Risk Ratings: The Missing Piece for DeFi & Onchain Finance Mass Adoption

Transcript

All right, it's half past so we will be going into the next talk. Um, my name is Marson. I'm co-founder of Redstone and Credora risk ratings and today I'm going to talk a little bit about risk ratings which is the missing piece for DeFi and onchain finance mass adoption. I've been in the space since 2017. I have observed a couple of cycles already from alternative ones through DeFi summer NFTTS and many many others and as you can see I'm one of the few guys in a suit right now and it's not that I enjoy that kind of outfit like on daily basis but it's rather right now we are entering the phase of our industry where

we have to also like be as professional as the people we want to cather to like we cannot be anymore either children or like just frog pyps on X or Twitter like we have to stand up to the level of uh those finance guys. So a few words about Redstone itself. We've been around for over four years right now actually five. Uh we have never ever had a single mispricing or downtime event as an oracle. Uh which is not so common. Majority of our competitors including the big boys did have smaller hiccups and usually an Oracle Oracle failure comes with like a very bad consequences but that wrongful liquidations protocols going down users being extremely unhappy

with how their positions are treated unfair. Uh some of our partners are for example securityize that chose redstone deliberately as the main and official oracle for all the tokenized funds including black rockckrock bidd Apollo a hamilton scope and many many others when I was going through the due diligence from black rockck that was like an experience I can tell you those people are extremely extremely careful who they are working with we are uh the first oracle that launched on the canton network we are featured a validator and we expand uh dayto-day uh within in uh that particular uh ecosystem. We power Morpho, Spark, Camino. Right now there is hardly any protocol in DeFi that is not necessarily like using Redstone.

We're live on over 100 chains over 200 clients trust us dayto-day. We protect north of $6 billion still in that bare market and uh deliver price feeds for over thousand assets. All right, so Redstone is an oracle. It's growing. about why risk ratings. So last year in September we announced that we acquired an entity called Cordora network and with that acquisition we decided to expand within this risk aspect. Reason we did it is in order to produce high quality ratings you do need to have uh extremely accurate and diverse data feeds and that is what we've been doing

for a couple of years already on the Redstone side. Before we did that acquisition, we asked ourselves, okay, well, how does the history of the ratings industry look like? Because you then can look for similarities. So, uh it started in very late 19th beginning of 20th century where S&P and Moody's come into existence. But being frank, when I was digging into some historical data, no one cared. So, it was pretty similar as it is uh right now in DeFi. In the very beginning, uh very little investors were actually paying attention to those ratings. But then the inflection point came in 1936 when the banks had the obligation to invest only into assets that were investment grade which meant that on a

scale from D to A from specific uh trusted and recognized risk ratings agency they could only invest in products that were BBB plus or higher which were called investment grade. Anything below was called high yield or junk and they couldn't interact with that. And that was the moment that the rating industry like properly took off. Then in the 70s, everyone already knew about uh when you're launching an investment product, you really need to get a rating in place for gaining investor trust, especially in the early days or to scale to uh hundreds of millions and billions of dollars. But I bet you all saw the big short uh that uh in 2008 illustrated the um maybe not the collapse but the huge hiccup of the

financial market where ratings agencies also played like a significant role not necessarily doing their job in terms of the quality of the ratings they were assigning. And in 2020, we've seen the DeFi summer, right? Like the uh various flavors of swaps from sushi, pancake or any other food that was appealing to the users. And now in 2026, we see the um in the um financial institutization of the industry. And whenever I talk with the big bank financial manager or like an asset allocator, they do ask, okay, but what is the rating of that product that I'm about to deploy my capital into? So this is a kind of uh area that we want to tackle. H Vitalik posted in September

last year that low-risk DeFi can be for Ethereum with search was for Google. And there are two concepts here. One is low-risk D5 and the other one that we promote is riskaware D5 which essentially means to attach next to APY. So the earning that you can do get the risk associated with specific um implementation for vitilik uh investing mainly in the lowrisk asset was the key but in our review you can still invest in something that yields you 25% or 20% but you should be aware that for example it has if it has C rating there's a high probability of default so it's not about uh taking away higher yield and higher uh risk assets it's just making the

