Michael Egorov - Fireside Chat

baby uh continue with the big brains we have Michael eggerov who is making it all happen um who's going to be joining for a quick fireside chat so are you good excellent excellent now very excited to have you here so I was asking a lot of people for like help with questions um and one of the things that people were really interested in hearing about was just your journey how it was you got to be here today you have a physics background you worked at LinkedIn a new Cipher you spend a lot of time on chain dodging liquidations which I don't think a lot of people know about right how did all of this uh inform like the you know initial design of curve and where it is today well curve started as you remember with the stablecoin uh stablecoin exchange amm right and that was well that demand for for that came really from myself using uh maker Davao at that time so I really used um well didn't didn't really like to sell crypto and crypto was eth so instead I borrowed and all used the funds from borrowing against Eve and of course the big difficulty there was that there was not much liquidity for for dye dye usdc for example and you cash out something like usdc usually so yeah that was a problem and so that was a demand for like why why building curve but also at the same time um they kind of played with uh like trading Bots a little bit so uh really observed how how this Dynamics works and those trading Bots was doing with a friend they were not like not what you traditionally have they were like really really amm inside the trading bot so and uh many many learnings from that inspired both the stable Swap and and actually their the soft liquidations as well so that's uh that's a kind of interesting thing but there was another angle to this to to their demand for for doing for starting curve it was thinking about liquids taken before it was the thing and I'm before before curve I co-founded a new Cipher and a new Cipher has has staking of new token the question was like if you take like wouldn't it be good to to actually create a stake in a state derivative and being able to enter and exit out of out of staking so and that requires amm which is also stable swap so that also was inspiring it quite a bit so that was like wide uh why it really got started and at the same time indeed since 2018 I do use cdps and I had lots of observations there and um like was scared of liquidations all the time well especially with eth in at the end of 2018 when it dropped to to like I don't know 85 or something so um yeah that was that was scary that position had 83 dollar liquidation price so but I instead I instead of repaying I actually bought more Ethan it was the bottom and later on I used that ease for for basically very early funding of of curve when it grew up a little bit right so and that was that was nice but uh scary to have uh you know those uh you know those carry hard liquidations but you actually just suffered your first hard liquidation uh just this week right oh yes well that was that was uh ironically on curve and it was on purpose because you you actually can you actually can change how big is your like range you you put liquidity in the bigger the range is the smaller LTV you can do but the less of a risk of this hard liquidation you have so I've made a small position made it in in the most risky way where it's very like narrow so you it probably would be would be fine if you just want to avoid like next day liquidation if you I don't know go to sleep and the prices drop um and it actually worked it didn't liquidate me at that time but I didn't repay after and when the price was going up it was deliquidating and experiencing losses during that process expectedly and well that was a hard liquidation there but that's fine it's it was I was kind of I wanted to test how how it works well I saw it happening for two different people but like probably should have tested myself so that's what I did it wasn't the giant stash it was just a tiny position um yeah yeah about relatively yeah so that wasn't like a super big stash but yeah well Liquidator got quite quite nice profit and I would say loss from Liquidation percent wise was lower than in other protocols so it worked all nice and that's something people don't actually understand a lot is that it's oftentimes on the price going up that you can get liquidated into the trading fees not due to the price true true uh well the thing is you like as you go through this liquidity which you put in for like collateral so you put cholesterol as liquidity right and this liquidity gets used when you have reversible liquidations and deliquidations and every time it gets used you do experience losses so and uh you kind of if price suddenly jumps up the losses are just as high as if it suddenly jumps down only if you've been in this soft liquidation mode so if you were not in soft liquidation mode doesn't really matter how quickly the price changes the price of collateral but if you were in soft liquidation mode so you already kind of presumably in a bad situation but you better keep an eye on your like health ratio and repay the position a little bit so that that helps you to avoid hard liquidation or maybe you can you want to repay it entirely if you if you got into into the price you didn't expect that's also also something and we're workshopping a better name than soft liquidation by the way so if anyone has ideas you're not actually being liquidated so it's a bit misleading um but um it is the case that um that when you're in like in that soft liquidation mode you can get like right up to the