Quantification of MEV sharing design OFAs - Danning Sui

foreign [Applause] [Music] my name is Danny I'm a data scientist at CRX and we are a Dax aggregator platform where basically we handle like users intent to trade across decks um and access the am liquidity which is where muv happens across the decks or me order flow and then that's why this is like very interesting um how to say topic for us and this presentation is more of like an open research in collaboration with flashbots and a lot of data are coming from flashbots and also I can find so shout out to them and so for my agenda today I want to share more like about um first of all what's a Meb sharing workflow look like um let me check the chat okay um how does the workflow look like and what are the rows there and what are the values there that we try to identify first of all and then is like how is the Meb Revenue spell it across each row today and we will review some numbers that can potentially have an overlap with what Ellen shared earlier so I'll go quick there and then next is that looking at this funnel of this Revenue spell it where can we potentially quantify and extract the user share back um potentially and we will share two approach here today one is by looking at the Searchers on gym profits the other one is by looking at the validators bit Surplus or sorry this is actually a type of Builder's bit Surplus so um this is a how to say like a diagram of common workflow of OFA today that we learned about um so basically we have this user have an intent behind a dab or a wallet or you can say also RPC provider can be also the entry point and then they want to make a trade that's accessing mm liquidity and then there potentially will be be this OFA or like a Meb share service potentially we can call it Matchmaker or can or solva or resolver and then they will send to a pool of Searchers and the Searchers will try to um bid for this transaction to basically send a bundle with bids and so to win this transaction and if there is no um search or respond with this transaction potentially this service will just fall back it uh back to the mempool so to make sure the execution quality um and then once the amino share of service figure out which is the highest amount which is the highest debate for this transaction um from one of the Searchers they will send on those towards block Builders and then there is this normal flow of block Builders waiting for a block towards the validators through a mini boost so um there's a bunch of like variables or like numbers we need to identify first so first of all there is this now we're talking about how much potentially revenue for user can be from this whole Meb flow and potentially in the future we will see probably a movie share service as a middle layer will take potentially a v as of Bachelor of Revenue and then we will see revenue from searches that kept to themselves and for this part potentially us we can see some of them that's on chain but a lot of them can potentially unknown or off-chain like for example like across centralized exchange index and then next is like how much revenue Builders will keep and then we will see relay but most of them basically as of today there is no revenue for them and then there will be the revenue for validators so to summarize this whole thing into like one table we have see this like table here basically to summarize all the rows and um it's sort of like comparison of like how it was in the past and then right now with I mean boost as present and potential in the future with a new Boost and also OFA sharing design in the future um there you see like more and more people are taking this piece of cake and we will see how we can potentially split further and this is uh on top is a simple invite how to say workflow that's from the order flow auction article for all Frontier research so um one thing very clear here is that basically Searcher now is how to say um segmented or separated from directly accessing wallet or Builders but um through the order flow auction um yeah um so to look at the numbers or basically trying to quantify the scale or total values we want to look at this funnel chart here that I'm trying to propose for framing all this thing um so first of all What's the total Mev here um potentially there is an unknown part that's often we won't see ever um that's particularly that's kept to Searchers and then the visible Parts own chain potentially we can see is what we'll call it like on-chem profits from the revenue for researchers and then after that the the next step in the funnel is how much they given out um to Builder and validators all together we will basically have that data from mme boost proposer payouts rewards uh data and then the next is after the Builder layer there's another another layer of bidding markets which is bidding for Block that the previous one was for beating for Bongo and the next is so uh we don't know for sure how much Builders want to keep or given out but as of today it seems like most of the builders will be competing to argue most of the rewards to validators and potentially there will could be a extra amount that's outside of this whole Meb value that can be coming from Builder as a subsidy um so this is a founder of how this value actually is spill it uh step by step or extract it step by step um so let's look at how big are these bars are today um so quick revisit on some numbers that um uh Elian mentioned earlier and this is also a call out to her slides here which is basically emphasizing again that we will not have a way to pretty much know how much is happening or on centralized exchange or decks um what we can see is how much on decks and uh across decks platforms and centralized to centralized exchange is also another sort of Med um but it this is outside of the scope of our discussion today because we're talking about how much we can potentially give back for Dex users and the centralized change Mev is and evolved with any intent from the Dex users so yeah um next data is coming from eigenfi and this is a pretty nice breakdown of different type of Meb we can see that like on this scale we can see roughly like around um daily of 50k to 100K of Mev orange chain profits that can be attributed to the searcher's addresses and apparently this is definitely will be under indexed uh compared to what's really happening in the reality because um based on all the heuristics um it can only index for example like Atomic um so there there can be a lot more to be on uh to be indexed but next is like so we want to look at how is the stability across Builder and validator