Anyone who tries borrowing in DeFi for the first time is aghast to know that the collateral value for the loan is always equal to or more than the borrowed value.

It kind of defeats the purpose of why most people want to take a loan, the fact that they don’t have enough liquidity in the first place.

Besides, that over-collateralization leads to two major problems at the protocol level:

• Most protocols suffer from a low utilization ratio particularly if the collateral ratio or the borrowing APR is too high.
• Protocols don't segment the users. An institution with negligible credit risk and a degen that gets liquidated left, right, and center pays the same APR and collateral value.

Now despite these limitations, fully collateralized loans are still super useful for traders and institutions. DeFi lending is only second to Dexes when it comes to total TVL, representing 27.3% of the TVL in the ecosystem.

However, as the space matures and DeFi reaches the masses there would be a massive need for under-collateralized loans and as of today, the ecosystem lacks any framework for it.

Enter Roci.Fi

Roci.Fi has created a fully composable-on-chain Nonfungible credit Score system( NFCS) that uses a complex algorithm to evaluate a user's credit risk, fraud risk, reputation risk, and assign a credit score. Users are rated from 1-10 with 1 as a hallmark of being highly creditworthy. The score is assigned in form of a non-transferable ERC-1155 token to the address. The protocol has already deployed under-collateralized borrowing, with the collateral ratio set at 71% for highly credit-worthy users (1-3). Another use case of NFCS is that it has a built-in fraud detection mechanism and only provides NFCS to whitelisted addresses. This can help protocols keep bad actors at bay. Now you might ask what if the users make a run on their loans?

The answer to it is social recourse, where borrowers agree to the T&C of disclosure of their personal information in case of a default. The defaulted users would also be firewalled from Roci. Fi and other protocols that deployed NFCS.

The growing importance of web3 reputation and identity would make social recourse a potent method to maintain a low default rate.

Traction

NFCS went live during the 3rd week of June.In just about three months.

• Over 20,121 NFCS have been minted - An avg of 223 per day.
• Current TVL of $106,581- Polygon top 13 lending protocol.
• 3000 loans have been created - An avg of 33 per day
• $95,747 loaned out.
• Institutions and DAOs constitute around 35% of total outstanding loan value with high to mid-credit-worthy users accounting for up to 88.82%.
• 2:1 current Deposit to withdraw value ratio - Good stickiness
• The bulk of the debt is being provided by NFCS scores 10 users- Protocol is better off this way than vice versa.

*No liquidity mining program yet,TGE event is only in 2023. Currently, loans are offered only in USDC with WETH as collateral. V1 is currently available only on Polygon.

(click the links for the data sources).

NFCS Architecture

courtesy: roci.fi

NFCS uses chainlink as an oracle to feed the NFCS scores as all the scores are calculated using an ML model while the fraud verification is done using Graph Theory and ML model

Here´s how behind the scenes of an NFCS mint call look like:

User address input -> verification -> credit risk oracle(CRO) -> off-chain ML model -> (generate LTV and user´s credit score) -> Relay it back to CRO -> Mint the NFT with score.

Security:

Protocol code is audited by two leading auditors- Chainsulting and Certik. Contracts are equipped with circuit breakers that allow protocol admin or DAO to pause it and disable any money transfers.

Future states

A vertiginous growth awaits RociFi.

• Mass Undercollateralisation can unlock the TAM of DeFi lending by at least 10X even by a conservative estimate. Given that the traditional lending market is worth $7.5 trillion and the Defi lending market is a mere $14.74 billion
• Being the early pioneer, RociFi´s NFCS can be the go-to credit score system adopted by other lending protocols. RociFi can easily set a small mint fee for every score mint. Eg: so far 4.8 million addresses have interacted with DeFi protocols, applying the last year's growth rate of 45% for the next 3 years, by 2025, 14.63 million addresses would have interacted with DeFi. Now assuming a SOM of 5% and that a user on average submits two addresses for credit score. The protocol can earn $1.82 million in just royalty fees over three years when the fee is set at $5 per mint. The protocol can again charge a nominal fee($0.05-0.20) every time the user calls for a revaluation of their credit score.
• Launch of the Roci and veROCI token in 2023 to move towards a DAO model. Enabling the staking mechanism where token holders could stake and get a share of the minting fees and the protocol´s lending platform revenue.
• Launching incentive programs and expanding the protocol to other L1s to grow the overall TVL. Armed with a unique product offering and high TAM, RociFi can comfortably attain the TVL of the top lending protocols (in 1-2 years) which would be in the range of $200-300 million.

Challenges

Besides the pessimistic macro environment. There are a few challenges at the protocol level that RociFi will have to solve.

• Low Average loan size: Overall avg loan size on the protocol is only $31. While avg for users with NFCS 1 -3 is $1786 This indicates that the protocol needs to onboard more high-value users. Only approximately 7% of total users have a credit score of 5 or lower than that.
• Address duping: Users are free to select the addresses that they would like to link for the NFCS score. They might not choose the addresses that they got liquidated in past.
• Insolvency Flurry: Though highly unlikely, a huge number of undercollateralized borrowers fail to pay back their loans. The protocol might not have enough in the treasury to cover the loans which can create a cascading effect of withdrawals from the protocol resulting in a TVL crash.

Team

With Konstantin Zagaynov , Pavel Filippov and Chris Bookins at the helm, the team has leadership experience at prominent fintechs like Bitwala, and Kreditech and has a deep understanding of both TradFi and DeFi lending spaces. As proof of their credentials, they have managed to secure seed funding of $2.7 million from Nexo, ArringtonXRP, and Signum capital. The stage is all set for the team to raise series A and take the protocol to the next level while also solving the challenges mentioned above.

Final Thoughts

NFCS is a promising step towards attaining the goal of under-collateralized loans. Roci.Fi´s algorithm has set high standards for the credit score and currently, only a handful of addresses are able to attain a score of 1-2, almost all of those consist of institutions and DAOs.

Given that the on-chain identity and reputation adoption is still in its infancy , RociFi cannot be too aggressive with its scoring approach or collateral ratio. The growth of NFCS would go hand in hand with the growth of on-chain identity.

The more personal information that the user has on-chain the greater would be the efficacy of NFCS.

While NFCS is one way to solve the problem of fully collateralized loans, I expect protocols to also come up with alternatives like using a payment stream mechanism by superfluid.finance mechanism to provide undercollatrized loans.

Stay tuned for my next article on payment streams. Peace✌️