users aware what they are signing up for when they are deploying the capital. And in order not to just put blank words over there, I'm going to bring three cases. The first one, anyone recognizes that chart? Anyone can say what it is? It went from uh 1.26 to almost zero like immediately overnight. It was stream finance that blew up uh last year. Uh it turned out that the protocol was missing uh 93 as far as I remember million dollars and their backing and everyone who had essentially the stable coin of theirs that was supposedly worth $1.23 1.26 sorry h turned out to be worthless and it's not a bad luck like I I've

heard people saying this event was a black swan or a bad luck. No, this is like one of the worst sentences you can say because there was so many uh signs that this protocol will eventually collapse and we as red credora unfortunately went live with ratings in November 7th and that happened November 3rd. Uh but reactive reactively we gave it C minus which is the lowest rating we've done like in the history because uh just all the fundamentals were extremely shaky over there. And there's another um chart I can show you that is way more recent. It's from the last weekend which is resolve a stable coin that got hacked and an authorized party minted $80 million of additional uh stable coins against the uh collateral

that was sitting in the protocol. Uh in that case risk ratings can be helpful because we did rate USR as a as a platform. Uh but it wasn't a risk of the strategy that they were running. It was rather like a hack. So it can be helpful to navigate but it's not solving the problem entirely. But still it's definitely better than status quo with no any indication of the risk. So the problem statement is that it's difficult to choose the right asset lending markets and vault for risk appetite of D5 users and we are going only going to have more and more headache because we see the evolution of stable coins. We started with USDC, USDT. Then we went through all the flavors of uh yield bearing stable synthetic stable coins,

ETNA USD, uh on USDY, hash node, USC. Right now we have over 200 stable coins and we are still living in this stable coin year. I'm going to expect only more uh the proliferation of real world assets. uh whereas we started with treasury bills, we expanded to private credit and now on daily basis I have discussions about trade finance, refractory, reinsurance. Naturally we see stocks being tokenized and in defi we also started with simple lending lend and borrow uh on a compound spark. Then we expanded into permissionless uh vaults with more for uler uh and chamino. Now there is over 500 vaults and yield strategies that users have to

navigate through. So the matrix is going to get only more and more complicated. So what you require is something simple for regular users if we go mainstream and to institutions that are not native to crypto. Something that is going to translate to them very easily the risk and the data associated with it uh from the market. That's a screenshot I took last week uh from Larink. Uh so Black Rock is uh betting billions that tokenized funds will do for Wall Street what the internet did to mail. If those biggest financial institutions and hedge funds say so it's going to happen. So it's not about if it's rather about when and at what velocity and this is what we want to help. Two weeks ago, I was in

Washington DC at DC blockchain summit where I witnessed uh the chairman of CC CEC sorry securities exchange commission and the chairman of CFTC speaking both at the same panel talking about the memorandum of understanding they signed and supporting the crypto expansion in America. This is the first time we witnessed that historically with Gary Gensler as a chair. It was very hard to do those kind of like crypto revolution. So all the pieces are putting uh it all uh together for uh the industry to properly take off. So Credora simplifies complex conra constructs into understandable information and provides simple uh to grasp risk rating. So you can think of it as a thermometer. If you

like it hot and uh sit on a chair that is uh potentially putting you on fire, you can go more towards the more risky asset. Naturally it's more exciting but it can come with a drawback. And then if you go on the safer side um you you're less exposed to risk but also probably the returns are uh more conservative. Um the scale when we were thinking about how to approach it we took the best aspects from the traditional market. So we mapped out the trfy scale from AAA double B uh C and so on. So a very uh difficult uh and very granular scale that evolved over time and made it way simpler between D to A+ and the main metric that we look into is something

that we call probability of significant loss. This is defining the cutoffs for specific ratings to be assigned and what comes into the rating are network risk. So the same two assets let's call it syrup USDC from Maple can get a different rating on very uh popular network like Ethereum and an obscure L2 or obscure L1 counterparty risk smart contract risk operational risk who is the custodian what is the legal setup oracle risk whether the vault has a hardcoded oracle fundamental oracle market oracle is it an oracle that is trusted in general in the ecosystem um and the key indicator as I mentioned is a PSL that measures likelihood of a market experiencing bad debt exceeding