and it could be pretty tricky because of different type of behavior patterns of Builders so I think a normal flow as how we are looking at all the data of the builders behavior is that a lot of time Searchers will put um the coinbase transfer which is basically a payout of the Meb address to builders and then Builder as the person who is basically the fee recipient on the Block is like um basically the fee recipient of Gatsby's and then basically Builders can potentially collect all of them but then decide how much of them out of all the total they want to transfer to validators so we will see a transfer usually at the end of the box which is the payouts Builder goes to validators um so uh with this approach with a potential index but there are some sort of like different like Jose um behaviors from some level of the builders for example we could see some zero profit Builders like they just don't take anything and um usually sometimes Searchers can also skip Builders they just put the valuers Via recipient directly in the transfer in the bundle and Builder will also sometimes just put the free recipient of the block um as the validators address so in this case Builder is not even shown in those block production but actually they were actually you know handling the block um there could be also Builders that's actually subsidizing um for example they they can just give the gas fee um set it to um a proposals fee recipient address and even extra transfer at the end of the block um so yeah um and so I I was gonna say basically it's pretty tricky to spill it the um how to say the spell it between Builder and bad leaders so we will basically just look at them together as a whole and look at the scale um so this is like just like generally when we talk about like a daily total of how much proposer Revenue uh look like and it seems like there was some interesting um difference uh compared to end of last year which is the first three months of um mme boost alive and then also the first three months of this year um The Daily was around 600k and now it jumps to um like a 1 million per day um and then I did some cross check with the data on and maybe boost um things like they also have this same scale shown um and then so using another piece of data from I mean inspect where we can clearly see how much is the um transfer out is versus how much they keep um it's also interesting to see like this clear time um timestamp cut off where like in the last uh three months of last year um the spirit was more of like a 70 to the Searcher and 30 to the um proposers uh as a whole and then starting from beginning of this year it totally changed to another like basically shifted towards um much less for a Searchers but much more for Builder and proposal together um so these are some basic numbers um that we know as of today um next question is like okay now as all the application teams and wallets who have the direct access of users order flow um thinking about entering and potentially take more autonomy on user's order flow where and how much like where can we even extract users Kickback um and where how much would that be um so thinking around these questions I think there's a few things to point out um that can be potentially the data that's also accessible that for us to be to be able to quantify um looking at this flow again one thing we can know is that like when the order flow auction take the order information from the application team or the users they will be able to also know maybe how much the user already requests or they want to be shared back um so user can potentially share a request of how much they want to be shared back and deliver that information to the build the bidders or basically Searchers here um and so to see how you know this whole thing play out um so it was one thing they can account is basically um if they are able to see all the engine profits from The Searchers then they can potentially even say hey I'm an exclusive order flow so like it's only how to say um a plus value for you so can I just get like 99 back um it's possible um so we're estimating more like an upper Bond here um and then the next place where we can see actually some sort of number that we know for sure that can be potentially enforced is at the um block bidding Market which is when the block Builders are bidding across competition to get validators get their blogs um and maybe boost actually have this open data set from the relay that they know what's the submission of all the block Builders look like and one thing we one thing we can know is that like basically the first bidder and the second highest bidder the difference we can call it potentially as like a bidding Surplus where if we subtract that part the top Builder is still the highest value validator can access and they will still accept the block and so that's a potential you know a value that we can basically cut and give back to users um so to how to say to fit the ideas into this funnel chart we can basically see like one thing is OFA infra can potentially allow users to request some share um that's visible on chain when they are giving out the order flow information the other is like I mean we share uh infra oh sorry that's actually I mean boost infra can basically cut and return a surplus of the high speed that's above the second highest um so yeah um so now we have these two approach to potentially quantify how much Meb we share can come back for user um first thing let's look at this search your own chain profits data it doesn't really matter the scale here because as we said it can it can potentially be under indexed as well um but when we look at the margin on all these um how to say Arbitrage profits out of the total trading um volume we actually are able to quantify like say how much how profitable per um Mev Mev attacked transaction or me victim transaction can be or basically how much Mev can be extracted on a margin like bips based um per transaction and we can see actually like it's interesting to see approach in general is can be potentially more how to say a higher margin than sandwiching and but all like a median stats but then when we're looking at like about the higher Bond of 95 percentile and 99 percentile there's seems to be coming into the same Trend and these numbers can be like huge variants for how to say for example like the median looks like around its three Pips and 95 tau is around a few hundred Pips and then 99 tile which is basically the upper Bond or like the max number is around like a thousand ish and then if we look at the breakdown of the margin distribution we actually could