Capital Efficiency in DeFi - Sishir Varghese

foreign So today we're going to be talking about financial reputation in defy uh I think we all know that D5 has shown incredible product Market fit uh it's here to stay and we're quite literally talking from a defy stage at one of the biggest crypto conferences in the world tons of great ideas have come out of this we've seen powerful disintermediating protocols and obviously things have slowed down with the market turmoil uh things with the centralized exchanges and trading funds have left a bad taste in Trad five's mouth but all along D5 has kind of chugged along quite smoothly we do believe that there is a missing layer in defy which is financial reputation and I'd like to outline here today why it's important what we're building and also start working with the community on building some of these use cases with this new missing layer so what we'll go over today is the question of what's the point of a web 3 credit score uh we'll outline a bit of what goes into building that score and what how we did that and how we built the macro score some of our Outlook into how it's used and since an on-chain credit score is a New Concept I hope we can come away with from this talk with interesting questions and some use cases to work on with the community and Developers so the first question is why a web 3 credit score uh what's the point isn't everything pseudonymous uh don't we want to build a better Financial system instead of copying and pasting over a broken infrastructure uh couldn't we just kyc everybody and kind of uh basically forego all the benefits of a global consensus and a distributed Ledger can we bridge and create attestations I think what we want to do is lean more into the on-chain side in the D5 side and stop relying on kind of fintech use cases and build this kind of take advantage of this kind of Paradigm Shift we've seen over the last 10 15 years in addition to that there currently is no standardization for financial reputation when you're traveling across countries or regions when you move or travel you can't necessarily carry your financial reputation with you if you're in a country uh that doesn't even have a credit score then you actually are kind of empty there if you're in North America or in some countries where you have a credit score and on-chain data or Unchained Financial history becomes supplemental but in parts of South America Africa southeast Asia we're seeing a lot of adoption of web3 but these people actually have never even heard of a credit score so there's an opportunity for us not to be supplemental but to become Primary in those regions and it becomes really interesting and so if credit is the driver of economies then we should actually enable this infrastructure for the marginalized and then there's another issue is if you already have a credit score the current system is actually monopolized by three bureaus and FICO and just a few months ago last year Equifax got hit with a class action lawsuit uh because it was a glitch in their score uh we actually don't know what the glitch was but they they get thousands and thousands of people the wrong credit score people got rejected from these loans and we don't know why and it's very likely that this is probably not the first time this has happened uh but that's actually the problem we actually have no idea and so the stronger the hold that these bureaus have and the governments have on your financial reputation the closer and closer we inch towards this idea of what's happening in the east of this type of like Chinese social credit scoring system where you can get banned from booking hotels you can get banned from uh Transportation it's on your dating profiles and so the idea is that you know what happens if there's a glitch in those kinds of things and it becomes a huge problem in this kind of black box that's there uh when Satoshi launched Bitcoin he was up against Banks and governments for this idea of transferring value or basically moving money without asking for permission Financial reputation is actually in a very similar bucket but instead of going up against Banks and governments we're actually up against bureaus and credit bureaus and governments and so we need to actually disintermediate The Gatekeepers in this system okay so C5 versus D5 is very much a Hot Topic right now and we're showing it here as like kind of juxtaposing Industries but while the infrastructure is significantly different and D5 is obviously far superior uh the end user is actually the same it's us it's everybody here we're actually using both rails but most people who are entering the space don't necessarily care what's on the back end and so they'll continue to use this but actually everything you do in D5 is actually completely neglected everything you do on chain is disregarded it can't be part of your larger Financial history and we need to start blurring that line and so part of our mission is actually to start kind of blurring that line between defy and C5 so you can start leveraging your on-chain history in the real world and vice conversa so having an on-chain credit score can give people opportunities to include their web 3 activity as part of their real world financial history and also give people credit risk infrastructure who have never had it before so for example when you apply for a loan or a mortgage at a bank right now they actually ask you to liquidate all of your assets in order for it to be considered real what we want to do is actually start to enable these Banks to leverage that information start including it in your applications for loans start applying it as regular underwriting mechanisms and supplementing your existing applications no one wants to sell their crypto at the bottom of a bear Market no one wants to trigger taxable events and we want to provide this data to these institutions to make it connected back into