DAPPCON 2024: Revolutionizing DeFi With MEV Aware Oracles - Ugur Mersinlioglu

Transcript

my name is uger um I work at API 3 jack of all trades uh for those of you who don't know API 3 or have never heard of it um an oracle project uh there are many of those out there I'm pretty sure you uh heard probably all of these names and uh more specifically in the context of let's say defi uh all of these in one way or another provide you with the same thing that all of you want please give me a price feed of let's say e ussd obviously there are nuances to it everyone does it slightly differently there's different levels of centralization verifiability and so on but especially during a bull market dabs tend to not care about that they just

want a certain price on a certain chain uh and all of these projects essentially doing the same thing are competing uh on uh beautiful chart let's say uh which is uh the defi Lama TVs chart and that is essentially the value that all of these Oracles in one way or another secure by providing price data to certain protocols and I uh titled this slide meaningful numbers without meaning and there's a certain reason for that uh we are somewhere at 9 and uh at January this year we weren't even on this list we had like 10 million or so I think in the next couple of days it's going to be at like 1.5 billion but uh our revenue

is unchanged so it's it was zero and it is zero and still and we could be securing 10 billion and it's still going to remain zero so the Oracle game essentially is a subsidized what I like to call mess uh nobody pays uh but everybody somehow secures value and uh The Meta so to speak to make money as an IAL project is the following uh you climb the corporate ladder uh AKA defi Lama uh you do that with all means necessary uh you bribe people if you need to you promise them tokens if they switch over to you once you have somehow attracted a couple of protocols and are somewhere there up

there and you have made your name for yourself you Market that heavily like look how much we're securing and the fact that you bribe people to get there doesn't matter the fact that you make zero money also doesn't matter and then you initiate phase three which is you just sell your token as heavily as you can because you need to keep these feeds running there's costs associated but there's effectively no value capture mechanism for your token or your token holders and uh effectively that brings us to this situation uh nobody makes money every Oracle project somehow just dumps on you relentlessly uh from a d perspective especially during bull markets nobody actually underlyingly very much so cares what they use

everybody is kind of the same just give me a price and business development is somehow reduced to promising people token like if you switch over we'll give you x amount of tokens worth let's say 100K and this is a very unsustainable business model or completely unsustainable uh essentially situation so if you want to build price feeds let's say there's no way you're going to make money with those price feeds because some other project is going to come they going to create a token with potentially a point system seem to be very attractive right now some point system if you switch over and they're going to heavily subsidize it because that has been the name of the

game for the last four to five years essentially this is something we did not want to take part in and also everybody kind of does the same thing they do price feeds so how do we do something slightly different like how do we do a something more offer anything more than just simply a price and that is why we essentially how we got to focusing on something called oev uh OE stands for Oracle extractable value uh you already got a little bit of an meev introduction in the last presentation so essentially any uh value that can be extracted due to Oracle updates and there is a lot of examples for this uh I'll do the most basic one that everyone will most likely understand uh everyone

here probably knows those GMX uh it's a derivatives platform so you can go let's say 50x long on eusd the nuanced thing about it is it's not based on an order book like there is no other side actively uh taking the trade against you like it's LPS but no essentially order book style you take that trade on the basis of an oracle price now the Oracle prices unchain and when the Oracle updates you see that in the menol so what you can do is you can monitor the mempool let's say if is at 3,000 and you see the next Oracle update coming in at 330 you place a trade go 50x long right before the Oracle update and essentially

close your position right after and that's guaranteed profit there's nothing that anyone can do to stop you from this and this has been going on for GMX in late 2020 where uh for 40 days essentially somebody just continuously uh opened the trade before an oracle dat closed right after and funneled off like a quarter million in potential profits just the most basic example of uh o uh how big is the issue uh this is a past number this is is closing in on when well of half a billion that has been lost across numerous protocols and it's lost in the sense that users of these Protocols are bleeding this value away needlessly and I'm going to in this

presentation go heavily into uh Landing protocols because it's the most easy to explain so Landing Protocols are utilizing oracles to determine what your collateral potentially might be worth uh when you then take a loan against that collateral let's say you put an eth in and take out 2,000 usdc uh you essentially have a health Factor as long as this health factor is above one you're fine you can just go on about your life use that usdc ape some more eth but the moment essentially e drops in and your health Factor drops below one uh you become liquidable or liquidable sorry and uh what determines that is essentially the Oracle price change onchain and in order to

essentially not risk getting into bad depth these Protocols are offering up a percentage of your collateral as a bonus to get anyone that wants to to compete to liquidate you and uh we say it's 5 to 10% I recently also saw some insane numbers 5 to 20% essentially of your collateral being offered up in case you get liquidated now the the bad thing about this is or like first of all the the flow like if you get liquidated on a landing protocol the flow is typically that the Oracle updates your health Factor drops below one and then uh Searchers see that your position is now liquidable and a bunch of other

