# What’s the fair price of ETH ? - Anastasia Melachrinos - Speakers: Anastasia Melachrinos - Channel: [ETH Warsaw](https://streameth.org/eth-warsaw) - Date: 2025-10-07 - Duration: 24:30 - Watch: https://streameth.org/watch/yt-t7PBUSOReLY - YouTube: https://www.youtube.com/watch?v=t7PBUSOReLY ## Description A talk on the importance of providing smart contracts with a robust and outlier-resistant reliable price, and how to compute such a price for any cryptocurrency, from very liquid ones to very illiquid ones. πŸ§œπŸ»β€β™€οΈ ETHWarsaw is a series of educational and entertaining events for an active community of blockchain builders, developers and enthusiasts with focus on Ethereum-related tech. Once a year, we organize a large conference and hackathon for the community in the center of the Polish capital with speakers from the best web3 projects and participants from all over the world. Follow ETHWarsaw on social media for the latest updates! X (Twitter): https://twitter.com/ETHWarsaw LinkedIn: https://www.linkedin.com/company/ethwarsaw Telegram chat: https://t.me/joinethwarsaw See you all at our events in Warsaw πŸ™ŒπŸ» ## Transcript hi everyone um so my name is Anastasia I'm from Kao so Kao is a digital assess data provider we provide a lot of data on a lot of cryptocurrencies from a lot of exchanges covering both words CI and Defi and my talk is about what's the fair price of e so this question uh I've been asking myself actually this question while working at Kao for now five years because it's it seems meaningless like that but it's very very important question and when you do have the answer you actually understand that this like having the answer to that can enable you to avoid many types of risks like um useless liquidations for example in D or CI so this is the topic of my talk today and well I chose e because we're at e Baro but I'm not going to talk only about e and uh I'm going to quickly Begin by showing you this chart that basically shows that e is a very liquid asset liquidity uh here is presented through the volume so uh there are dozens of billions of dollars of eth traded across centralized and decentralized exchanges which makes that we can think at first sight that uh there is basically price efficiency market efficiency for this asset however we're going to see that it's not actually the case in terms of numbers of markets uh despite the volume we do have a lot of markets here a market is either liquidity p on on a decentralized exchange or a pair owner centralized exchange as you can see here we have more than 3,400 markets that include e e as a base asset or as a code asset so e again a stable coin a fiat currency or any other cryptocurrency however there's still markets on which e isn't traded at the price uh at a price near what we call the consensus price for example in this case we do see that on binance us um the execution price of the trades that happen on pairs like e usdc and e usdt is pretty far from uh the price on all the other markets we see there and this is a sign of lack of efficiency in E markets even in the most liquid ones because here I really took uh e slusd or stable point and when we include Define there uh so even Unis swap V3 one of the most liquid um decentralized exchanges out there we actually see that it's even worse we do have outliers in execution prices that are are way more far away than these we had on binance us so when we look here indeed there is a lack of efficiency but when we actually look here it's even worse um so this was about eth however the situation is even worse when we look at liquid assets or at least less liquid than e here I took a medium-size one which is Doge Doge against uh usdt on several centralized exchanges which are represented by each color and you can see that despite the outliers anomaly that we have seen in the other charts here so basically some prices going uh away from the consensus we do have another type of anomaly which is the fact that on N exchange we have a price an execution price that follows the exact same Trend as the consensus however there's a lag a huge lag uh and constant lag between this price and the rest uh for now ignore the line in in Black uh but basically if you do the average price of all of that you end up having something in the middle and this is the point of my talk this something in the middle isn't actually the fair price of of Doge here uh and this pattern is actually present in many many markets and doesn't always concern the same Exchange so here I've represented for in a random um time frame the the aggregated price of e against the largest stable coin um so vwap for those that aren familiar it's volume weighted average price and you see that even if it's e it's it picks down and up which doesn't look very normal um and here I represented the exact same chart with kao's aggregated price for the same asset and we see that it's pretty smoother and this is because exactly what I just showed you there are a bunch of FL liers in executed prices and so if you do basically a very simple aggregation like a vwap it doesn't work for crypto for mainly statistical reasons and these reasons is that um is that crypto volumes and prices follow a very different statistical profile than traditional Financial assets vwap works very well for Equity or FX but actually doesn't work at all for crypto assets here you have the distribution of um e prices and you can actually see uh so in oh I didn't put the legend but basically the mean is in um is in green uh no sorry the mean is in in red and the median is um in green and you actually see that uh compared to the actual distribution of prices the the median represents better uh if this distribution than the mean and vwap is based on a mean so you you can be like all right then uh let's instead of a vwap let's do a volume weighted uh median price but it's not that easy because actually if you look at the volume so basically the weight of each EX executed trade within that price distribution because it's volume weighted average price you actually see that we have the exact same problem we have a volume distribution which is which has heavy tails and so in the end we end up if we do an average or or even a median we end up giving too much weight to prices um that have very high volume and that are outliers with respect to the consensus so we provide the solution to