Grant Proposal: Credit Scoring Integration

Introduction

RociFi is a protocol building two key pillars of Web3’s future:

(1) on-chain credit scores

(2) blockchain-native under-collateralized retail lending.

RociFi credit scoring scores users according to their likelihood of default on an under-collateralized loan. The scale is 1 to 10 with 1 being the best score and 10 being the worst. To-date, we have minted over 22,000 credit scores, 1200 under-collateralized loans, and 2000 capital efficient loans (higher LTV).

RociFi credit scores could allow Moonwell to offer higher LTVs to certain borrowers – generating greater revenues – and analyze user-level risk which can be married with the macro, e.g. Gauntlet, for a holistic viewpoint of protocol solvency risk.

The best part is that the NFCS API is open so the protocol can query scores in real-time of any NFCS (credit score) holder for risk monitoring or better term offers. The specific implementation of scores will ultimately be voted on by governance.

Motivation

Below we offer 2 ‘proof of concept’ benefits of the proposal to Moonwell:

(1) Credit scores for a sample Moonwell users, including sample metrics, can be found here. TL;DR: a lot of opportunity to segment higher value users for better loan terms which could increase loan demand.

(2) Comparison of the revenue Aave and Compound could have generated using RociFi credit scores here. TL;DR: revenue increases are large with a negligible increase in liquidation risk.

Call To Action

We propose that Moonwell integrate RociFi credit scores into the protocol as to unlock additional capital efficiency and risk assessment while differentiating the product from other competitors in the space.

Furthermore, it offers additional utility to Moonwell in the future, e.g. credit/trust-based voting incentives and token reward segmentation.

Proposal Details and Milestones

RociFi stands ready to support the integration as needed, but the existing open API makes the process straightforward and customizable based on governance needs.

If additional customization is required, the timeline, milestones, and grant size request will increase.

We’ll await feedback from the community and protocol on specific integration needs or desires before putting a dollar amount to the Grant request.

Conclusion
We’re open to feedback and questions. Thank you!

Thank you for posting. This grant proposal would benefit from more specific implementation details around how you would implement a credit scoring mechanism as an improvement to the Moonwell protocol. Here are some suggestions to help improve it:

  • Please share analysis based on the current on-chain data around how credit scores could be used to improve capital efficiency of Moonwell Apollo and Moonwell Artemis
  • This could include data supporting the theory that more credit-worthy borrowers are less likely to get liquidated, which might justify increasing their collateral value or LTV.
  • The data should demonstrate how implementing a credit scoring mechanism could improve capital efficiency without meaningfully increasing risk to the protocol.

The proposal is also lacking implementation details:

  • How would the credit scoring be implemented, from a technical standpoint?
  • What changes would be proposed and made to the Comptroller and other smart contracts to facilitate this new feature?
  • How would those changes be audited prior to an on-chain proposal that implements them?
  • How would you take community feedback and use it to improve the proposal?

I welcome further discussion on this. An on-chain proposal could upgrade smart contract implementations and directly transfer WELL to the proposal author if passed, so there is a mechanism to receive grant funding for these improvements, but I think a broader discussion and more specific implementation details would benefit the community prior to any smart contract changes being made.

Best,
Luke

1 Like

Thanks for the feedback @Luke!

Re: *>> Please share analysis based on the current on-chain data around how credit scores could be used to improve capital efficiency of Moonwell Apollo and Moonwell Artemis*

*This could include data supporting the theory that more credit-worthy borrowers are less likely to get liquidated, which might justify increasing their collateral value or LTV.*

*The data should demonstrate how implementing a credit scoring mechanism could improve capital efficiency without meaningfully increasing risk to the protocol.*

I would say 90% of the information you mentioned is covered in the 2 links above.

To explain, the second link offers all the information you mentioned, but showcases Aave and Compound given they have a much larger dataset to compare the risk-adjusted capital efficiency over a longer time period.

The first link shows a sample of current Moonwell user addresses which have been scored by us, which can then be used as a frame of reference for potential capital efficiency enhancements by offering higher LTVs to the best rated users – credit scores 1 to 4.

For reference, based on the sample analyzed, 38% of Moonwell borrowers could qualify for these higher LTVs. See the distribution below.

However, I can appreciate that the community might like to see an equivalent revenue simulation on Moonwell data before making a decision. If so, we can provide those results in a few weeks. Keep in mind that the time series data being limited might make the analysis less comprehensive compared to what we attached for Aave and Compound. Let us know and we’ll be happy to oblige.

Re >> *How would the credit scoring be implemented, from a technical standpoint?*

*What changes would be proposed and made to the Comptroller and other smart contracts to facilitate this new feature?*

*How would those changes be audited prior to an on-chain proposal that implements them?*

*How would you*

RociFi can provide a framework for how these changes could be integrated Moonwell’s smart contracts for a dynamic user experience. But, we’d rely on the Moonwell community developers and contributors to perform the actual integration of this new feature into the contracts; along with the subsequent auditing before pushing to PROD.

Given this portion would be driven largely by community developers, we will take feedback quite seriously and provide as much help in the specification process as possible to enable a smooth integration process as best as possible.

Happy to discuss in more detail with community developers in DM. Thanks!