Searchers might also see it and they compete in Block space auctions to have their transaction included first so they bribe block Builders validators to have their transaction included above others and then essentially you get liquidated the Searcher makes whatever is left after paying out uh the bribe the uh interesting thing here however is that these compet this competition for liquidations is uh quite significant this is a quote from uh moneros report called the M book and uh basically on the right side over here you see the largest liquid on the Ava protocol this year in January for 4.2 million uh you see essentially

here how much was available as a reward for this liquidation and you see below how much the Searcher paid as a bribe in order to execute this to walk away with $60 59 uh sorry 945 cents so essentially this is a common occurrence uh Protocols are offering up up to $141,000 in rewards but the Searcher is willing to take $60 and use the rest to bribe his way to $60 and the funny thing about this entire concept that is happening on chain is all of this is only possible because Oracle projects so far have decided not to take advantage of their

privileged position and what I mean by that is that the entity deciding what the price on chain is could in theory simply bundle the Oracle update together with the liquidation so all this of people seeing a position would never happen because together with the update automically the position would be gone so there's no chance for anyone else to catch that position which obviously opens uh the question uh if you have essentially this super entity with uh deciding the prices why should anyone else compete like you're going to have nobody else looking for liquidations and now you end up in the situation where if the Oracle

project ever fails to liquidate there will be nobody else to liquidate and you're kind of wrecked so Oracle projects have collectively so far decided to ignore liquidating on lending markets for this potential danger so they just update blindly they don't care what potential OE happens across whatever many protocols so as a summary of how liquidations look right now we have established that the fees are to D High because we know that Liquidators are giving away 99% of what they're being offered and in theory the oracles hold all the power so how do we solve this and we solved this or are trying to

solve it there's very very different approaches to it by building an auction mechanism on top of the Oracle by simply moving the auctions essentially that currently already happen on the Block Space level and making people compete for the Oracle update that allows for the liquidation instead and how we do that is uh essentially with what we call the OE Network the OE network is an oracle market and uh like a rollup with an auction like an auction platform that is essentially hosted on a rollup uh here you see essentially our uh General infrastructure uh API 3 operates uh what we call first party oracles these are data businesses in the space like coin

Geo operate the oracles directly themselves and they maintain a feed on for instance mantle like ethusd this feed is slightly delayed by 15 seconds and they also sell in real time real-time data without that 15-second delay on this rollup and now you essentially as a Searcher can uh you can see it as an Express Highway if you see a certain price that you like where you can perform a certain action for let's say liquidation you can get the price immediately here for competing for it go to the chain update and essentially front run you don't have to wait for this to happen for this 15sec

delay to pass you can just essentially scoop in and uh get the potential liquidation and whatever you paid for it here is enshrined in the meta transaction that we give you and you pay the dab in the process to do that so on this rollup uh there's essentially a auction ran for each data feed on each chain for each stab so essentially there is mantle USD for inet Capital which is a dab on mantle you could have uh e USD for compound on arbitrum you could have BNB USD for Venus on BNB chain and the Searchers essentially uh can place Bits let's say they say they pre-calculate

there's a massive position that is liquidable at 3,000 eth uh it it gives me a reward of a million dollars I'm bidding 900k the moment the oracles essentially hit 900k the answer is posted here you can take that answer and that answer enshrined has essentially how much did you bid which chain did you win on which dab did you win on and uh all of the Oracle answers that you need in order to update the data feed and you can essentially bundle This Together directly with your liquidation uh the dab gets paid you get the remaining reward and uh the position is liquidated and what then happens is essentially that these uh 41,47 that currently are used as a bribe

to uh effectively uh or like the cost rather here to uh pay validators is now returned to the dab so that's a massive amount of money uh that you now have at your disposal and what do you do with that and that is essentially the core of the essence of this presentation I just needed to explain a lot of how that works so you have millions sometimes even hundreds of millions now coming in like how can you what can you do by simply changing an oracle you essentially now make a lot of more money which allows Landing protocols to potentially customize themselves to become let's say more advantages in certain areas uh one uh potential

solution is uh I'm pretty sure you you all are aware of compound Finance uh compound is essentially uh a landing protocol and they hand out essentially comp tokens as a liquidity mining reward for you to for instance uh lend them your eth and uh from inception to date compound has quote unquote lost or paid out 90 million in liquidation bonuses so 90 million users have been overcharged by 90 million since Inception if you actually essentially got all of that uh back essentially what I've written here uh if

compound recovered the oev and brought comp bought comp tokens at the end of each month since Inception you would have now an additional roughly 700,000 comp the daily issuance of compound is currently 16 1,652 tokens so essentially if we keep that number up they could keep their liquidity mining program which will essentially last maybe for two more years or so running for another one year or so another uh interesting idea is H this I think is my old slide sorry for that but you could essentially take OE from liquidated borrowers and offer it to borrowers that are more reasonable so

