Steffen Schuldenzucker - Designing Resilient Reserve Portfolios for Decentralized Stablecoins

[Music] so a very fascinating talk we know that the centralization is very important but the question is how do you build a resilient uh portfolio to back your decentralized stable coin so you might need to turn to some fine German engineering to that end we are very pleased to give a warm Guten Morgen to Stefan three minutes all right three minutes we'll wish a warm Guten Morgen to uh not to tease it too much the guests will be Stefan um Sheldon Zucker Stefan shildenzooker from gyroscope um so in the meantime um does anyone know any good stablecoin jokes I don't think that there's yet been a stablecoin joke no all right what have people been doing so far in France uh earlier there was a few blurry eyed people who uh came in and they'd seen the fireworks last night the Eiffel Tower was there anyone who trickled a little bit later and was watching the fireworks last night yeah so some of you showing up just a little bit later for that so um how much fun was that to actually like just be in Paris at the same time that there's all this surge of energy coming into the crypto ecosystem for the first time in like a year and just see fireworks shooting off everywhere and everyone from web3 was uh really really a great uh great moment um so we hope the URL of course going to be sticking around for all the events throughout today and tomorrow so if you missed Julian's introduction earlier this is in fact the first stablecoin summit which has ever been convened to our knowledge so maybe there's been like some small events that took places and cafes or coffee shops but this is the first one of this magnitude and this size and it's such an important important subject um in my estimation if you look at the front page of coin gecko you'll constantly see within the top 10 number a number of stable coins and I don't think it's any mistake that we all here recognize that stable coins are the killer use case for cryptocurrency like it's not even a question like we've had nfts which are kind of fun crypto kitties clogged the ethereum network a long time ago they're still seeing an ft mints clogging it nowadays but if anyone's tried to do a transaction using a bank you're doing a wire transfer especially more than a few thousand dollars you know that the ability to effortlessly send currencies across the globe 24 7 is just by far the killer use case so this extends to the developing world where you know if you're in a country that's ransacked by hyperinflation it's a real problem that you have to face on a daily basis how do you keep your value stable coins have become a Lifeline to so much the developing world that way and then for anyone who's in the more like a of a world superpower you know you might just like stable coins because it's nice Refuge when you're trying to flip stable coins 24 7 between the newest you know Meme tokens or if your favorite cryptocurrencies back to stable coins to try and hold value so it's um you know I don't think I need to explain the import of stable coins to everyone here in the audience we all get it um then the question just comes like how do we actually design it uh there's been so many uh aspects of design we've seen already most of our speakers talk a bit about the stable coin Troy Lemma so anyway without further Ado uh we're going to discuss more designing a resilient portfolio with before mentioned Stefan of gyroscope Labs Stefan please thanks a lot do you use this oh wow so great to meet you everyone um I'm Stefan I'm from gyroscope and today I'm going to talk about designing a resilient Reserve portfolio for decentralized stable coins and let me give you a little bit of context about gyroscope um the idea of gyroscope is that we were trying to improve upon many problems we saw with some stable coins um I mean you you know like like some of them so risks are concentrated governance is slow um scaling issues and so on and um so today what I want to focus on on is the concentration of risks and this is why we think we need Diversified portfolios um it's working so often in stable coins that are collateral backed um one fade collateral has a risk of dragging down the whole stable coin and of course here doesn't have a laser so so um here in in March of this year this is of course what we saw that usdc went down and by implication many other stable coins also went down with the exception of usdt um so what you might be saying right now is well what this means is that building a diversified portfolio of mostly stable assets is actually very hard because we see the correlation right there right um so one thing that's going to make this a bit easier for me is that what we care about is long-term State sustainability more than short-term correlation um and the other thing that I want to highlight is that so this thing is obviously there but it's also incredibly rare So if we look at Price data and correlations and all these kinds of things that's something we can do but we also need to look at fundamentals because stable crashes of course these very rare kind of Black Swan events which means that you don't have a lot of data um so gyroscope the idea of gyroscope is to build a set of new Financial infrastructure advancements that will unlock a new third path between um centralized and algorithmic stable coins and so the the conclusion of this will be a new stable coin that we call guid by the way launch is planned for August maybe Q3 we have a Proto system right now that's on polygon that's under this address it's working pretty well this is mostly for testing so so gyroscope has these three pillars and I'm going to focus on one of them the first one is resilient defense of the stable coin itself the second one is liquidity Network to keep the token liquid and the third one is all kinds of redundancies built into the system and today I want to talk about the reserve design obviously and yield