Capturing Oracle Extractable Value Through Specialized OFAs | Ugur Mersinlioglu - API3 DAO

from API 3 da he's going to be talking about uh super super relevant topic super new topic uh Oracle extractable value so yeah let's give uger a warm welcome test test one two perfect uh thanks for having me um the title sounds very intimidating but I Tred to break it down to non-technical Parts as much as possible uh it basically just describes a Oracle system that pays you for using it um so uh I'm uger leading strategic effort efforts at API 3 and uh what is API 3 we are basically just an oracle project uh that do does things slightly different so imagine prize feeds but instead of being operated by intermediaries we go to the data sources and make them operate the Oracle directly so instead of getting data from somebody that gets it from coino we go to coin Gecko and coin gecko runs an oracle directly uh price feeds are a very competitive scene and there isn't a lot of money to be made in price feeds which is why we decided to dive a little bit deeper into a separate topic uh which brings us to OE oev is uh simply a subset of me and similar to me it's a rather Big Industry it describes value that can be extracted due to Oracle updates or the lack of Oracle updates uh sounds complicated but I'll give you a very simple example everyone of you probably knows GMX here derivatives platform so you can 50x long eth and uh the rather unique thing is that you don't take that trade against an order book you take it on the basis of Oracle prices which means on the Oracle price without any slippage you can go 50x long the core issue is that Oracle updates are visible in the me poool so what you can do quite easily is you can look for the Oracle up updates you see one in the men pool you place your 50x long before the Oracle update you wait until after the Oracle update and you close your trade and you profit always and there's nothing that uh the other like the protocol can do to prevent you from this uh on the right side you can see uh a tweet about this from chains side it was a threat explaining this issue on GMX it was going on for 40 days and they were consistently able to essentially take away 40% of the no 10% of the profits of GMX which amounted to like a quarter Bill million during that amount of time so it's a big industry uh how big are we talking uh in this presentation I tried to narrow it down to Landing protocols if you look at the big Landing protocols Ava has already lost well over 100 million to oev compound is slowly closing in and 100 million and Venus on BMB chain also lost roughly over 60 million uh so just to explain what oracles have to do with uh a lending protocol so Oracle transactions on a lending protocol determine the worth of your collateral and your health Factor what that means is you deposit some eth uh that e has some value and you're for instance allowed to borrow 80% against that at one at some point that collateral value might drop with e price going down so your uh loan to um collateral value essentially drops and at some point you become liquidable typically the calculation mean uh in a calculation way this means you have a health Factor when it is above one you're good once it drops below one you're not good anymore and you can be liquidated uh all of this is uh determined by oracles and once you become liquidable as you can see on the right side what typically happens is that people are incentivized to liquidate you by being offered 5 to 10% of your collateral value and that doesn't effect the value that protocols like Ava and com compound are losing so you get liquidated so you lose a bunch of money and on top of that you also pay 5% additionally to somebody liquidating you just to explain the flow a bit so typically for you to get liquidated the first thing that happens is that the Oracle updates onchain and initiates the change that drops your health Factor below one once that happens a bunch of people essentially monitoring the half of positions see uh that you are you have a liquidable position and they often times or most of the times compete for that value in uh block space auctions uh in those auctions they try to get their transaction included first and uh bid a lot of money for that and once that block goes through the user essentially gets liquidated the person liquidating makes profit and they have to pay whatever they bribed so their transaction gets executed first uh so how much are these people that look for your liquidations earn uh there was there's a pretty good uh me book by monoceros that goes into that uh just a quote out of that book for you so liquidations are highly competitive as a strategy and they're simple to perform this means that it's quite common to see liquidations paying up to 99.9% of their revenue to the validator so what does that mean uh quite simply this means on the right side you see one of the biggest liquidations happening in January on the Ava protocol V3 uh somebody got liquidated for roughly 4.2 million uh the reward for that liquidation was 141,000 so just for liquidating the position you would have gotten 141,000 but at below you see the cost to execute this so somebody had the chance to make 141,000 but they bribed somebody else almost equal amount of money so that they are allowed to execute uh liquidate that p uh person and they made $60 in that process so this means Landing protocols like Ava offer you 141,000 and the people actually liquidating you are giving away almost all of it they are fine making $60 on a $4.2 million liquidation but AA is giving them $141,000 and all of this is actually only possible because the or ignores that this value is there theoretically the Oracle project can update and liquidate in the same transaction which means that all of this what happens on the right side won't even happen we could update and liquidate the major issue why a lot of Oracle projects don't do that is it centralizes liquidations if the Oracle project always updates and liquidates other people will stop looking for opportunities which means that if the Oracle project ever fails The Landing protocol will not potentially liquidate that position and incur losses so how do you take advantage of this privileged position while still keeping it somewhat decentralized um or like rather let's talk about the status quo before I talk into that so we we established that the fees are too high AA is paying way too much which is evident by 99% of that being given away and the oracles hold all the power but they didn't execute it thus far because of centralization