Michael Egorov on Curve, Yield Basis and DeFi

Transcript

All right, everybody. Welcome to the Fully Vested podcast. I'm your co-host E here with Patrick Scott. On this show, we explore everything in DeFi, the world on chain, and where it's going off. Uh and on this episode, we have on a very special guest, someone who's been building in the space for a long, long time. We have on Michael of Yearn Basis and Curve. And in this episode, me and Patrick, you'll notice we really just sat back and listened. I felt like I was back in school, I was back in college, and we just and the professor brought someone in who had experience in the real world. Uh and he was just sharing all his stories and all his knowledge.

And for all the OGs, I think they'll really enjoy this one. What do you think Patrick? >> I I thought it was pretty surreal. I mean, I was I was a a pleb back in 2020, 2021. Um Yeah, and and so to actually hear some of the inside stories behind what was actually going on behind the scenes, uh how Curve came about in the first place, uh it was pretty awesome. Uh and and I I learned a lot. >> Yeah. >> I knew that it was math-intensive. I didn't realize um quite how intensive the things uh were behind the scenes.

>> Yeah, uh and just hearing about his thought process with everything. And keep in mind, he's a physicist, and he was working in labs doing major post-doctoral research. So, and just brought all that knowledge and experience to DeFi and building these types of systems on chain for all of us to enjoy and play around with. Uh and for him, it's actually a big deal because he lives on chain. Like, that's how he's been living his life for years on end. And he's one and we said this multiple times in the pod, he builds products for himself. And once he built Curve for himself, he's it's like, "Oh, this actually may have use for other people if I built it for myself." And so, I you know, I really respect builders that build with that kind of framework and mindset.

>> It's the best way to build products. >> Yeah, and hands down the best way to build. Yeah, and uh so before we get into it, make sure if you're listening, watching, wherever you are are, make sure you hit that like button, hit that heart, reply, comment, share, repost. Uh make sure you follow us on all platforms, X, Twitter, uh YouTube, Spotify, Apple, wherever you are. Uh and this is a Snapchat. >> [laughter] >> Fully vested Snapchat. It's soon. [gasps] Uh uh but and you know, this is a DeFi Llama podcast then. Uh if you're interested in going dive into the world on chain and need a little bit of help, we have uh Llama AI and we have a special discount we have a special

referral code below uh DeFi. So, make sure you go to the description or reply below and click that link to sign up for uh the Llama AI. Uh and so, yeah, we're about [clears throat] to get into it. Uh here we go. >> [music] >> Welcome back to the Fully Vested podcast by DeFi Llama. I'm one of your co-hosts, Patrick. >> Here with DeFi Dave. And today we have a special guest, Michael, the founder of Curve and Yield Basis. Welcome to the show, Michael. >> Hi.

>> Um it's great to have you. I think I've been you know, I think Curve is probably one of the protocols I've used the most over the years for swapping. So, it's really cool to actually be able to talk uh But yeah, and what are you >> truly a foundation of DeFi in the world on chain. I think anybody remembers back in 2019 and 2020, Curve really being the foundation for making a lot of things happening and pioneering a lot of things. And the fact that you've been building since then, not just Curve, but Yield Basis, uh really excited to get in to all this with you today.

>> Cool. Um but before we get into that, uh maybe you could share a bit more about yourself, introduce yourself to to people, let them know what you've been working on recently. >> Um sure. Well, I mean you did the very short intro of myself as founder of Curve and your basis. And well maybe most people know by now that I'm physicist by education essentially. And um Um, yeah, I uh kind of got uh a little bit into crypto in I think at the end of 2013.

Obviously with Bitcoin like And of course I saw all sorts of how they were called shitcoins which are now uh I would say called mostly meme coins, right? Well, I didn't buy them, but you could mine them and back in the days and uh Yeah, and swap for Bitcoin. So, yeah. Uh anyway >> I have a question. What initially got you into Bitcoin and really interested in it? >> Yeah, well, I I just started looking at it when I still did my postdoc in Australia at Monash University. And I've noticed that

well, one of the one of the PhD students told me that he traveled to Europe and he actually took Bitcoin instead of instead of cash or whatever as travel money. And he he was able to essentially swap them to euros in some Bitcoin ATMs which already existed in Europe and it actually worked. So, I started looking at it and uh Yeah, um I don't know. Even mined some some altcoins on some essentially a video card.