Source design so the the aim of gyroscope is to be 100 collateralized and backed by a diversified portfolio of crypto assets that are deployed in low risk yield generating strategy in a risk segmented way so you have actual useful diversification there um and the main stabilization mechanism is mint and redeem Arbitrage let me get a little bit more in detail so when someone wants to Mint or redeem gyd they go through a mechanism we call the dynamic stability mechanism and what this is going to do is that they need to provide let's say they want to Mint they need to provide a balanced portfolio of assets according to certain weights now they can deviate from these weights in some limited way if only small amounts are minted you can also just knit against one asset up to some tolerance Etc but by and large we want to keep this balance of Reserve assets and then what's going to happen is that the reserve assets are priced using a system we call the Consolidated price feed and they they receive gyd in exchange for the value of these Assets Now what does this Reserve portfolio look like um we have these buckets we call vaults like everyone does every Vault holds one or more Assets in them and then there's a strategy that tells us how these assets are invested usually if you see more than one asset in a vault that means that there's a liquidity provision strategy um so what what are these assets um at launch these are mostly going to be other stable coins so gyroscope will start out as a kind of meta stable coin with a diversified portfolio of other stable coins backing it and this is going to be the main topic of my talk um and then going forward we can start including other Assets in some risk limited way um right so I want to talk about now is give you a risk-based overview of the task of choosing these different stable coins that should back the system then I'm going to talk about custodial stable coins non-custodial stable content finally talk a little bit about strategies um yeah by the way so sources for this are of course our ongoing work at gyroscope and another sources the PHD thesis of my co-founder Raya glagasman who actually did a lot of research which you can see here on the Dynamics of stable coins and stability and comparisons Etc um the basic principle of Reserve selection is um like this is this is the point I want to stress let's consider stable coins as risky assets that usually do not manifest these risks in price um and then we want to build some risk segmentation to to get a robust and liquid portfolio um this one already said let's look at the fundamentals here more than Time series if we look at stable coins we can group them into like the the largest buckets you can think of the first one is custodial several coins like tether let's say um you there's an entity that holds certain assets um that's an off-chain process you have a token on chain and you need to trust that entity to well manage these assets well and do the right thing and the risks here are oh sorry and the risks here are reasonably well understood in traditional Finance so you have like counterparty credit risks you have censorship of course and then you have other traditional Finance risks and then as of course this non-custodial world where you don't have to trust the protocol the important processor on chain um and and then you have like new risks like deleveraging um you you have Oracle risks governance risks etc etc and of course these systems also tend to be more complex so the risk of smart contract bugs is higher so let me talk a little bit about custodial stable coins and um so we before I before I go through the slide I want to say that custodial the risk profile of custodial stable coins is typically a little bit different than for uh let's say very non-custodial very crypto-backed systems so what we would expect for a custodial stable coin is not that the whole value just disappears so something like UST you wouldn't really expect for tether but you would maybe expect that there are some haircuts that there is some some some kind of discount they could be freezing off assets but that also doesn't mean that these assets are gone forever um so so kind of what we want to do is make sure that not too many of these assets go too far Downer or too much Frozen for too long a Time um and and so with that in mind we can think about these these different risk categories most of these resemble a money market fund and then we can think about that so of course there's like asset backing risks there are risks that that are maybe not just about the price of these ads but there's also something like duration risks if they hold like um like fixed period bonds for example um there's a counterparty risk obviously censorship um the upgradability of smart contracts of course there's a risk that the custodian will just decide to make the system into something else on their own volition um there are these regulatory aspects and and I mean typically of course usually we think about this right what if a government intervenes and and the thing gets banned and they are forced to censor or something but of course there's also a positive thing about regulation which is accountability a regulated entity is typically maybe more trustworthy I guess it depends on the entity but other things equal that's the case um and then there are liquidity aspects so custodial service coins are typically reasonably liquid off chain um and and of course we also need to think about on-chain liquidity of these assets uh what did I want to add oh yeah right so when we are choosing here because we want to build a portfolio we're not so interested in a ranking in the sense of what's the best one but we are more interested in Risk structure so that we can we can choose the risks we are exposing ourselves to um for for non-custodial ones obviously the story is a bit more complicated and I want to again like distinguish two uh broad categories um the first one