what that H what happens is um that if somebody goes DJ let's say 5x long on AA and actually gets liquidated and you now have this new money that you distribute among the other borrowers you drop their costs to potentially near zero and you having the cost at essentially near zero will attract other borrowers and all of this will create a positive flywheel where uh lenders are just essentially making a lot of money uh the third thing that people could potentially do is token Burns uh just showing this at the example of Venus uh also a lending Market on BNB chain uh from inception to date Venus

has lost roughly 70 million to OE if they recovered that every single day and at every at the end of every single day bought their own token xvs uh they could have burned roughly 4, 600,000 tokens uh that is essentially 15% of the total Supply and the current circulating Supply would be down by like roughly 30% so there is various different ways how lending protocols could use that uh money that is now coming in they could make their own token more valuable they could make the markets that they're offering more valuable uh they could keep liquidity mining programs uh running for longer uh and the fourth way that I actually don't

have on the slides here is you could also just give it back to the user that was liquidated uh and drive the cost down immensely so why would you use a liquid why would you use a landing protocol that charges you a 20% penalty if there's one that charges you less than a percent and that is essentially uh the summary of this presentation was a bit quicker uh made some time up from before if you have any questions uh shoot um our feeds are live on like over 25 networks right now and the auction platform is uh set to launch imminently uh it's already operational and liquidations are already being performed

on most of our customers through that it's just not publicly announced [Applause] yet great uh thank you for the lovely presentation we have time for questions so uh we start off here in the front yeah thank you for the talk uh could you explain maybe a bit more so mention that the oracles can technically exploit their privileged position um but they just decided not to do that and my thought is okay if I'm an Oracle and I want to to bundle my Oracle update with a transaction that exploits this uh like marketing efficiency how do I ensure

atomicity of these two transactions because in the end of the day they will be included in a blog by some Builder or by some like validator or whatever you call them and it's up for them eventually to maybe unbundle my transaction and replace my transaction with their own transaction it's not up to me to construct a right yeah but uh theoretically essentially the Oracle can still limit who is allowed to update that specific feed and what action follows it so essentially if you only allow the first action to happen after the second one there's like no way for anyone else to take advantage of that in our system specifically for instance if we sell the rights um the right to update is attached to to a certain uh address that is allowed to do

that so that that person essentially then uh bundles it together with like the Oracle update together they only they are allowed to update and after that follows the liquidation which means that if you essentially try to split the first one like apart like you can't actually do the second one because the first one has to go through for the second one to go through okay yeah we can have a chat after uh just come to my booth great yeah next over here on the left in in your example with the uh the high the fast Highway and the 15-second delay for mantle what is the incentive for the uh the layer 2 in this case to

uh you know to agree to that 15-second delay or is there some incentive it's not something that the layer two decides like the Oracle noes uh together essentially uh just imagine it as two end points one is selling the instantaneous data and one is selling the 30 seconds delayed data similar to some um let's say uh stock API providers some have like this 15 minute delay regulatorily that they they can't sell you real-time data but they can sell you 15minute delayed data and similarly here you can essentially see it the regular eusd feed is maintained with a 15-second delay but you can have instant data whenever you want over the rollup and just pull it into the chain whenever it

is beneficial for you to do so so like it is nothing that the chain essentially can decide but obviously the chains are incentivized to um do this with us because we are essentially uh keeping first of all tvl on their chain the liquidation bonuses are typically just leaving the chain if you pay out let's say 100K as a bonus because you liquidated someone rarely does that stay within the ecosystem you essentially took away tvl from a potential D dab all chains are competing for potential value to climb up again the defi Lama corporate ladder and here the chain is incentivized because money just stays with the dab and essentially the dab is incentivized because they

climb the corporate ladder and so on and so on great uh highly engaged audience we have yeah I was just wondering how do you guys bootstrap the Searcher market for back running the the liquidations that is a very good question I think that is the most challenging part uh with any system it's also what when people typically say every dab can just run their own auctions this is the first problem that they arise to like how do you actually get Searchers and we're in touch with a lot of them uh essentially uh incentivizing them through uh agreements with us uh to get it bootstrapped and then the the idea is that when it grows large enough uh similar to what happens on Main net uh

you will actually have natural competition appearing you already see dabs uh making these sorts of deals on their own uh I believe like angle is a good example they have a private deal with uh I think it was propeller heads so dabs are already essentially pulling in uh Searcher teams themselves and making deals with them so initially to bootstrap we will have uh deals with a lot of Searcher teams uh making sure that there is sufficient liquidations and some of these protocols that we work together with already have established deals that are just going to essentially use this new highway to make use of the privileged position and FL money back to them okay with that uh thank you very much we'll take one minute but uh stay in your seats so than