these real world applications so we talked about the fractured state of credit risk infrastructure and how we also want to blur that line between C5 and D Phi and be able to leverage your on-chain history but within defy itself there's actually a ton of room for improvement within Capital efficiency right now loans in D5 are completely flat if you go to any protocol everybody in this room will see the exact same rates regardless if you're credit worthy if you're a complete Degen it actually doesn't matter so while there is a benefit of D5 being open accessible and everybody being treated equally it's actually in reality is that everyone is actually treated equally bad we all get over collateralized Capital inefficient terms and the idea is that being flat across the board doesn't necessarily benefit a lot of people so we want to create some sort of incentivized mechanism where people are incentivized to be credit worthy start repaying their loans and then be give and then offer preferential rates to those people uh and while it makes sense since it's pseudonymous lending these lenders are saying if you don't want to tell me who you are you have to give me more money than I can give you there's still a lot of room for improvement in capital efficiency with collateral rates with interest rates and things of that nature we're not actually saying we could solve for lack of recourse because just because I give you a three-digit number it doesn't necessarily mean that you're not going to run away with my money but we can become like the first stepping stone towards that larger journey and also just bring a ton of capital efficiency on the way um and a company called Sofi actually in traffic it's a large financial institution actually started in a similar way where they looked at students that were graduating from Stanford and said that their student loan rates were actually the same across the board with everyone in the country these kids were graduating with six figure salaries they had way lower probability of defaults and yet their lending terms and interest rates are actually the same for everybody in the country there's a similar opportunity here in the on-chain world where we could start to offer these more credit worthy individuals more preferential rates if we start to analyze their on-chain transactions and start to find these types of individuals and create this type of incentivization for them we want to start building this infrastructure so lending platforms lending protocols can start to leverage this data and then start offering these services to these users so all those reasons that I've given you is kind of why we need credit risk infrastructure now we want to kind of get into what the score is itself uh our macro score which macro which stands for multi-asset credit risk Oracle covers two and a half million each borrower events uh 1.9 billion each transactions over seven and a half years of each data and over four years of D5 data all of that data has been structured into 138 plus model features that go in and come together to Output this three-digit credit score uh you can also see on the screenshot or consumer app that we just launched a few months ago you can go right now to spectral.finance check your score with connect your wallet you could see your outstanding loans you could see your health factors on these loans see if you need to top up your loans if your health factors are low soon we'll actually have third parties basically pitching to you and giving you specific interest rates specific lending terms based off of your financial history and your on-chain credit score on the app itself we so far we've computed over a hundred and ten thousand scores so far for individuals just in the past few months we're also overhauling our back end so that scores can soon be delivered in less than a second this is really important just in terms of delivering real-time data and real-time risk insights so we have to parse an individual wallets transactions for all these billions of data points that I mentioned and then analyze and calculate all these different attributes which then output the score the hard part actually is that getting all these data points is not so easy while blockchain data is raw and openly accessible there is definitely an emphasis on the raw part and so which then brings me to my next point of on-chain data sources we spent an abundant amount of time just sourcing clean data and this is from D5 protocols uh General ethereum transactions uh and we then we had to create all these proprietary features which then assess wallet transactions and output the score so how do we get all this data right now we use a ton of different apis and data sources since the breadth of our data that we ingest is quite wide and varied we use platforms such as the graph Alchemy etherscan coin gecko all these different apis we also run our own ethereum node we run subgraphs on the graph and then we started with Ave and compound data since the majority of lending borrowing data was there then we expanded quickly into a lot of other protocols nft trading data dex's evm compatible chains Etc polygon lending and borrowing is actually multiple times larger than Ethan avax combined in general though this is quite an industry-wide problem data is scattered there's a lot a lot of companies building right now to kind of structure and clean this data but we've did a lot of this and from the perspective of credit risk insights then the next question usually is like is there enough data at all we've covered and back tested on over two and a half million loans uh with over 200 000 borrowers that have been liquidated over 130 000 times uh actually we have a blog post on our blog right now if you go to blog.spectral.finance uh you can actually see all of our back testing we've taken every single loan that's ever happened on D5 uh did a deep dive into how we built this