line it seems like and you chose to design curve USD to be over collateralized yes um but do you think that you could have actually like dropped that line even closer to make it more Capital uh well not really because like if you if you put it right at the border right at the border of this uh range then your LTV which is which you are allowed to do is basically based on the value of um of stablecoin you would get if the price slowly goes down to the bottom of this range which will be uh something well still less than one let's say if the range size is I don't know 10 then this would bring you uh well 100 minus 10 over over two uh 95 LTV but you also need to leave some some room because you you actually don't want to to get liquid hard liquidated right away like you don't want to a tiny loss to cause hard liquidation so you uh we usually we add a room of three percent to that so then um you you are basically down to like 90 percent uh or 89 maximum LTV so that's what we have and for different assets it probably should be different like if something more volatile should be probably lower than that so you don't want the system to to like to be with bad positions which you which somebody could liquidate better loss to the Liquidator that who would do that right and the moment it's about 9x leverage that means you can do yeah I think it's a 9x maximum and you've talked also in the past about wanting to introduce like curve LP tokens perhaps later as a collateral but like much reduced and we've seen some presentations on endogenous collateralization for example is that what's going through your mind with that uh yeah well uh LP tokens were I thought of were three crypto LP tokens mostly so let's say uh like usdt or eth BTC or usdc esbtc so then it would be backed by by this volatile collateral collateral I would say but it's less volatile than eth or BTC so LTV probably could be a little bit higher so that's uh but yeah that's this is still um still something external to the system so it's still still like crypto or well cryptofi ad essential but curve eth might be more risky curvies yeah one needs to think about it it's I think it's still not getting into into this territory of um of being like um endogenous or external how it's called yeah so it's still uh it's a little bit different if if you imagine one curve is backed by one curve is it's sorry one curve USD based backed by one curve USD it's actually cancel it actually cancels out so you uh well you kind of change nothing if you deposited one curve USD as collateral and borrowed one curve USD against that but nobody would do that of course but you uh that's a nice model to think of if you have a pool which contains curve USD but still it needs to be done really carefully so I didn't really think too deeply about about pools which contain curve USD as collateral so that's um but yeah well I think after good thinking it's possible and the thinking is exactly this that curve USD kind of cancels out I think the big Keepers do kind of that stuff because they back when you have pack Keepers getting funds they back essentially some curved USD implicitly by LP tokens which contain curve USD and it's exactly like that curve USD cancels out right right and it does seem like the peg keeper mechanism which is like very Innovative it seems like that a lot of the construction of curve USD was designed with this idea of keeping the peg as tight as absolutely absolutely yes and in doing so like you know I've been running some basic simulations and it's um you know still early days and then I'm often usually looking like coin gecko data which is not ideal but it's still showing like as tight of a peg as a lot of the one-to-one backed redeemable stable coins yes yes so then the trade-off there I guess would be the borrow rate is right uh well uh yeah borrow rate depends on how much you have in pack Keepers it incentivizes to have thing but not much if back Keepers get too much borrow rate gets really low like below one percent which we're so which we've seen and if uh back Keepers get empty right now they are the borrow rate gets to uh some preset value which is for stake is collateral is uh 10 which is a lot so this instant and the reason is that if um from this point when pack keepers are empty here we use this starts going down we need some people to to buy CRV USD to essentially repay the loans because then CRV USD will go back up and uh my original thinking was that this mechanism would be slow because that's what we've seen with maker Dao it was maybe taken weeks to uh to be like um uh for for their rates to affect the peg for some reason uh in curve USD it's acting much faster so I've seen like it was maybe maybe a few hours after which uh the CRV used the price started being significantly changed because of rate going up and rate went up exactly for that reason to to actually try to try to stabilize CRV USD when pack keepers are empty and yeah just works a little better than I thought which is a surprise so maybe that's just because there are people who actually actively look at borrow rates and maybe Arbitrage borrow rates so those people who Arbitrage borrow rates between different platforms they jump in and out depending on the borrow rate and price very quickly reacts on that so maybe maybe that helps a lot and it's kind of low risk thing to do for them because the price of CRV USD is actually pretty close to one yeah and I haven't got a chance to publish it yet but it does look like the incentives are working in terms of you see the rate of change of tvl just whipsaw depending with the changes in borrow rates