1% of principal and bad debt essentially means that there is a position in a vault or lending market that liquidators are not incentivized to close so it's unhealthy um on credora the I believe key beneficiary in terms of compet competitors is that we rate not only asset but also markets and vaults hardly anyone nowadays in crypto just holds an asset like holds ETNA USD. No, people usually loop it. Go to Morpho, go to AA and try to create a leverage strategy.

So it it doesn't tell you the whole picture if you just rate an asset. You should also look at the protocol level. What's the liquidity related to a particular asset? H what is potential risk for liquidation and what are the paths? So in the asset methodology, we first look into the category. So looking at the asset is it a derivative token like for example liquid staking token is it a stable coin then if it's a stable coin is it non-yield bearing or yield bearing h is it an RWA so a tokenized asset like treasury bill private credit uh and so on then we implement uh modifications which is anchor probability of default so we look into audits how many are there like are

there from reputable parties where they're repeated where they're added when new code was released. Asset quality that is within a particular rapper. What's the custo risk? Is the custodian from the trusted s trusted um set of uh parties and if the uh strategy is actively trading. So it's way different when you just deposit assets into one strategy and it's over there time or if it's actively uh traded strategy like the one with stream finance was like where people were trying to do basis trade on on leverage.

Uh and then we have modifiers uh looking at reserve management. So things that happened over the past so recent incidents was there any DPEG was there problem with redemption uh what is the PG peg record? So an example where whether a stable coin lost pe for some time what's the governance strategy and then mixing all those aspects into all together we get the final probability of default that gives you a rating for a specific asset. So this is the first step we're doing then okay so imagine we have an asset for example syrup USDC from maple I can tell you it has an a minus rating and then we can have USDC the circle stable coin that has a rating. So you have those two asset and then we can have a look at the market on

morpho for example looping of syrup USDC to USDC. So in that particular case we were going to look at uh loan to value thresholds and also what is the bonus for liquidators should the position be underwater to liquidate that position close it as fast as possible and keep the market healthy and oracle characterization as I mentioned is it the market oracle is it hardcoded we look into allocations distribution so determining how aggressive uh specific active uh user borrow positions are so they're very close to liquidation threshold. So if liquidation threshold is 70%, are they close to it like for example 65% or are they like only on the 20%. Um what's the rebalance profile? So

uh we model borrower behavior from the past. So we look if they got closer to liquidation thresholds. So imagine they started with 20% and they get closer to 60 whether they reallocated again into that position to make their um health factor uh improve. What's the collateral asset rating? So what I mentioned for example syrup USDC whether it's a good rating or not necessarily like the favorable one uh rewards distribution. So uh we model what's the returns distribution considering the volatility and extreme tile events and par liquidity data. So imagine there is a liquidation uh between the pair of syrup USDC USDC. Then we look what's the situation on dexes and Texas in order to

liquidate that particular pair. For example with RWAS this is something very common right now that liquidity is almost non-existent. So if there's a larger position to be liquidated, no one exactly knows how it's going to be performed. For price simulations, we implement something that's called Monte Carlo simulation. You can imagine uh rolling a dice uh many many times and then just looking what are going to be uh the outcome of um random events and then we plot the pair volatility but also tail events. So we see what's going to happen if Bitcoin for example drops 30% or like a specific asset is going to appreciate 80%. We also saw those those kind of uh moments in example when compound comp

was listed on AB bit I believe it appreciated like 70% within a couple of hours which was naturally an anomaly but as an oracle on the redstone side we had to keep uh posting that kind of price speed uh on chain uh and also what's the uh collateral default probability and then we plot uh within the simulations uh the probabilities that a certain position is going to breach the uh loan to value trigger and get liquidated. This year is a year of vaults and a vault is simply an onchain equivalent of an investment vehicle that provides exposure to a defined strategy.