score and then have basically back tested on seeing how our score operates and functions and performs on this kind of historical D5 data so we do know that there is enough data but we've taken the effort to clean and process all this data and then we created all these hundred plus data streams that feed into the model and then output the score one thing we learned along the way during this process is that this three-digit number can give you a high level understanding of what this credit score can tell you in terms of risk assessment but this granular level in between of all these kind of data packs and attributes is actually really insightful granular information that we want to provide to Partners and to the community to start building out more our customized solutions for different types of use cases so we need something which bridges that data from all these different available data sources and then translates them into actionable risk insights real-time customizable data streams that Partners can pick and choose from which attributes they want and then plug in these on-chain risk insights for lots of different use cases so anything from Wallet transaction analysis credit mix D5 Health factors as I mentioned nft data Dex data and each of these groups actually have another subcategories of much more granular data themselves so maybe you're an nft lending platform that wants to know nft lending data wealth and risk factors plus maybe ethaned asset balance over time you'll be able to pick and choose which data streams you want to use for your protocol and then supplement that in your underwriting mechanism and then provide more preferential rates for these users on your platform so now that there's a heap of useful structured data at your fingertips you know what do you do with it what are the use cases how can this be helpful for you and how what are the new applications that you could actually build that weren't actually possible before well it depends let's say you're a web 2 fintech company and you know a lot of your users are interacting in web3 uh it'll sell a lot of these retail based apps such as Kasha venmo PayPal Etc start to roll out services to more normies and normal people around the world that means more and more people will have web3 Financial history as part of their broader Financial history but right now all of those transactions are actually completely neglected and disregarded we want to be able to bridge that back and supplement it for a lot of the traditional use cases that you would be able to do a lot of these companies and platforms can increase their addressable Market if they see a lot of their users are actually having larger balances that they didn't know about they can enable new product lines for them increase their addressable markets so they have a broader kind of a platform for their users and then in addition if you're a credit bureau in a traditional credit score then you could actually supplement an existing credit score Experian has a program called Experian boost which actually takes in alternative data sets so something like are you paying your Netflix bill on time are you paying your rent on time are you paying your utility bill on time what we think is that web3 financial data is just another alternative data set and so you should be able to plug in another alternative data set to supplement your FICO score supplement your existing credit history and then boost your score in that way by leveraging your on-chain history in the web 3 world if you're a web 3 protocol there's opportunities to create entirely new applications you can make new lending platforms that are more Capital efficient based off of your past history or perhaps maybe start credit scoring Dows Dow ratings are kind of like credit ratings versus credit scores in the real world credit scores are for individuals credit ratings is what you hear about when institutions or governments start to have ratings you see AAA ratings B ratings Etc but the equivalent of that in terms of like Collective credit worthiness if you think about what is the equivalent of that on chain it's actually Dows so how do you actually credit score a dow what does that actually mean you know do you take the governance participants of the Dow everyone who holds a token do you only talk to take the officers of the Dow or maybe the multi-sick holders a dow-based credit score credit rating should allow a dow to leverage their treasury their financial history and then offer them Financial Services to take out loans via Collective credit worthiness there is no specific way to do this yet because actually it just doesn't exist but this is ongoing research that we're doing and this is one of the examples that could be enabled with on-chain credit risk so in general on-chain credit risk infrastructure can enable an entirely new application layer to emerge within defy that wasn't possible before and then bring new capital efficiency to the space while Bridging the worlds of C5 and defy right now you can go to our website access our API for free play around with it or go to our app and check your score at spectral.

finance you know we've done a lot of work to clean up and open up this raw blockchain data you can check out our documentation let us know if you have any questions if you scan the QR code you could actually access the API right now for free we look forward to working with the community coming up with new use cases within D5 for this new application layer you come by our booth check us out let us know if you have any questions thank you I think we have a few minutes right now like two or three minutes for questions if anybody has any uh so the question was what is our Revenue model so we provided API plus an oracle call so right now we are doing a kind of a per call basis pretty much traditional kind of SAS Oracle model cool there's no other questions thank you everybody foreign