yeah yeah it's pretty exciting to see that system actually works as it as intended yeah I mean the whole design is like really relying on user incentives like I don't think many people know that they themselves can call the peg Keepers to update yeah make money in the process right well yeah not much money though because there is a competition already well uh so I think usually it gets called just well when their profit is just just above the gas spent that's what I think Arbitrage Traders do and Arbitrage Traders are really responsible for all the trades to keep it in Peg yeah absolutely so it's a what else have you learned from kind of deploying this year you're famous for testing in prod so you you know you test it a few times and had to relaunch based on what you learned how that's been tested in prod for a few months what have you learned that surprised you um well I think the biggest surprise was this that or that in rate incentives work uh much better than I thought so that's uh that's a good one I don't know if it will be like that uh when it scales up well another surprise was was that rates actually Jump Around uh very quickly so that's not something I necessarily expected but that makes sense so volatility of stablecoin price which would presumably come from volatility of collateral actually gets absorbed in the volatility of of the borrow rate is there thought that the volatility might decrease is like the overall scale increases possibly yes um yeah and I think the thing is that if if demand for for stablecoin increases then it will start uh well if if there are no borrowers who generated this stable coin it will start feeling that from pack Keepers and when that happens then a rate goes down and then there will be more borrowers so the rate will equilibrate somewhere and um yeah and I think right now it jumps around because um the well uh it's the borrowing demand sometimes becomes pretty high so well this week well no this not this week was it um yeah so the uh at some point there were like um I don't know like weekends right and on on weekends not many people borrowed but at the same time uh demand for CRV USD uh went uh went up um maybe after after Thursday so incentives changed right and then uh after that um people saw low power rate and on Monday for some reason they many borrowers start borrowing on Mondays so they they borrowed really a lot and it was comparable to their uh to how how much is in penkeepers so it's a really uh changed borrow rate a lot so there are some this uh you know kind of artificial limit time uh time limits which like you know what people do on weekends and what what they don't which apparently affects affects the borrow rate but it changes like really really quickly I would say so we're still dependent on tradify a little bit yeah yeah we are dependent on tread file a little bit I think or maybe on just people's behavior really and but I think as soon as the amount which people will probably borrow in one day will become smaller than I don't know five percent of total borrowed for example or then maybe the rates would be a little bit more stable yeah so that's what I think yeah let's talk a bit more about the scaling uh so you've mentioned that by the end of the year you're hoping if Market dependent will get to a billion but we'll see right hoping yeah we can't predict next week we cannot predict at all but what happens if it gets bigger than that what if it gets bigger than tether usdc with the pig keepers still be able to function in USD curviest was dominant yeah I think at the moment packkeepers assume that the amount of amount of redeemable stable coins impact Keepers is uh maybe maybe around five percent so so maybe we would have a problem if it becomes 20 times bigger than tether which I I doubt we will see that in the nearest Future at all but yeah we'll see it's just a very hypothetical question yeah of course of course of course um anything else on Kirby SD while we're on the subject because I'd like to talk also about like try crypto NG and some of that but any final thoughts I'm sure we'll get user or audience questions on it too um yeah I think that those were good questions so okay that's good to hear good to hear um so let's uh Circle back to you know back in the V1 days and a lot of the inspiration for the V1 algorithm was just the slippage question right it was very tough with the kind of classic uh x times y equals K formula to be able to do large stable coin trading that's right it just wasn't optimized so you went to the drawing board and you came up with the V1 algorithm right and then from there you said well let's also rebalance the liquidity internally and came up with the V2 pools and that's been going fairly well um but uh at the moment like that's not yet flipping uh like the uniswap V3 right right because well I mean it's something to rebalance uh yeah yeah it's uh and it probably shouldn't because you um if you concentrate liquidity you necessarily increase impermanent loss and as soon as you rebalance your impermanent loss becomes permanent loss so and let's say if you squeeze liquidity to let's say a one percent range you maybe amplify your impermanent loss by factor of I don't remember 200 I guess right so uh and you hope to get compensated for that by by fees but actually if you keep just concentrated liquidity you you don't you you actually need some of liquidity being concentrated and some of liquidity being non-concentrated maybe you need to you can do some smooth transition between concentrated and non-concentrated liquidity and that's what we do but in very briefly you