are leverage based or collateralized that position systems this would be something like the the die loan wall so the the original Multicultural Dye System where someone wants to take leverage against their position they lock their asset up and they get a loan that's overcollateralized and the other one would be swap based systems I mean robot Auto Outlet in this in this talk before um where an arbitrageur sees that let's say the price of the stablecon is a little bit higher than one and so they Mint new assets um at the at the mental redeem mechanism to get Arbitrage and but the answers themselves are then protocol controlled value um now if we want to make this much more complicated we can use something that I co-founder rioted in this research and put everything into this three-dimensional Matrix I want to briefly talk about the dimensions I'm not going to talk about everything here the first question we should ask about non-costolic subconscious is um what's the asset backing obviously and they're kind of kind of two Realms here right the first one is asset backing could be exogenous something that's not related to the system itself and the other one is as a banking could be endogenous or something something like Luna where a system is backed by its own governance token um then we should think about who absorbs risks are there agents that need to absorb risk like and collateralize that position systems like these ones or are there protocol assets that absorb risk and and gyroscope is a system that that goes here so we are exogenously backed with a mentality mechanism so the protocol would absorb risks and then this is third dimension which is who decides about issuance and and again Robert talked about this in the talk before if the agent if there's a self-interested agent with some let's say leverage interest who decides about issuance um then you need somewhat somewhat more involved system to make sure that the supply of the stable coin is at the right point um now I should talk about two more the first one is die of course which sits in in both of these regions started out here and now it's mostly here um and the other one is curve USD which currently primarily sits in this region with this interesting new mechanism called Lama essentially these are collateralized depth positions but the the liquidation of course is handled in a very smooth way in a very Innovative way so that you don't have separate liquidation events but liquidation kind of happens smoothly over time which which I I think is quite nice because it limits losses of people from these liquidations um and then down here are these systems called pet Keepers where of course the idea is if in certain markets in certain amms the price of curve USD is too high and the pet keeper would Mint new curve USD and deposit them one-sided into that pool of course the implication of this when you do this is that then the pack keeper holds a portfolio that consists of curve USD and whatever the other pool is and that means that effectively that's an asset backing by a stable coin um again in this in this very nice smooth way that I think is very efficient um when you know what the primary market for this for this coin will be the the main trading menu now right I should show this uh we have kind of this like diagonal of death here um so so that's that's not the only reason why we thought that that so exogenous maybe this is obvious but exogenous asset backing seems to be a pretty good idea and endogenous asset backing seems to be um very tricky to get right um right I want to talk a little bit more about the the risks and these systems so the basically that same the the kind of cousin of the slide I did before um so we we focus on exogenous collateral only because we think that makes sense and once you do that um of course again you need to think about asset backing risks um there is of course composition going on right so um if let's say die holds a lot of usdc and I hold a lot of dye then I'm implicitly holding a lot of usdc but I'm also holding other things so I'm exposed to a certain portfolio and you in your minds you can come up with this like graph of exposures where at every node risk is remixed and transformed and new risks are added and new stability mechanisms are added um there seems to be there seems to be a trade-off between having stability and having some custodial collateral we saw that die before they introduced the DSM had a certain stability problems for example the price of If drop the price of die spiked and it seems that you need some custodial collateral to make the price at least short term stable um otherwise you you get like these delivering spirals negative rates Etc um then there's a question how does the system handle under collateralization events and again this refers to both um kind of the The Leverage side so collateralize that positions and and swap swap based mechanisms so how does a reserve back system handle undercoateralization of its Reserve portfolio and how does it collateralize that position based system handle liquidations um then there's there's governance of course unfortunately there's not a custodian who can just decide to upgrade the protocol but of course as the governance process um and and Regulatory risk um maybe depending on the involvement of other entities um we we have this Oracle risk which is of course unique to DFI that you rely on some other price Oracle to pull in prices um what I want to say here is that um again I don't want to say that all of these risks should be avoided because when you're building a system it's impossible to avoid all of the risks but instead we want to build a diversification across all of these risks um and then smart contract risk and liquidity strategies so the main the right I should repeat the goal so the the goal is to generate good risk-adjusted yield so if risk is low and the real yield is also relatively low that's actually fine um the the main strategy that