And yeah, I don't know. It turns out Oh, it's just on a single video card, so it turned out to be one Bitcoin in the end, which was not very expensive in the day. So, well, anyway, it was more than the video card was worth, and I was sold the video card anyway, so it was fun. And later I did come across a use case for myself where I had to use Bitcoin to transfer money across the borders. I mean, I could avoid that, but banks were giving me hard time with reversing transaction for no real reason when I had to reattempt it,

losing some losing some funds on reversed transaction, plus the bank wasn't really some bank with like where you could avoid going to the branch, so it was like Um, yeah, it wasn't very convenient, and then I thought, "Okay, well, I need to do something else. Maybe maybe I should use PayPal to to transfer funds." And it kind of worked once or twice, and then it got stopped working, and I thought, "Well okay it's maybe not working for this purpose.

Uh Um, well, why don't I try Bitcoin? Like I can buy it here." Well, I was in Australia at the time, and I could sell it I don't know, on the other side, and then it's like you know, like like transfer of money, and it actually worked. And um Well, that's what that was my use case, I would say. But obviously I thought, "Well, okay, maybe Bitcoin is actually pretty valuable if it can do that." So, I probably should keep some.

So, in 2014 I actually went to to work in the US, and I did think about starting some startup and not like not necessarily thinking about crypto. Um but that that that was the time when I first learned about crypto. And uh well, when I came there, it was pretty fun actually. I went to uh well, it was uh in uh Silicon Valley. So, I went to Hacker Dojo in Mountain View, which doesn't exist today, but as I am learning, many many many founders were just in the same Hacker Dojo like uh founders of um I don't know, Dfinity,

uh of uh Cosmos, of Filecoin maybe, or like uh uh all sorts of all sorts of stuff. And uh I did found the NuCypher with uh my co-founder MacLane back in the days as well. Um and that's when I switched to well, not not to fully crypto because like it wasn't about crypto first, but then it went more towards crypto. And um well, um I guess it was good.

And um well, later in um well, I would say sort of much later in at the end of 2019 actually, I already was in the process of moving to Switzerland. And I did think to start a company in Switzerland, which was first something I thought about liquid staking essentially cuz uh well, when in NuCypher you uh there was some interesting staking mechanics, which was not so interesting when you want to exit the position, and liquid staking was kind of a natural answer. And I thought, well, maybe it also makes sense in upcoming Ethereum staking, right? Uh but then I thought,

well, how do you market make for that? And then um I came across um a use case for Well, I mean, uh algorithm for which we which is Curve's stable swap. But at the same time, since 2018, I guess, I was Well, I mean, I was crypto maxi, so I didn't want to sell my precious coins, right? And I was using Maker DAO for borrowing um DAI against Ether. And well, when you borrow I borrowed DAI just to not spend Ether, so I had to redeem it to some dollars. So, I had to swap it to USDC, for example, and redeem USDC on Coinbase.

And uh well, I didn't like uh how liquidity looked like. Uh well, it was all right, but it seemed like it definitely can be better, and I thought, well, why don't I just use that idea with uh liquid staking AMM? Why don't I apply to stable coins, which is my uh well, my which is my need? And well, maybe liquid staking is uh going to materialize sometime. Nobody was even staking uh Ethereum at the time. Um but maybe uh maybe I can fulfill my immediate need for swapping stable coins.

So, I launched it in the beginning of 2020, and we Well, it started with stable coin to stable coin swaps, and then actually doing liquid staking uh swaps as well. Uh this is what we now know as Curve. Um and of course, it went beyond these algorithms over the years, and we are as it's happening with Yield Basis, we are rediscovering the power of uh algorithms which uh which were created within Curve later in the years and maybe we will maybe we will rediscover some more.

>> Yeah, what was the origin? >> Oh, you go ahead, Patrick. >> Uh, what was the origin of those algorithms in the first place? >> Mhm. You mean stable swap or others? >> Um, stable swap or, uh, liquid staking swap, whichever comes >> Well, oh, okay, okay. Yeah, well, I mean, I was, as I said, I was first thinking about liquid staking and then, uh, um well um, stablecoins and I I learned about AMMs at the time and I thought, okay, so we have constant product AMM, which is, um, which is obviously having pretty high slippage and we could, hypothetically, have a constant sum AMM, which is just having constant price, but then, uh, if

the price changes by a little bit, you're out of liquidity. So, what how do you do something in between? And, uh, well, thinking about that, I came up with stable swap algorithm. >> Yeah, that's incredible foresight that you saw liquid staking becoming a thing. You're like, okay, Ethereum that's going to turn into proof of stake, uh, I'm going to have want to figure out a way to redeem out of it. And then, what sticks out to me is, uh, you being a user of DeFi early on, like you said, with MakerDAO, with Dai, you're like, oh, how can I switch out of Dai to live my life? Um, and that's, you're like, oh, and then you come up with something like Curve. Um, I wanted to ask, too, uh,

but that we talked about the, kind of the mechanics of Curve, but I want to talk about the DAO aspect of Curve and how that came about and how V came about. Um, why did you decide to go that direction? Like, how did that >> Yeah, yeah, yeah. So, in, uh, I started working on the DAO in May 2020, I think. And it was pretty clear to me that one coin, one vote is pretty dumb system. So, need something better.