that um uh so we don't have it only for Dogecoin or for eth we actually can see that for btcusd even if you compare our solution that is the blue price to a vwap or a volume weighted median which is in Gray you actually see many picks out there however with Geo's uh price and uh crypto tailor methodology you actually see that the price uh is quite normal and actually made uh for these types of assets same for stable coins which are very sensible uh when you uh represent their price with with respect to USD you see the same exact uh type of uh trendings so I didn't only compare our prices with different types of methodologies based on our data you may be like all right this is because uh the data set isn't complete enough and so maybe there are not enough trades to represent correctly the price with a Vivo app but I actually compared it with other sources like coin market cap or coin gecko after and uh this is for a relatively liquid asset uh in of injective protocol and very recently actually two days ago I received this notification on my phone telling me that uh the price of in went down 97% uh this solution is made for retail thankfully um however if this had happened uh on a onchain derivative protocol for someone who was long on on in then it could have led to liquidations um and so see here you see at the exact time uh same time period the price of in against the USD compared to coin market caps uh coin market cap has corrected this this issue but again such use cases are very sensible to realtime data so if you correct it afterwards well it doesn't have really much value and we obviously can see that even for very very liquid assets so Ash is an asset um related to an nft um collection and uh it has only two markets uh on un Swap and uh if you check Coos price historically uh there are many outliers uh the price goes up to $14 against its normal price fair price which is around 1.5 and Kos price doesn't have the same behavior now why outliers matter and why having a reliable fair price actually matters I'm going to show you two use cases onchain derivatives and onchain landing and borrowing so for onchain derivatives how it works is that these are settle on we call the Mark Price um The Mark Price is used to compute p&l on these products but there's also managed liquidations and this Mark Price is computer over the Oracle price and where Kao helps uh generally is either by providing directly this Oracle price but we also are working with oracles like Redstone for example on providing them with our prices that are tailored for crypto assets um and they push these prices on chain for protocols like hyper liquid or other D protocols I took this one because it's on if uh to consume and at least not active liquidations for nothing um I have here a a case that maybe most of you uh know it's a case where um there was a market manipulation for a very liquid asset the mango token um due to the fact that the aggregated price uh was manipulable and so basically we call this an oracle exploit so basically what happened is that someone uh called um Eisenberg he actually had uh 40 plus million uh 400 plus million dollar long position on this asset on Mango markets and so he manipulated the price on spot markets on FTX as sendex and other uh exchanges the price went up um and so he made a profit of uh over100 million out of it because basically the asset was very illiquid but also because the price uh on which this Perpetual fures uh platform relied on uh was not reliable uh another case where uh having a fair market price which is reliable is important is lending protocols so lending protocols like a for example um how they consume a price is basically they take a price of uh some of the assets they're lending or enabl borrowing for and they value the positions so the landing and borrowing positions of every user that uses a platform and so basically a liquidation can happen if the price and so the value of for example The Borrowed asset uh goes down and with respect to what has been lent in dollars um becomes non-sustainable so if the position becomes non sustainable and this is evaluated with the price of the two assets uh in dollars then there is a liquidation happening so imagine if uh well on these platforms you have if but you have a lot other assets um if the price actually uh is based on very bad data or doesn't manage outliers very well then it can lead to liquidations and this already happen do dozens of time in z um so here I've put for those that are uh very technical uh the full paper that describes our methodology so how we actually uh 100% transparently went we ended up producing this very strong and robust uh to Fire's price um and I've summarized here some of its characteristics and I think the most important one is uh Pony it's basically a a price that is a methodology that is based on real data uh there is no parameter that we have used and that can change this methodology works for any crypto asset and is based on their statistics um so it makes it way more robust than other types of methodologies and also it's um it has been trained on data sets during several periods stress periods but also normal periods which is very important for it to actually work at any point of time um however having a fair price of a cryptocurrency is also difficult for another reason um many cryptocurrencies actually are not traded against Fiat currencies and most of the time uh if you run a wallet or if you do uh tax reporting or if you actually provide derivatives on chain or if you actually have a lending and Bing protocol uh you need this prices to be expressed in either a stable coin or a fiat currency and so uh we actually use this methodology and extended it so you can have the price of any cryptocurrency and we cover more than 10 10,000 of them uh in any fiat currency including uh the Polish fiat currency so here I have the direct price so uh of eth against uh polish slotty so we see that given that there are no many direct e uh uh pln markets you have a lot of gaps because there is no trade happening so there's no price uh however if you uh understand this methodology that I um that I explained in what we call um using what we call a liquidity path you end up having actually the price at any point in time for this pair which doesn't isn't that traded directly on the markets um so here's a summary of basically