have a combination of concentrated and non-concentrated liquidity and if the price is uh well not far from what it was yesterday right then concentrated liquidity is working but if price jumps and you know somewhere far away then maybe you you have a non-concentrated liquidity working and if you had no unconcentrated liquidity then if price jumps away then you you don't earn fees anymore because you don't have any liquidity and if you didn't earn enough fees in the past when you've been in concentrated liquidity then you just cannot move to the near concentrated liquidity to the new region without a loss so and the design Choice was to uh to basically keep both concentrated and non-concentrated liquidity and the fraction of concentrated liquidity is actually not very large at all so but nevertheless it's enough to to keep low slippage within like I don't know one percent from the um from the current price or something but you know if volatility hits then you sometimes in concentrated liquidity sometimes not but always making fees to to move concentrated liquidity there right right and it still blows my mind that the V2 pools have their own internal Oracle just based on the you know understanding the trading behavior and finding the price Oracle and price scale can you talk a bit more about that mechanism for those who aren't familiar with it right right so there are two prices there one is a normal price Oracle let's say exponential moving average right but there is another which we call Price scale which is somewhat mysterious right but actually what price scale is is it's just the price where the concentrated liquidity sits so when price price scale changes it just means that concentrated liquidity moved and that is the price at which you have lowest slippage and ideally you want your price scale to follow the price Oracle as quickly as possible but also if well there is a problem though if you make your concentrated liquidity um huge then you actually it will be too expensive to move it around so you will not earn enough money to move it without losses and it will just be stuck somewhere if you if it is too large if the concentrated part is too small then of course you can move it around very cheaply and you can always make earn enough funds to to move it that's fine but then it will be not really helpful because it's if it's too tiny so and crypto pools try to find a balance in between which which is sort of optimal between the two but on unit swap three if we go back to that you have you have like only concentrated liquidity so it's up to you how you manage it in principle probably it would be possible to implement Implement curve V2 on top of unit swap 3. Omega unit swap for but my feeling is that if you do that the swaps would be an extremely super expensive when it comes to to rebalances and tracking all the information which is needed because like if you if you do it like really on on the low level in the on on whatever smart contract language then I think it will be it will be always cheaper than if you go over over the abstraction layers so um let's also talk about something that sits a little bit outside stable coins which is uh the ve CRV mechanism which you know back in its day everyone laughed at it and then everyone adopted it um but back then you're building it off of like an or I think Viper 2 um brownie was still in the nation state and there's been a lot of improvements to Viper that have allowed for some of the more recent Renovations so if you were rebuilding ve curve today is there any changes you'd make well I mean of course there could be uh really like switching to new compilers of course helps yes but maybe some code refactoring also wouldn't wouldn't really hurt uh well what would be well I think I think there is some Innovation actually happening in the space so some different projects tried different things um well there was even idea to do like the nft uh kind of uh position well I don't know if it's a good idea or not but um well that's an interesting experiment nevertheless um so and um but yeah I think it's probably would be fairly similar to what it is today I think that's because I tried to wait to make it as minimal as possible so that not much changes are needed right right so I got a lot of questions about some like other protocols and your thoughts on them so uh first up um go is launching their stable coin and there's been a lot of questions just like your take on go and what's going to do for the other ecosystem oh um yeah I mean that's a reasonable step given that they uh they had liquidations covered they kind of they should have them mastered right they they have all like sorts of collaterals and systems for like Risk estimation and onboarding of boards and collaterals so it's a very logical step for them to do a CDP and in principle in principle single coins they they give all the all the interest earned to the system so that's that's a nice income for the Dao I guess that makes sense what about a balancer people are oftentimes using balancer for like their boosted pools or they're like imbalanced 80 20 pools is there any thought that curve might start launching similar type features atop their pools Oh you mean um the pools which have uh non-equal balance right right right yeah it makes sense uh actually um actually like we we usually don't try to copy uh the like uh inefficient uh things like uh constant product is not Capital efficient although it's very fundamental and about what balancer does with those pools they are also um well not very Capital efficient but they're good as a reserve kind of thing or