Um and but it was clear that some decentralization is needed here because I was actually well, it was the time when um well, SEC looked pretty aggressive. They always look pretty aggressive except except for today, right? So, today they don't seem aggressive, but at the time they did look aggressive. So, I thought, well, maybe um maybe we need well, some decentralization is the right response to that, right?

And I also read um an interesting report by FINMA, a Swiss uh regulator. They've been looking at um well, I think at Libra, but they really went much much broader in their thinking and they had a very kind of big comprehensive response or I would say their thoughts expressed about what's what like what can happen, what can work in crypto and what not and when it's decentralized, when not decentralized.

And I read it and I thought, well, actually that sort of makes sense and it did point me to like to the need of decentralization. Um well, ironically, regulators can not only be bad, but they also can be good and they can point to the right things. Um who would have guessed, right? >> Who would have thought? >> Yeah. Uh so anyway um but obviously, if the system is fully autonomous, you just don't cannot have so many controls. You really, uh, organize people who are driven totally by, uh, by their economic interests, right?

So, pure game theory. So, um, well, how do you how do you do that, right? And, um, well, I had some example in front of my face, I would say. Um, uh, just before leaving the US, I lived, uh, near Seattle, where Boeing is located. And there was quite a lot of drama with, uh, Boeing Max Max aircrafts. So, you probably, uh, you probably heard that they've been, uh, like, crashing and whatever, and, uh, that was really, well, I truly was curious what was the reason for that.

And, well, on the surface, it was the the reason was that, the company was spending a lot of money not on improving the quality, not investing into production, but actually into buybacks of their stocks, right? And after, well, certain amount of buybacks, um, some shareholders seemingly exited. And then, it appeared that, uh, the technology is under invested, and it there caused some flaws, and, uh, all the crashes, and stocks obviously went down. And, well, after they went down, then, um, seemingly government bailed them,

and uh whatever. But, what was the the the issue was, at least it seemed to me, that the investors of Boeing were not so much invested into the future of the company. Because, if they couldn't be become not owners of Boeing, they probably would have done something to make it like to invest in in in the production and to maintain good quality and essentially to not have this situation.

Right? And I thought, well, how would I do it to how how would I implement a system which has this impossible to happen by construction? And then I thought that well, maybe I should introduce time into into ownership. So you like your voting power would be not just number of tokens, but maybe number of tokens multiplied by time. But what would be that time? One variant would be like the time you already own the token, but another one would be the time or the minimal time you will own the token. And that was actually a more interesting one. So you

can lock it, you will not be able to move it for some time, and maybe maybe over that time you would be invested in this token and in this system, and your vote should count more than someone who will exit I don't know, tomorrow, right? So that was the main idea behind the DAO and well, and I was not not the only thing. Another one was um distribution of emissions. Um well, I wouldn't say there was a particular um well, a particular motivation there apart from decentralization because obviously we want to control everything

in decentralized manner. So there is no single person controlling it, right? How would you do it? And then we we need to use the votes, and then it came down to it's really distributing votes across pools using this gauge system. And it's actually called gauge not because like not just because it is called gauge because I worked in a laboratory and in laboratory they were many gauges which measure things. They measured like vacuum pressure or like I don't know laser temperature or like magnetic field or current or whatever,

right? So there were multiple gauges. And I thought, well, maybe here we measure how much liquidity every user holds for how long. So let's let's call this thing this smart contract which measures this gauge, right? But but now it become weird because users say, "Oh, deposit in a gauge, stake in a gauge." It's total nonsense for someone who is coming from physics actually, but that's that's how it turns out to be. But it actually went from this this is slang which appeared from laboratory terminology.