um we're going to use at launch are liquidity provision strategies we have these custom liquidity pools that we call elliptic concentrated liquidity pools on balancer they integrate with balancer and the the idea is that these are special liquidity pools that allow to shape the liquidity in a way that fits the asset pair to a relatively High degree um what what we should keep in mind is that when we provide liquidity between those between two assets that's not diversification right that's unioning up the risks of all of these assets because if one of them goes down you end up with a cheaper one um so we um these these liquidity pools Union risks between asset Pairs and then but you can have multiple of them and between these risks are going to be segmented um the idea then is that we would provide liquidity between assets that are correlated anyways so assets that are exposed to similar risks already and if we do this we are not introducing new risks in the mix um I I should maybe I'm not sure if I should talk about this feature but um one kind of nice side effect of doing all of this is that um the assets are held so balances designed such that the assets are held by the Vault um so this reduces censorship risks a little bit because we only need to be concerned of the risks that balancer would be censored by some component and not us or or um yeah not not ourselves um then we have some other strategies um you can you can put things into lending Protocols of course here you need to think about collateral backing we could put things into the idsr essentially um this is a way of getting access to real world assets because maker holds them you get all of the risks and also um a significant part of the yield when you do this um and then we have some some other ideas for the future so now let me give you an example and I want to stress this is just an example right this is not what we're going to do but this is an example I use to illustrate um all the things I talked about and here here is something one could do you have die in the DSR you have usdc and flux you have a liquidity pool between usdp and gusd LUSD instability pool and then I try to be provocative so I put tosd with your CT in there of course in the of course in the future um you may not want to do this anymore or maybe you do I mean we we need to see what happens so why do I claim that this embody is the things I talked about this risk diversification um well if we look at dye we have decentralized backing here um over collateralized we have real world acid exposure and we also have some exposure to custodial stable coins through the PSM if we look at Flux Of course this is custodial again so there's a risk overlap there but it's limited um we get some custodial risks again and again we we can get some exposure to real-world assets this eclp has two I forgot what this is called the New York regulator regulated stable coins which arguably should reduce the risk and they should also be similar risks so if they suddenly let's say the US suddenly decides to go Rogue on on all of defy then both of these assets would have a problem but we don't have one of them in a pool that would then create a problem for the other asset um LDS in the stability pools of course a very different different story fully centralized um you get some of the price volatility of Eve you get the risk of deleveraging spirals which these don't have um and and then we can put so here I wanted to put some custodial stable coins that are not based in the US which are then of course exposed to different regulatory questions different regulatory regimes both in the positive and then the negative of course so you you might argue that at least are less well less transparently regulated perhaps um at least from what we see and but at the same time of course you also don't have the the geographic risk of the US but you have other Geographic risks um exactly I want to add two more things um the first one is that once we have built this we can in the future extend it and so this is what you've already seen and we can then extend this with a second mechanism I mean no one says that gyroscope forever has to be purely this asset backment and redeem scheme um but instead we can also have over collateralized Landing and what we are thinking about here is to have an eclp that trades that provide security between let's say eth and a liquid staking token or between two different liquid staking tokens and then of course here you're exposed to eth as a risk so you need to over collateralize these you cannot put them raw into the reserve but we can do this and we can do over collateralized lending here and then we have a second a separate Branch how gyd can be created I want to give like a quick uh well like a quick peek of what we're building right now so right now we are working on this like big sheet of all of the different asset risks and all of the different stable coins and so again the goal here is this is a bit different from from previous efforts that um Steakhouse it for example the goal is less to kind of rank the stable coins to figure out which one is the best but to give you an overview of the risk structure and the connections between the different stable coins um right and so basically you see everything in here that we talked about so to to conclude the main message from this talk is consider Sable Thrones as risky assets think about the fundamental risks and the structure fund fundamental risks I talked about custodial Sable coins non-custodial stable coins I talked a bit about strategies and if you found this interesting please get involved um we we have like like a website and a Twitter and a Discord please come to our Discord it's going to be great thank you very much [Applause] yeah sure sure so um would be okay so um I draw I need to draw a little bit into the air here and I'm sorry for that um let's say you have unit swap V3 right and if you have a position unit for P3 you