see like at least clearly half of the trades that were attacked by basically Meb um Bots can potentially produce zero to five which is like we know the median is three and so in fall into this bucket around like Ibis Pips at most um and so Tina asked us to put some number here so I'm just putting some both numbers here um if we think about concretely say now the wallets are actually entering the game and based on their base volume per month um if we look at this like two biggest players um for example if you have like around 200 million volume per month um user can potentially expect a total of like say 100K or based on the five Pips Point data set um and potentially maybe 2 million as like a really huge like upper bound upper bound kind of estimate using the 100 bips that we saw on the 55 95 percentile and if you're a Jose platform has like a 1.2 billion volume per month um then the number can go up to like 600k um or five bips estimate and 12 million or 100 bips like old estimate so um but caveat here is that this is like a totally an upper bound because it's not all the base trading volume will be immediately extractable because like say it's like super small trade of like two dollars and I don't think that can really produce like even five episodes uh Mev value so yeah this is the upper bump um the second approach we want to check is basically the block beating Surplus and on the right side this diagram shows like what that means basically on each slot or like per block all the builders um they have like the the latest uh bid and um we will check like what's the highest debate per across the builders and what's the second highest bit Builder and then the difference between them as we described earlier is considered as like a bidding Surplus basically if you extract any percentage out from the high speed it will still be the high speed and validators had no other choice but to accept it as it's still providing the highest value for them so um this is the methodology we will we will check the fresh freshest bid from each Builder and then take the highest two bids and take the uh I'll say the difference and that's like a estimation of the Surplus where users Kickback and then um caveats is that it's definitely a lower Bound for example um what we could say like why not just take a higher percentage of all the payouts of for the proposer and second of all is like it's probably a very practical and enforceable portion um that can be implemented through any boost um so yeah I'm trying to look at like how much is that bib Surplus look like um so I've encountered some elementation on working on this data set I can it's too big so I have to actually take only uh every time a slice of data of like a day um but then like I checked across time frame and it seems to give like a pretty consistent number here so averagely actually uh we can see like per block the Richer estimate which is basically the Surplus amount between the first and second bit is around 0.003 each which is five dollars as the price of today um and the median can vary how does it vary around like a dollar ish um and then if we look at like 100 Jose a thousand blocks it will be around two e's and then if we just do something paper masks it will be like 25k per day um it seems like not that you know exciting but it's again it's like a lower Bond estimate um another way to look at this is basically okay what's zipper percentage of this Surplus outside of the you know the total amount of the highest bids which keeps the answer around like a three percent ish or like a two percent to five percent um Range uh seems like out of the total bidding amount um so yeah it seems like they're giving us a sort of like a pretty consistent um result um so what do you what what does that mean here so if you want to put some concrete number here if we look at like a DEX trading volume um total flow and say just quantifying by the percentage of the Dex market share you are um say if we look how the total bit Surplus per month basically across all the blogs in that month the total will total up to like 750k and if you own five percent of the Dex market share it will get like 37.5 okay and if you own like 50 it will be having like 300 uh 375k um again it's a lower Bond it's the enforceable part that's coming from the block feeding layer um so putting them all together as a full picture um is um this is what I'm like thinking in my head or to share here um when we looked at that and maybe Revenue split chart again we saw that shifting of the 30 percent for uh proposal to 70 of today and if we just put that five percent of surplus bidding um as a perfect part we kind of like can spill it then further into um the numbers that we can quantify here um you would see that the bidding Surplus as a lower bound estimate for daily will be 25k and then monthly will be like 750 okay and if we're really be really um and then on the Builder proposal Revenue part I put like a red X percent question so that is like the part that we don't know how much we can request but potentially you can put number there um but then the search your own chain profits part is like um through the OFA design we know that users can already specify how much percent they want to give it back so so that's like a green X percent Mark there I put um if we total them all together as the only chain visible mbb Revenue you can see a daily of 1.4 million as a upper Bond um and then a monthly total of 42 million as an upper bound nice number um and then another thing to point out is definitely the iceberg part that's below the water or basically off chain um we will never know how much that will be um I think there are some assimilation um uh potential um sort of how to say service that can potentially run across a centralization centralizing exchange versus Dex and so to tell you potentially how much that is um but I I don't think that data is like going to be very widely accessible and near-term um so I put a question mark there and one thing to keep in mind is that like potentially with this OFA Design coming up this um I'll just say Iceberg percentage versus the one that's above the water can be changed changing as well like shifting up and down um so yeah that's the full picture um of the numbers that I I've studied on this topic um and last notes is thanks to I can find data flashbots data and Tina Elaine Sarah chinyuan and Robert and Tom for the help um this Meme here I just want to show that like I think from the users perspectives like this is our Meb um we want to get a percentage back potentially because we are the value originator um thank you