>> I have a question because we've you mentioned how instead of just doing buybacks, you decide to do this gauge system. Today you see a lot of tokens where buybacks is the cornerstone of their tokenomics, whether it's hyper liquid, whether it's pumped. In a way I like to think of hyper liquid's buyback as essential as Bitcoin's 21 million. It's like, "Oh, you know that they're always going to buy back. You know that they're always going to buy back." What are your thoughts on those systems using buybacks? >> Well, actually I was well, gauge system is not like it tells you nothing about about buybacks or not buybacks. I did think about buybacks first actually.

All all so it was the first idea, but then I thought, "Okay, uh these buybacks, are they going to remove much of tokens from circulation?" And uh it didn't look like much, really. So, it it it looked like it like it's it would be um unless there would be some market maker who is really drawing your price, and we like I didn't hire any at all ever uh with Curve. So, now I thought, "Well, um maybe um uh maybe uh this is not going to move uh any to move pri- the price at all to

to affect it in any no- noticeable way, and except if it's working for many, many years. So, maybe uh it's better to actually distribute these revenues, right? Distribute to who, right? And then I thought, "Okay, maybe we should distribute not to just all tokens, but only to locked tokens and only proportional to the voting power, which encourages people to lock for longer." And indeed, in Curve, um most of the CRVs locked, they are locked uh essentially forever. They are locked for 4 years, but they are extended again and again and and again. And uh um the this way, the number of tokens

taken away from circulation is um growing. And um later on, later on, I think uh Sam from Frax um sort of challenged me uh in that thinking like that. Like, is it is it really more efficient than buy and burn, right? And I thought, "Well, we we have It was many years later." So, I thought, "Okay, we now have a lot of data. Why don't we just measure it?" So, uh so we did.

And it appeared that actually actually this system where you distribute revenues to token to those people who lock tokens. Right? Which incentivizes us locking essentially. It um It actually takes away three times more to tokens from circulation than buyback and burn. Well, I mean it's not burning, right? So, it's it's locking, but it's locking kind of semi forever. So, you you have this caveat that yeah, if everyone stops extending in four years, they will be able to unlock the tokens and sell.

So, it's not exactly burn, but it's it's actually very close to burn because they still extend this these locks. And it's it's a little bit like I got to know like those countries who don't really sell real estate, but they rent it out for like I don't know 100 years or something. And you know, four years in crypto is like 100 years in the real life. >> I I would say like a key difference is with buyback and burn, it's the team deciding like okay, we're going to or like a committee or the group internal deciding what to do with it. Meanwhile, with the locks, it's a much more democratized version of that.

>> Yeah, yeah. But you can have buyback and burn happening autonomously without the team involved. You can. >> Yeah. >> Um Yeah. But yeah, I'm really even with like the hard hyper liquid for example. Well, buyback and burn is a little bit of a meme right? >> Yeah, it's a meme. Yeah, like like 21 million is a meme. >> And that's and that's actually the most powerful part of it. Especially the burn part, right? So, you you You stocks stock buybacks, right? But when you when companies do buyback of stocks, they actually hold those stocks. So, and here it's bought and burned, and burning is, you know, such an emotional and

destructive event that everyone is kind of growing going crazy about it. Um but yeah. >> So, in both cases, would you say the benefit is from how users react to it either than having more confidence because of the buyback and burn or them locking because >> Uh yeah. Yeah, I think I think locking still still has to have some more work on sentiment. In principle, tokens locked forever, isn't it? Um isn't it crazy? Isn't it as crazy as burning them, right? Uh so, in principle, people should react emotionally to that, but apparently not so much. But physically, there are

more tokens removed from uh circulation by this, and it's actually much more. For example, in Yield Basis, um we actually calculated it it works for not not for a long time. Uh And we had um I think more um more tokens locked uh than tokens produced by emissions over this time. >> Nice. Um and you mentioned Yield Basis. I was doing a bit of research on it, listening to a few interviews of yours, and you basically took uh previous primitives and products that you built, whether it's CRV USD, the FX pools of Curve, uh

and added some new primitives as well to create the whole Yield Basis system uh where basically the yield is generated through market making. It's uh generated through volatility, and you have this uh yield-bearing Bitcoin asset from Can we >> Essentially, yes. >> Essentially. And so, can we actually go into yield basis a bit more and every stats that we can like point to it the traction and it working? >> Mhm. Mhm. Yeah so um well, I guess it's probably fair to start from what preceded yield basis.

>> Yes. >> Which were Curve products. So, first there were crypto pools. Crypto pools created in 2021. Right? And I didn't want to do simple XYK pools for volatile assets, but it seemed like something is needed for volatile assets, right? So, I created uh um crypto swap algorithm which is having concentrated liquidity.

