hart. Hazu, uh, strategy lead at Flashbots, as well as just ecosystem researcher and crypto economic enthusiast, Hazu. Welcome back to the podcast.

Hey David, nice to meet you. Hi Hart. Don't think we've met yet on a podcast, but I'm looking forward to this.

I I like the uh crypto economic enthusiast uh moniker. That's a good one.

I think that's the right title for Hazu. I think that's the right title for most people that come on this podcast, but Hazu, especially.

That doesn't make it very special, David, you know?

The lead cryptoeconomic enthusiast that I know. My preferred.

Okay, guys, we're gonna get into a conversation that I think bankless listeners will be familiar with, longtime bankless listeners. Um, this is inside of the subject matter of MEV, uh maximally extractable value, minor extractable value, uh later rebranded to maximally since we no longer have miners in Ethereum. Uh, but we're talking about a specific kind of MEV, which has been dubbed uh Oracle extractable value. Uh so I think this just falls inside of the category of MEV, but it's uh around a certain uh set of how the MEV arises, which is through Oracles. Uh and this is going to be a conversation that I think talks about uh the way that applications on Ethereum produce MEV through Oracles, uh the sustainability of applications, the healthiness of the aggregate DeFi applications, uh, a lot of these different subject matters. So it's probably gonna be a decently technical episode, uh, but I think we can also make it uh pretty understandable for people who already know what MEV is. Um, Hart, can I ask you to maybe just set up the context of what we are talking about, the foundations for what's going on? Like, like I said, bankless listeners will know what MEV is, uh, but applying it in an Oracle um framing might be a little bit new for them. What is the landscape? What is the problem set that we are going to be addressing here on this episode today?

Yeah, absolutely. Um and we can make it concrete and then maybe generalize it a little bit more too. But um today, Oracle updates, and let's talk about like chain link updates. Chain link updates uh drop into the public mempool on Ethereum.

Chainlink

Yeah.

Price updates, correct. Chainlink price updates get dropped into the public mem pool in Ethereum, and then they get used by DeFi protocols.

And if a chain link price update triggers a liquidation, if that price update drops below a liquidation target on a uh Ave or compound position,

um that liquidation then

triggers a bunch of bunch of things, but basically collateral getting sold.

I mean uh

and it triggers MEV. It's it's not dissimilar from a user submitting a Uniswap trade with slippage.

Um, into the public mempool, which we all understand creates MEV and like creates a sandwich attack and creates other things where

searchers will construct transactions to extract the maximum value possible from uh that uniswap transaction.

Well, basically we have the same thing going on with chain link price updates, where the chain link price update looks very conceptually similar to an unprotected uh Uniswap uh transaction with slippage. And it creates MEV that gets leaked and ultimately captured by Ethereum validators.

And I believe that the amount of MEV that's captured from Ave compound liquidations can be quite large. I think this probably happens more. This amount of MEV capture is more when there's times of volatility and probably times of bull markets as well. And really, just to set this stay the stage here, we are talking about MEV. We're talking about MEV. This is MEV. It just uh we're talking about specifically MEV that happens as a result of.

Price oracle updates, which most of DeFi is chain link. Uh so like Ether is at $1,000, it drops down to very quickly $950. A bunch of DGMs get liquidated, but then there is the fight to be the liquidator of uh these positions. What uh Hart, is there like a some numbers or some measurements or metrics about like how valuable this MEV is, like how much value is being squeezed out of these applications?

Yeah, totally. I mean

estimating MEV uh is quite difficult. Um but within like Ave and Compound, we can have a very clear theoretical kind of bound. Um

and the way Ave and Compound work is when a chain link update price update comes in that causes liquidation, uh Ave and Compound then sell collateral at a discount.

And the reason why they sell that collateral at a discount is because they really want to make sure the collateral gets sold to keep the protocol healthy.

So that discount is between 5 and 10%, depending on the asset.

And if we look back over the history of Abe V2 and V3 and Compound V2 and V3 over the last about two years, we can see that $150 million ish of liquidation discounts have been sold. So there's been liquidations, a billion and a half dollars of liquidations that have caused uh discounts of about $150 million. And that money has been lost to MEV.