have price bounce and you provide liquidity between those two price bounds so what we felt is for some asset pairs basically you know what these price bounds are probably going to be these are not very volatile assets let's say something like so we had actually some great success with eclps for liquid staking tokens against eth or liquid sticking tokens against each other and it's reasonable to expect that the price is not going to be like five right um at least when you scale to get rid of the yield so what the eclp allows us to do is that you have two price bounds and then what you can do is you can shape the liquidity in between so you can say um where should the pool have most liquidity and then how quickly should it fall off towards the sides and then there are many things you can do with that the first one you can do is you can actually match historical price distributions reasonably well it turns out that they're kind of bell-shaped and this one is also kind of bell-shaped so that that works okay um but you can also match assumptions that you have about the future um reasonably well um and so I think so we are we are using this for liquid staking tokens I think it's also very useful for stable coin pairs where it's basically you know right where the that it's probably going to be one to one most of the time and then you shape liquidity around that and you provide that liquidity so right so um I'm looking at the um the your demo Reserve buckets and something not like um was a bit confusing to me was um like you let's look at it from a regulatory risk perspective you try to diversify on that but what about the implied risk so take an example um the U.S going rogue um it's always first of all when everything everything is correlated down to minus one when everything is going bad right if you look at the flocks die and let's say um the other stable coins they're also negatively negatively affected by adverse regulatory Legacy stuff but then totally different buckets so how do you manage how do you manage for that yeah that that's a good point um so so maybe what you're telling me is that my example was not super complete and that my example is actually exposed to that one risk yeah exactly exactly so so um I mean I I I think I said this that you composition risk is definitely something you need to think about um I guess regulatory actions when they are severe enough at some point just become very hard to diversify I think it is I mean it is diversifiable in some assets like liquidy for example which is which really shouldn't be affected by this too much um I think also having this bucket with stable coins that lie outside the U.S is also a diversification there um I I think it's I think it's like um I I mean I hope everyone in this room is going to agree that a very severe regulatory action by the US would be would send like very severe ripples across the ecosystem so maybe at some point of your scenario is is very extreme it actually becomes very hard to diversify um but I mean in something like in something like die die would not die when uh when um you when uscc goes down right it but there would be a haircut so there's some protection already there I'm sorry could you speak up a little bit yeah yeah um maybe you're right sorry what's the scenario can you can you say this again I know what happens am I missing something um because the way they're structured right now they can compound things um I guess the only exception I would make blur from my understanding but but they all at a basic assumption that the price goes up continuously and uh and it never gaps the margin requirement which is the reimbursable cloud if if that happens the way the way the Liquidation Center is structured is that you're going to create liquidation Cascades they're going to mathematically I'm sorry not not pulling something but they're going to generate themselves yeah um I think I can't explore this scenario live but I would be very happy to explore it further with you offline it did happen like that with um price going up and shorts leaving some insolvent positions well not that large scale so that was managed but it it did happen in the past but I guess I also wanted to mention one other risk which which is probably temporary not fundamental but uh also relevant micro PSM risk like if you observed price Dynamics during usdc DPAC not just usdc went down but also also usdp gusd and die and that's not a coincidence that's because maker PSM has really a lot of money in it like I don't know billions right and all those coins are essentially pegged to each other via the PSM until the PSM gets are gets arbitrushed out so at least for some days all these stable coins went down together with of usdc despite having absolutely no exposure to svb yes yes um so okay okay I'll I'll try to come up with a strategically clever answer um gyroscope isn't doing this um so what gyroscope is not so this like mental redeem scheme is we're pricing the assets um so if uscc is at 70 cents we're not pegging one to one to usdc but the system would see that there's a stable coin it's underpack is under collateralized it would see that and not on a collection Story the stable coin is on the peg um right now it would actually uh just prevent people minting against that stable coin because it would basically decide that this price is not meaningful let's put it like this you could also switch the system and it would just allow money can redeem at whatever the price is but I think that's very dangerous so the solution to this problem is be a good example to to make a and make them change the way the PSM works thank you thank you uh big round of applause for Stephanie McStuffin and just a quick reminder since uh now people are trickling in it's standing room only here but there is upstairs uh raft Finance is on the other stage conducting some uh some demos some workshops so if you are in the back and can't get a seat then go ahead and check that out