So, how it works, you concentrate liquidity around certain price if you have two assets. If you have three assets, it's like you have certain um certain pair of prices, right? But anyway, you concentrate liquidity around certain price and you move this concentrated liquidity along with the price, but only if you have profit enough to move it uh to move it there because apparently um rebalancing concentrated liquidity is a lossy process, right? So, every time you move it towards the current price, you experience a loss.

And you'll but you want to to make profits, right? So, if you track profits really well and if you measure these losses really well, you are you would be able to to have like positive P&L, right? So, you you just keep track of P&L right in the smart contract. So, I did that by making um well, by actually making it at at the invariant system level, right? So, you I just defined everything in such a way that profit is up only. It cannot reduce, right?

Well, probably simple way to say that is this. Let's say imagine XYK bonding curve, right? The same as Uniswap one and two. If you have even if you had zero fee, if you have price changing, you move along the curve and square root of product of balances is not changing. And the square root of product of balances is a very very good measure of your money in the pool. Like if you put two times more money, you have two times bigger invariant. But invariant doesn't change as you change the price. So, it's and if you introduce some fee, then this invariant goes up only, right? So, even with managed concentrated

liquidity, you can introduce some measure which is up only. So, which is what is measured in the pool and then you can split off, you know, 50% for example and spend it on rebalances on moving this concentrated liquidity. So, it was done back then. It actually worked. At the peak, three crypto pool, pool with three coins, USDT, WBTC and ether, right? It's got bigger than 1 billion dollars in size.

Just one pool. So, oh, yeah. It worked pretty well. But obviously, um well, it was more efficient than constant product pools, but it still had this problem of impermanent loss, which is basically if the price is going up, then you are selling the coin which is going up. So, you are um inevitably getting less value than if you didn't do anything, and you of course hope to make it back by fees, but maybe you don't necessarily make it back. So, anyway, impermanent loss was a problem, and I think it's that's what

made these pools pools not so big at the end. Um at the same time, back in beginning of 2022, I came up with algorithm, which turned into Curve stablecoin and Curve lending, um Llamaland. >> [sighs] >> And um this algorithm uh it allowed to do reversible liquidations. So, you can if the you know, when the price of an asset is going down, that it gets converted to stablecoin and converted back to collateral when the price is going up.

So, um and later, in 2024, I figured that actually uh this um reversible liquidation, it um can be applied uh if you make it in a totally different way. With totally different math, but the same idea. Like when you uh um well, uh some math is uh math is similar to formulation, but not at the result.

Um you can actually use it to fight impermanent loss. Um like if you look at impermanent loss from a different perspective, then it's just a mathematical artifact, where the price of your collateral is proportional to well, price of your LP token is proportional to square root of your asset price, rather than asset price. And you want to straighten this square root, so you want to just remove it. So, you need to you want to make some position which makes essentially a square function. And uh um Well, that is >> by a mathematical artifact? How would you define that term?

>> Right. Right. So, imagine that you hold Bitcoin. Right. >> Yeah. >> You hold Bitcoin and your value is just Well, your value in US dollars is just equal to the Well, proportional to the price of Bitcoin. Right? Price of Bitcoin went by factor of two, your value went up by factor of two. Very nice. If you hold US dollars, your value is constant. Whatever the price of Bitcoin is, it's just constant in US dollars, right? But, if you put two in the AMM, in the constant product AMM, then value of your um holdings in this AMM, because it will be arbitrage and everything, it will be if you neglect the fees earned, it will be proportional

to square of Bitcoin price. Right? So, it appears that your like portfolio value is some mathematical function of your of the asset price plus something, plus something earned. Right? And this mathematical function is something you don't want. You want something like holding Bitcoin or holding US dollars. But, you don't want to hold a square root of Bitcoins. I guess. Right? Um Well, I mean, I was okay with holding square root of Bitcoins, but I figured that maybe people don't.

Um Uh right. So, anyway um I I I I thought Well, I need to get rid of this square. How do I do that? And then I thought, well, maybe something which can get rid of it is compounding leverage. Because if you compound leverage by of factor of two, then it will be it will turn into a square function. So, maybe if you have apply square function to square root, this will just disappear. And that that's kind of the idea behind Yield Basis. I did it with a special automatic market maker, which is just like Llamaland converts your collateral into um well

into US dollars, for example, or the other way around. And it has that and you know, it essentially reduces or increases the stat depending on where you go up or down, but it keeps the leverage constant. Keeps the leverage but compounding leverage of what you know, factor of two. So, at at all times you're like well, it tries to keep that to be equal to half of your collateral value essentially. And it appeared that it works. And you do have profit left after you do that.