like rebalancing portfolio but one thing I've seen uh very interesting which I couldn't really come up with myself is what a wire earn is doing with Y8 they actually found how to make stable swap keeping not 50 50 mixture but any mixture so essentially what balancer does for constant product they managed to do to stable Swap and that's actually a very interesting approach so probably probably doing something like that would be curious interesting just a couple of quick questions before we open it up to the audience mostly on the subject of decentralization so I don't think actually many people know this but curve is extremely decentralized to the point where you've never even met most of the core team is that right that's right yeah it's also for developers it's like a developer's Paradise because there's no PMS no meetings because everyone's just coding right yeah that's right um so is your thought that you'd like to kind of like follow the vitalik approach and like scale back and let the community take it over more because 99 of the Innovation is coming from your brain yeah yeah so probably that would I probably still have some some things to to bring um well I mean not uh well just some ideas so it's basically when you build something it's a uh you you almost immediately think about something else to uh to create which is not necessarily always huge but requires some some effort and it's hard to resist to not do it so interesting interesting do you think more companies are and like within web 3 even this kind of developer-centric uh structure of a company is still fairly rare do you think that like but it's working for curve because you have about a dozen people managing billions of dollars in assets um just through writing code alone like do you think that's going to become more the model for successful web3 companies possibly possibly um but I think it's just whatever works for for different companies yeah makes sense and then I guess finally before you open the audience like just what are your predictions for the D5 cryptocurrency uh over the next three years to three centuries which is roughly the emission schedule for curve okay well I guess for for three years probably would continue as more or less as it goes uh well interesting to see that Regulators started noticing crypto um well maybe maybe that's good maybe that's bad who knows but they they do notice and because they feel it's that it's probably the future and they of course try to regulate and uh well who knows maybe maybe in natural ways of self-regulating are are good but I think that this kind of Regulation versus self-regulation and how it will interact with traditional systems it is probably something we will see in in some coming years but as for distant future I do think that this is the future of the world economy so I think that well everything will will be defy at some point I agree and on that subject to regulation and you've basically said smart regulation would be welcome the problem is a lot of it's not very smart right yeah yeah yeah but but you know many many Regulators are many Regulators are non-malicious maybe they they are just not very um not always very educated but uh they have sometimes right goals so and sometimes Community has the same goals so um well ideally they probably should watch what communities do and learn and yeah but being in Switzerland the laws there are much more clear and have allowed you more of a framework to actually launch a lot of this right oh yeah yeah that's uh that's part of the reasons why I'm there it's uh well still definitely far from uh far from being all clear but they seem to be moving in the right direction that's good excellent well thank you so much for your time like every time you do an interview I just learned so much so it's such an honor for me to get the chance to talk with you like this let's open it to questions so they're uh see right there all right ah yeah I think I think they are learning at the moment so it's pretty much research and uh like whether they will launch at least this way soon well remains to be seen but what's uh what I find interesting there are two kinds of cbdc's one is something you see in China which is which to me feels like not like cbdc it feels like some state-owned PayPal essentially and uh oh well many pilots on hyper Ledger seem to be that way as well but also there is something they are doing on ethereum which is obviously happening with Mariana while they use evm code so um and that's that means that that this is another kind of stream in in cbdcs but also this uh Mariana project seems to be focused on uh on large organizations Banks um like basically big guys so it's not doesn't seem to be uh something uh like consumer oriented so and that also makes sense and kind of uh kind of hopefully uh hopefully alleviate some cbdc concerns people may have like about surveillance and whatever this is not something uh I I think this is not that kind of cbdc over there but maybe we'll have two types who knows so it's uh they are still exploring like what to do and I think they all started with cbdc's like PayPal and something running in in like super super confined environments not very decentralized and this project Mariana what is uh what I think is interesting there is that they seem to to be running on on ethereum they deploy an ethereum test net so uh maybe that means that they are they become more receptive to using uh to using something like public blockchains hopefully I don't know like don't have any hard indications about that but uh that that's what it seems given that Swiss stake owns