Um, if we zoom out and go a little bit further and look at kind of you kind of think of Ave and compound since their existence, I think that number is much bigger. It's like quarter million or quarter billion, 300 million uh uh type number uh that has been lost because of liquidations that are are really selling collateral too cheap.

Hazu, I want to get your perspective on what would happen if this uh were left unchecked. Uh Hart talked about um some parameters about collateral ether or maybe less liquid assets being sold at a discount. Maybe you could talk about um why it's being sold at a discount uh and what uh why that's important to the protocol. And then overall, uh if this version of MEV uh is left unchecked, Oracle extractable value, like uh what would happen into the future? Because we all want, you know, Ethereum, DeFi, any DeFi on any chain to be the world's global financial system. Uh, why do we have to solve this problem? Why is it why is it important?

Yeah, so to your to the uh the first part of your question. So why is it sold at a discount, I think. Um

Hart actually gave a very good explanation here. So when you want to sell or when you want to buy a token on Unisop and the token is quite volatile, then you don't exactly know what slippage to set, right? And so if you set a very tight slippage, then you risk not getting in the next block because maybe the price moves against you, right? And the same is true for a lending market. So if they want to sell a token now, they have this trade off between basically setting a higher slippage, maximizing the chance to

get into the next block, uh, or setting uh basically a lower slippage and risking that maybe the uh the price doesn't take place in the first block, maybe it takes for longer time, right? And so you have this.

So it's just a protective mechanism that they need to sell this collateral because they must not have bad debt.

Yeah, so I mean this is really a design choice by the protocol itself, I would say. So if you're on Unisop and trading shit coins, okay, so what's the worst that can happen if you don't get to buy this position now? Okay, you maybe incur some kind of opportunity cost. But the lending market, um uh you could argue has a much higher responsibility to that capital, right? So they uh manage the the money effectively of many different depositors, right? And so if they let someone get away with bad debt, then that must that has to bad debt has to be socialized somehow. So uh in the first tranche probably would go, in the first instance, it would probably go to the token holders.

of that project as a first line of defense. And if if that doesn't help, then at some point you have to socialize um the money on uh depositors effectively because there's nobody else there, right, in in crypto. You can't you don't have any legal recourse or anything like that. And so that's why traditionally

Uh lending markets have chosen to be very

rigid uh about how fast they want to liquidate in order to minimize the risk that they

internalize any of that bad debt either to their token holders or to

uh even to their depositors.

And um

I think that paradigm

I guess is changing a little bit. Um so I mean if you look if you look kind of uh at at the the slippage that, for example, a protocol like Aave or Compan has been quoting.

Five to ten percent is a lot, and it's more than maybe they need to. And so they tend to bleed a lot of money by doing that. And um

we've seen some protocols basically handle this problem by spinning up their own auction mechanisms.

Um, like Maker, for example. Makeer has um

I believe uh a reverse Dutch auction. So basically the the money that they uh the the price that they charge

um to liquidate this collateral basically starts low and then it increases every block. And and so this is at the opposite on opposite end of the spectrum, right? So where compound says uh I must liquidate this in one block, and so I'm willing to pay a lot for this, or I'm willing to accept a low price, you could also say on the other end. Uh Maker says, you know.

I'm fine. You know, what Make actually does is they only bring Oracle updates on chain with one hour delay.

So the Oracle update comes on chain, but it can only be used for liquidation after one hour. So they actually say, Hey borrowers, we're gonna give you a big fat warning.

Your position is underwater now, right now.

And if you like if you don't actually top it up within the hour, then we'll start liquidating you. But even then, they have this

increasing price auction where they give the

Borrower, even more time, right? Basically, to top it up. And sometimes

transactions take uh or like that takes longer to uh liquidate in Maker. So Maker is a protocol that's very borrower friendly.

And then

a protocol like RV or compound is much more lender friendly, I would say. And so you have these different sides of this of the spectrum. And it's really a design choice, you know, because they compete in the market and