Um well, there are some caveats here like you like which point you apply the leverage around, right? And but but it appeared to worked to work and it was proven well, first in simulations and then in in in real life. So, in in like like when Yield Basis was operating, I was comparing to simulations and actually found that results given by simulations are very very similar to results achieved in Yield Basis in terms of what kind of what what profits to expect essentially right?

Um And yeah, so well, we are still improving it using these simulations because just recently it appeared that maybe it's earning a little bit more than simulations show. And simulations actually those simulations are like all simulations I ever did, they simulate just arbitrage. Um and arbitrage is often considered toxic flow. So, they only simulate toxic flow and how can you earn make money when all the flow is toxic? But actually it's not toxic. The arbitrage is not toxic flow. It's It does make you money. But there is some truth to it if you have flow which is not arbitrage, then you actually make

more money than just arbitrage. And that's what we're seeing right now. The flow is not just arbitrage. So, maybe you'll basically >> What are the other types of flow that you're seeing? >> Yeah, well, just uh aggregators bringing some uh bringing some volume. Well, uh I would say that my thinking was that when people trade in aggregators, it's equivalent to arbitrage. And it is almost true, but not necessarily. Uh arbitrage traders are still I would say hm still uh still more like um um still smarter in some sense, like um economically more logical than real

users. So, real real users may buy something when arbitrage traders already figured that maybe it doesn't make sense to to buy at this price. Right? So, um anyway, um so that's why real like real flow brings a little bit more than arbitrage. >> I'm curious if you've seen a a change in the user base since from your days at Curve to the days at Yearn Basis. Um I feel like, you know, the world of on-chain before early on is a lot of experiments experimentation, uh a lot of degens, uh things were still being figured out. There was a lot of opportunity, but, you know, now over the

years, uh the world on chain has gotten much more commoditized, more efficient. Um and that leaves less opportunity. So, I'm curious if you've seen uh any changes or any differences in, you know, building for users now than 5 years ago. >> Yeah, I would say that um probably probably that's true, actually. Um and I've seen that seemingly that um affects Uniswap, actually. Because on on Uniswap, the volumes were used to be much bigger than that just arbitrage volumes. And right now, they are much closer to what what is given by just

arbitrage. Um and I guess on Curve, that's true as well. But I mean, um I was thinking that all of the volumes are just similar to arbitrage anyway. >> Mhm. >> And um I think um volume uh like trading volume similar to arbitrage trading volume uh is just um a sign of market being very efficient, right? So, if you have more trades than just arbitrage, then it means that there is some inefficiency in the market because like maybe um

yeah. So, but but anyway, right now, I'm seeing that market is becoming more efficient than before. >> Mhm. >> And I I'm curious uh what challenges you've seen uh in scaling Yield Basis if if at all? Have you thought about >> Yes. Yes. Absolutely. >> Yeah. Uh what I thought was it Yep. >> Something I didn't think of first uh too much is um not the fundamental limit to Yield Basis scaling, but how how it affects stability of the stable coin it uses.

Because if you think about it, you have a gigantic crypto liquidity, right? And if crypto is moving in any direction like whichever stable coin you put it, the crypto will will try to pull stable coin price with it, right? If it's going up, it will try to pull to to pull stable coin up. If crypto is going down, it will try to pull stable coin price down. But you want stable coin to keep the price constant, right? And of course, if it was redeemable like USDC, well, sure, it would be arbitraged by via banks and everything and it would be totally fine. But if it's Curve USD,

then you need to do something to stabilize it. And doing something for a CDP stable coin means uh having big enough supply sinks. Uh that was what actually limited scale of uh yield basis when it started. And only recently I figured out how to make uh how to make these supply sinks for Curve USD scalable and scalable on demand. Like if you have some imbalancing imbalancing happening, then you want to increase the supply sinks and the supply sinks uh well, have to

um have to let's say support the system while this price pressure is happening. And um it was launched with uh incentives distributed through Merkel. And these incentives are going up if there is a price pressure down on Curve USD. If the price pressure on Curve USD is up, it's not a problem because back keeper system cares about that part. So, you only want to care about pressure down. And uh with this in place, the scalability limit is actually not the stablecoin used anymore. It's uh the fundamental limit of how much uh