Investors deposit an asset of a stable coin and receive a receipt token that represent prata claim on the pool similar to fund units. So it's very similar to how black rock and many other people operate. So you can imagine there is a curator there is a vault and the curator allocates the money that people put into into specific stable coins into specific yield strategies into specific markets and then we are able to uh rate the whole vault based on uh all the uh comp composition of these strategies that are put inside and managed actively.

It's not just theory, it's life. So uh we've got over uh $10 billion rated uh over 125 vaults on morpho and the coverage of 90%. uh over 555 assets rated uh 25 250 mor markets and 100% coverage of spark which is one of the largest landing market in the space and in practice I mean I talk about ratings okay nice there is the letter DA and whatever but if you go to moro app right now app.mmorpho.com I believe or IO h and you check out the vaults over there you can go to specific ones like gauntlet USDC stakehouse USDT Um, okay. That's supposed to be a video, but

it works as a as a picture. So, unfortunately, video is not going to play. But if you go to a specific vault on morph right now, and you see an API of, for example, 3%, you can go to the risk section and see exactly what's going to be the rating allocated to that particular vault. And what we are working with on with morph right now is to display the rating over here so that you have both the APY and the risk rating next to each other and then you can try to calculate new metrics such as for example risk adjusted return right given the uh rating and how it's been behaving in the past and also the APY you can uh create an allocation that uh works for you the best. Uh another example is also spark unfortunately it's

also a video that's not playing. So essentially if you go to sp up.spark com right now and you check out the spark saving product USDC is yielding 3.75 APY. You check out the risk rating tab you're going to see that majority of the strategies over here got a minus rating which in general is a very good one like a rating means that something is extremely extremely safe. H a minus is still like a very attractive one. uh BE+ is starting to deteriorate uh like going towards the more risky section and an asset that recently got A+ uh which is like very unusual is the stake if so we looked into LIDO and the

history of the asset itself how it's being operated and we created a campaign with them to showcase that they are asset that are uh very well with a very good uh track history and low risk associated with the investment strategy. Okay, so I can talk and talk, but what's the difference with the rating? Well, so actually we looked into the uh data of the performance of specific uh vaults that were rated and unrated. So those are two vaults over Morpho that uh offered the same exposure to asset by two different curators. And the moment they got a rating, you can see that one the TVL started to grow quicker. But even more important in the moment of

drawdown we see that less people exit their position. So there are less afraid that something bad is going to happen there and they're also allocating less uh I would call them tourists or mercenary capital that just go after the airdrop hunting and I really appreciate that you all came to the uh presentations. So I'm also happy to share an alpha uh that we are launching next week. So those ratings so far have been uh present on the protocol level. So you could go to specific website. But what's the going to be the most useful for individual users is actually a platform that aggregates all the positions that you can allocate into defi together with ratings and the APY associated. So uh in

the credora platform that we are releasing uh next week you're going to see vaults markets curators and assets and you can compare them. So you can try to even um simulate a portfolio if you're deciding to uh allocate into specific uh areas and then compare them before you're actually executing the strategy. Uh we have wallet rating uh feature where you can just plug into your wallet. We are going to analyze the positions that you have active on the wallet and uh get an information about the uh rating of of your particular holdings. And the most important feature after last weekend is going to be the last one over here alerts. So we are going to expose an API that will inform users if a rating of a specific asset is

going to deteriorate because it's not just static. You just put it out there. It's being monitored constantly. So with resolve we were able to um spot that something was happening with the hack of only after a few minutes it happened. So it would given uh many of the users the possibility to exit quicker because the resolve team specifically took four hours to do a public announcement and identify uh the issue. What comes next? So builders are going to create new products on top of those ratings. For example, insurance programmatic strategic strategies. uh AA agents will uh utilize the ratings to define the strategies they're going to allocate

into and new lending protocol designs. Thank you very much and if anyone wants to chat more, I'm going to be outside. Cheers.