what this uh extension of the methodology is about uh but basically uh if you have e as a and um Poli ly as looking b um you can check where the most liquidity is between uh eth and usdt usdt and something else and create the most liquid path to find the actual Fair uh and the most correct price uh between these two cryptocurrencies uh I mean this cryptocurrency and this fiat currency here's a summaries so what I've said is that um we provide very robust uh prices for a bunch of use cases so our Solutions uh cover a lot of cryptocurrencies and are very robust are made for this we have been working on this for at least five years it went through uh acquiring a company uh a research company that was specialized in quantitative analytics to spending years actually working with oracles like chain link on improving this methodology because we observed all the risks uh we were exposed to all the risks um every day by publishing data since 2018 to chain link and so it was years of efforts that led to these prices which now serve a lot of use cases already including these that you see on the screen and we even go uh even further uh providing prices that are regulated what we call indices um um or benchmarks and so these can be more used if you typically develop uh etps or more institutional use cases um so it's kind of an extension of the other Solutions uh which rely on different methodologies because it's different use cases uh but still can be used for very sensible use cases like you see on the screen that's it thank you very much we have I think think some time for questions yeah all um first of all I would like to say before I ask a question that um like I'm a subscriber to kle reports for a couple of years now and you don't know if you don't know their reports uh you should go to the website like now and subscribe uh because it's the best analy existing on the market currently uh Superior I'm not affiliated with them anyway it's just like pure user pure pure user experience seriously um so without further Ado uh question uh I have two questions uh one question is um so uh establishing for prices is very important and it's a big challenge uh for like centralized exchanges uh but it's even more challenging for um distribut liquidation among cross chain exchanges um in in Defi and well like have you done any research towards that and what do you think uh is like the maybe some kind of Golden Arrow for establishing these prices uh for uh in in the context of amm's uh cross chain very interesting question actually I began my career in crypto five years ago by asking myself this exact question how to actually create fair prices in in a d World which is which is meant to be cross-chain because back then it wasn't really cross-chain but now it is uh I haven't ever finished this research because it's a very difficult question and however I have some ideas on why it's that difficult um and I think the main reason is because every blockchain has its own uh rhythm um so blocks on ethereum are validated and so trades are validated at a very different Rhythm than on l2s which makes total sense because they're not meant for for the same use case but on the two types of platforms we do have imms um there's that but there's also the costs that are very different uh gas fees are very different from one to another and when when we're talking about costs uh like even gas fees are a cost to take in account when you trade and finally Bridges um the fact that currently there is no like real efficiency in bridges uh and that also it costs money to Bridge and so to benefit from Arbitrage opportunities and so to make the market efficient all these barriers I think make it that today there are price discrepancies between blockchains that can be filled in a in a in a certain future uh but that add a big challenge for when you're trying to actually find the fair price of a given cryptocurrency that is traded all over the place all right and a second question um like from what you can reveal uh since like all all the reports are so great like um can you give us maybe a little bit of a of a heads up on like the biggest research we're doing now uh and what we can see in a soon in the future like in in in your in your reports yeah of course um so the biggest research we're preparing now is on Market of use uh so we talked in the in the last panel about Micah uh it's something where I'm based in France uh but we have customers all over Europe uh and it's something that we really really care about uh Regulators care about it but also exchanges care about it because they need to be compliant uh and they need to consider Market of use in many forms wash trading spoofing So currently we're using uh we're researching on uh finding so finding how to replicate uh and identify um traditional Finance type uh Market abuse in crypto um wash trading and spoofing are the easiest to find insider trading is a bit harder but also we're looking at crypto specific types of Market abuse um like anything that happens on chain so yeah this research will be published in October okay yeah um I just wanted first thank you for the presentation it was nice very nice I I wasn't expecting that by the title of the of the lecture and the question is about the representation that you showed about what approach is better for the representation of the value of certain assets so you gave this by the mean and the median and which approach is more uh like proper and uh for establishing the the correct I think it's the next slide or not I don't know yeah this yeah yeah this one yeah with this this green and red uh and I I assume this is kind of the uh your the the vi approach to price establishing or this kind of evaluating of the assets this is the general approach SK approach and this basically shows that neither the mean or the median are enough when you do a price aggregation that is volume weighted so volumes are heavy tailed price are heavy tailed but overall if you see gray is median volume weighted median orange is volume weighted average price so mean and and blue is kind of a mix uh with the methodology paper that I put there uh that is kao's approach so it's really not easy to to do it yeah but if you have to choose between a volume weighted average rice and volume weighted median of course median because it looks way better uh but still it's not perfect thank you