revenues can you get from Bitcoin volatility. And this limit is uh much bigger. I don't know exactly how big it is, but my estimate is I don't know, some several tens of billions of dollars probably. Uh well, I hope to explore this limit. But anyway, at the at that scale, you would kind of start dominating the world liquidity for Bitcoin, and that would affect volatility of Bitcoin. It would start going down, and that's when you try you you will start decreasing your your revenues. So, there is some saturation point which will happen at that point at that time, but but it's,

you know, at least 100 times bigger than now, so I think that's fine. >> Yeah. Patrick, it looks like you're about to say something. >> I I I I I was, but um if you have a follow-up on that, >> I didn't have a follow-up. Yeah, you can go ahead. >> Um yeah, I was going to say we uh I know we've been talking for for a while. We have we have a couple fun questions to to close things out. Um so, it seems like a lot of the products you've built are products that you would use yourself. >> Absolutely.

>> Which which is I mean, just as an anecdote, great great way to design things, right? I think almost every product DeFi Llama has came from because someone on the team wanted it to exist, and so they built it. Um if you weren't building Yield Basis, what would you be building today? >> Yeah, good question. Um Well, if you look at things which Well, may maybe I will actually be building them uh because they just beg to be built right? One of the things um something I'm using today right?

Just like I was swapping between stable coins of the same denomination, right now I'm very often swapping two stable coins of different denomination, like to euro stable coin or to Swiss franc stable coin, right? And you can actually launch individual pools based on curve pools with actually the same modification as yield basis uses. They are FX pools, right? And I would totally Well, I'm I'm I'm using them actually not every day, but maybe every other day.

Um so, I'm buying euro and I don't know, sending it to some IBAN or, you know, paying some bills or buying Swiss franc, paying bills. It's It's actually possible to do. Um but the problem is liquidity as usually. There are curve pools for that. >> Always the problem. >> Yeah, there are curve pools for that, but and they are not bad, but sometimes they feel like not enough. And um well, to to get liquidity better, I think I have some thoughts how to how to make a system for foreign exchange better, right? But then it it would involve actually several

currencies, not just two. And yeah, it feels a little bit like rewiring how foreign exchange is happening. It probably wouldn't really be needed several years ago, but right now there are a bunch of fintechs who issue cards in different currencies, who can send money in local bank accounts in local currency, and I do use several currencies there. So, it's it's something I need and looks like I probably should just Well, I I sketched out actually how how this system should work and I probably should really launch it, right?

Well, another one is Well, not me, but actually one of our uh one of the people in our in Curve team uh figures out that actually um well while leverage is good in Lamaland, it's not as good as it can be. Um you can actually get spot leverage as big as hyperliquid, but without perps, just on spot. All right. So, I guess that deserves to be launched as well.

Or even you know, something probably uh something probably silly, but I was I'm not satisfied with any crypto wallet, right? So, I'm I mean, with anything JavaScript. I don't like JavaScript. And I don't like using software written in JavaScript, however weird that may sound. But, all the Ethereum wallets are written in JavaScript and I had to suffer for years. And uh right now with LLMs, I thought, "Well, why don't I ask LLM to write me a decent wallet which doesn't use JavaScript? Why cannot I just have a wallet written in QT and whatever? Uh it's it should be possible."

So, I did and now I'm using that wallet and it's uh um and it's been awesome. So, I'm not going back to JavaScript based wallet. >> Interesting. Is this Is that a Is [laughter] that a fork of an existing wallet or is that >> No, it's it's it's from scratch. >> From scratch? >> Yeah. Yeah. QT. Because like I I just thought like, "Okay. If I was an LLM, I wouldn't like JavaScript. It's such a horrible language which is so imprecise. I would use Python because as a human I use Python. So, why don't I ask LM to use Python? And yeah, actually I did I did like the result. The code is actually readable.

>> [laughter] >> Wow. You know, maybe we'll see a wallet soon. Who who knows? Uh >> [laughter] >> Well, I mean you can I mean you can try today on my if you get it from my GitHub. >> Yeah, that's Is it Is it Is it open source on your GitHub? >> Yeah, yeah, yeah. >> Okay. >> Well, you heard it here. Uh you know, we're looking at the future, but now for my final question, I want to go down memory lane. And if you had a certain on-chain memory that stuck out over the years of you building, if you have a favorite memory, and you know, there's been a lot that has coalesced over the past you know, 5 10 years.

Is there any memory or moment that sticks out? >> Mhm. Yeah, let me think. What would that be? Mhm. I think this would be some moments in DeFi summer in 2020 because it was such a wild time, right? I think it was mhm Well, it looked fairly wild when um I think when Andrea launched a worthless governance token.