risks and what we've been working on is essentially A system that simply takes the auctions that happen on the block block Space level and move them to the Oracle level and in simple terms this just means instead of closing your eyes and blindly updating the Oracle without caring what happens on the Chain afterwards you allow people to compete for the updates that have value and pay for them essentially in the same manner depending on the value that is available on chain how we do that is through something called the oev network uh which is a market for or transactions in essence uh it's a ZK rollup that is purely there to auction off Oracle updates and the entire system uh works like this imagine you have Oracle nodes and they maintain a data feed on mantle at uh which is the e price feed for instance at 1% deviation we delay the data by 15 seconds uh and additionally these same Oracle nodes push their data in real time to to what we call the OE Network this is our ZK rollup that essentially uh auctions off realtime data what can happen now is that when people see valuable positions on chain they can go Place bids on this network like searches see somebody that is liquidable at 3,000 e they place a bid at 3,000 e the price actually drops to 3,000 and then the oev network Awards you the win you get a transaction which I'll go into later and you can take this transaction and update the price feed so the price feed is not only updatable by the Oracle nodes they can also sell cryptographically signed data that allows external people to update it and everybody can compete for these updates to dive a little bit more into it so what happens on this ZK rollup uh like I mentioned it's auctions auctions auctions every dab uses a specific data feed and auctions happen for each data feed on each Network that each St uses so essentially like you see here there are different auctions that happen for AAP like init then there are for adap like compound then there are for ad daap like Venus and even if we if all of these were compound uh essentially data feeds since they all happen on different networks or different uh data feeds there are different auctions happening for them like we mentioned real-time data is pushed onto this chain whenever somebody wins an auction and the data that they get is a meta transaction that includes uh a lot of things but I tried to break it down so this transaction has how much did this person bid which is enshrined in it on which blockchain did they win for instance mantle or arbitrum for which data feed did they win so which tab utiliz data feed that there's the O proxy and then uh that's signed data sorry for that uh the all of the Oracle answers that allow them to update the uh the data feed they're only allowed to update the data feed when day in the same transaction transfer the money that is up there which goes directly back to the dab so just to get get back to this example that I've shown there if somebody would have won over the oev network the update rights for that liquidation AA would have essentially been paid back 141,000 of the 141,000 that they've given out and what do you do once you have such an immense new source of Revenue uh you could use that money to essentially build an insurance fund you could put it into your treasury you could distribute it to lenders or to stakers but ultimately uh and this is a bit of a personal opinion obviously every project will potentially integrate this however you want however they want but the money comes from users that are being overcharged for liquidations so the natural thing would be to return that money to the users so instead of of using a lending Market that charges you 10% on a liquidation you could have one that tries through these auctions to recuperate most of what they're overcharging you and you could end up with a lending Market that charges you 0.1% as fees instead of 10 which is uh quite huge and it's my personal belief that Landing protocols that do this create a more efficient market which will attract more people because why would you use one that penalizes you for 10% if there is one that doesn't uh this concept is quite similar to something like me blocker I'm not sure if a lot of you are familiar with uh this picture it is from a product called M blocker it's essentially uh a product where you change your RPC to their RPC they promise not to extract me from your transaction from your transactions they actually execute all potential M that happens through your transaction and give you 90% back and keep 10% uh all you have to do is change your RPC and you're essentially me protected and if there is me you essentially earn it back through them with a small cut for them the oev network is quite similar in that so instead of utilizing an orle that is effectively blind you you read you utilize one that is allowing other people to extract uh or to bid for Oracle updates so you simply integrate a or data feed like eusd you get improved granularity through that integration because people can pull real-time data whenever it matters most instead of just waiting for oracles to update and you get paid in the process and you can achieve all of this uh quite literally and it is very much that simple by reading a new data feed contract that is it all you have to do is read a new contract and that allows auctions to happen for your data feed on this layer two that we have and whenever there are opportunities like a liquidation that you saw on AA the winning party will pay you once they execute that um this basically already my last slide it was quicker than I anticipated uh this is something that you can try today we have bounties uh live for the buildon and we essentially created on the sapoia network a Ava fork with 1,000 positions and we created the fake data feed that every hour will drop in price and you can essentially over the oev network bid for Oracle transactions take them whenever you're able to liquidate someone and once you liquidate that position you get fake oev tokens and the winning team is the team with the most oev tokens at the end of the day so it's different positions with different amount of oev tokens and you can compete for Oracle updates with other people that are competing here and you can win up to $111,000 in bounties for that on the right side is a QR code if you want to scan the challenge and participate and that is pretty much it if you have any questions uh feel free to come up here I've been told to ask the questions into the microphone if not that's also fine thank you very much gur super interesting