the intellectual property for the curve V2 algorithms would they be obligated to at least compensate you for this use um well I guess but that remains to be uh to be seen they they are doing just research at the moment it's not really uh it's not really used right now but they are just trying to see like how how applicable applicable would be different uh Primitives to uh to their goals but well I mean something is possible but in in which way I think it's it remains to be seen once they once they get to probably next steps and when when those happen I I don't know because it's like really their research essentially right right of course so yeah obviously there are a lot of stable coins different models of those like they are feedback decentralized even in decentralized World there are different approaches to how you can build it more Capital efficient less Capital efficient more decentralized yield generating etc etc do you think the landscape will be that there will be a specific Market fit for each of those types or it will be just a world there is one stable coin or like one third-backed one decentralized what's your view on that developer right well I mean curve initially was formed on the idea of having multiple stable coins so I I would hope there would be many stable coins really could you say a little bit more about what you think it'll take for stable coins to get to hundreds of billions or trillions of dollars of market cap and maybe talk about it in a phased approach right so like if that's the medium term aspiration how do we practically get there what types of innovation do we need to see what type of Primitives need to be built to support that like is that a how do we get to that goal yeah very good question I think what we really need is uh exchanges to expand exchanges for stable coins to expand from just crypto Market to to Foreign Exchange markets and for that we need stable coins of multiple denominations we actually do have that we have several Euro Euro stable coins so that's good but also uh also it means that liquidity should be bootstrapped and question how to bootstrap liquidity for foreign exchange is a very good question because like you for foreign exchange to work really efficiently you need to have access to a really good foreign exchange markets for Arbitrage trades so Arbitrage is actually bootstrapping markets crypto Arbitrage is very efficient even between decentralized and centralized exchanges foreign exchange Arbitrage would mean access to those same markets which which banks have access to and I don't think uh I don't think biggest Arbitrage well biggest Market making firms have uh well at least connect those markets to crypto stablecoin so and that's I think something which needs to happen for stable coins to scale further yeah I I have the mic so uh yeah I recently tried CRV USD I'm a big fan already uh but like what I've noticed a lot of stable coins they still go to this like Fiat packed Paradigm uh which I don't know me personally I find a little bit uh maybe sub-optimal because of you know recent like Bank collapses and stuff and like how unstable the Global Financial system is uh I was wondering what do you think about uh unpacked stable coins like uh Rye high and could curve be exploring that direction well I that's a interesting question I think uh it it makes sense in principle uh but I don't think it has much utility because you cannot really um cannot really redeem those or pack those to to something you expect to to stay stable maybe something like a stable coin pack to Gold make sense in principle but I think still not very easy to bootstrap the the demand for that so but on the other hand if you borrow USD against crypto you are actually shorting USD so you are betting against uh against their I don't know against the stability of financial systems so if you can imagine that USD goes to um like to zero because uh I don't know because of some Econo because the U.S economy gets mismanaged and whatever then that's good because you will have to repay less all right so you can think that way if that makes you more comfortable I guess hi um I have a question because we talked about cbdc's we talked about decentralized double coins etc etc um but we are seeing Regulators coming in more and more and so if they recognize stable coins it basically means that any private company can issue stable coins and we're seeing a lot of interest as well from actors like BlackRock Fidelity whatever for the crypto sector and so my question is do you think that we can see stable coins not like cbdc like but something close to it which could be issued by these actors and do you think that is something that could be integrated within um the current markets easier than the the cbdc's yeah I think I think absolutely I think uh also well I've heard that these are called deposit tokens right so tokenized deposits and they um I think they make a lot of sense because you have risks of essentially one company in one token so if you if you are into combining many of those into Diversified Diversified baskets then they actually make a lot of sense I'm not sure if they make a lot of sense for like end user but they make a lot of sense for these Diversified portfolios and making like resilient like other resilience stable coins so that's what I think all right so I'm sure there's a ton more questions the good news is Mitch is here in the premises so feel free to ask him afterwards but we're going to move along to our next panel which is going to be hosted by uh Christoph orbansky of l2b everyone thank Michael for throwing this whole thing