I think that's how he called it. Which we know as YFI. He um created Well, there there were already Y earn pools on Curve, right? So, he started distributed to those and created what he called pool two. And you know, wrote this little write-up. And I think initial price of that token was $3, right? And after that it was growing and growing and this created insane APRs on Curve, like I don't know, 500% but the token was still going up in price and you know, pools also got

bigger in size and yeah, it was so crazy to observe, right? And and and it was like uh the price was of that this what worthless token was going up by, you know 100x 1,000x and I thought, well, it probably should be it will stop here but it didn't and this was kind of a and also people were depositing funds in the pools and it looked like a crazy, crazy DeFi summer moment. I would say that and uh I mean, you just saw the system scaling very big

uh just because of I don't know, because of that weird moment in time. Um it's interesting that today um we saw some measurements of how many users Curve has. It actually has many more more users than back in that day. It's number of user individual users on Curve was actually very small but it's actually growing. The TVL was big but the number of users was small. So, now the number of users becomes bigger and bigger but TVL is probably not as big as uh as back in like 2021, for example.

Uh but uh yeah, it's uh um I would say memories of DeFi summer are probably uh so far the most precious crypto memories. >> Yeah, that I agree there. Someone that was also a part of DeFi summer, and witnessed everything from Wi-Fi taking off to all the food farms and Yams, and it felt, you know, it things were a lot smaller, there were a lot less users, and the only place you could do stuff was Ethereum, basically. >> And I remember forks like there was Wi-Fi, and then Wi-Fi I, and Wi-Fi II, and that was so so crazy like forks of forks of forks. How is that possible?

>> Yeah, and then you had Sushi and all that. It's crazy like you know, being in the space long enough, there's only like a certain subset of people that were a part of that, and when you like continue working in the space like all of us have, people weren't a part of that, so they don't understand what happened back then, and it basically set the foundation of everything that we're building now. Everything is kind of like an iteration or improvement of those early days when things were so small, still starting off. Uh, still like we were still in the beginning of the experiments and experimentation phase, but now since we're still experimenting, but we learned a lot, and we we've taken those lessons uh, with us. So, yeah, I do remember those days fondly as

well, and I'm sure Patrick, you you do, too. >> Definitely. Um, yeah, final Any final words, Michael? Where can people keep up with what you're working on with Yield Basis? >> Yeah, I guess um, I guess you can follow what I write on Twitter for that, and I I do share all the newest developments on Curve and Yield Basis, and maybe I don't know, maybe some different kind of stuff will will happen there, or maybe just interesting thoughts, and

that's uh, um, yeah, people are uh are invited to to read and comment and or even ask questions. >> Absolutely. This is the the beauty of the space. You can just go ahead and DM and comment a founder directly and you know, usually they will answer. Um you can, you know, follow also now you can follow me at Defi Dave, you can follow Patrick. I don't know your username. I should know it by now. Is it Dynamo Patrick? >> It was Dynamo Patrick, now it's Pat F Scott.

>> Pat Funny story about that. >> The way it's So, just funny story is like I was the first I guess you call business hire at Defi Llama as far as I know. Um and the way it came about is I was basically unsolicited DMing uh our website and DMing my product feedback for like years. I'd be like, oh, like, you know, like I think like this would be a great feature. Like, oh, have you ever thought about this? And and I was making a lot of tutorials on it. Uh and then eventually he was like, oh, like, you know, I'm looking to bring someone on board to run marketing. How would you like to do that yourself?

>> Perfect. >> That's the perfect way. >> Yeah, and uh now, you know, follow Fully Invested too on all platforms, X Twitter YouTube Spotify uh Apple Podcasts. Uh and if you're interested in uh the products of Defi Llama Llama AI, we have a referral code in the description. So, make sure you get that as well. >> Maybe one interesting comment. Um I don't know if it if that's exactly true, but I think uh Defi Llama started out of uh out of frustration of the of wrong data of Defi Pulse.

>> It was. Yes, it did. >> For Defi Pulse was uh kind of like >> About Curve specifically, I think. >> Yeah, yeah, exactly. So, they always got it wrong and uh it was just unbearable and uh yeah, so apparently that's that's how Defi Llama started. Well, I mean still Defi summer days. >> The Defi pulses ended though a while ago. There's no more pulses. No more pulses.

>> Only llamas now. Only llamas. >> Only llamas. We have two llamas here, so. >> [laughter] >> Yes. >> Yeah. Michael, thanks so so much for joining and hope to see you again soon.