Increased Capital Efficiency Idea

RociFi is a protocol building the two key pillars of Web3’s future – on-chain credit scores and under-collateralized lending. The credit scoring is powered by our non-fungible credit score (NFCS), which ranks users according to their likelihood of repaying an under-collateralized loan. The scale is 1 to 10 with 1 being the best score and 10 being the worst.

For reference, NFCS is built using RociFi protocol repayment data and other DeFi protocols; including Aave and Compound. The depth and breadth of data allows the NFCS to be the clear leader in on-chain credit scores which has applicabilities to other lending protocols.

In particular, lending protocols like Aave and Compound could use RociFi NFCS to achieve better capital efficiency by offering higher LTVs to higher scored borrowers – generating greater revenues without a demonstrable increase in insolvency risk – akin to a micro-level Gauntlet.

The best part is that the NFCS API is open to any developer or protocol to call for real-time scores of any NFCS holder, thus they could integrate this score based on each protocol’s individual needs.

Revenue Increase
Below we provide comparisons of the revenue Aave and Compound generated historically and what they would have generated using RociFi NFCS scores for higher capital efficiency. The following LTVs used for revenue simulations are per score:

*As of writing (2022-10-04). RociFi does not currently offer lending of non-stable collateral. LTV’s provided are representative of what protocol could choose to provide to their users. Simulations are estimates which may differ from actual results.

Using a simulation period from May 2019 to Aug 2022, we compare the estimated total and monthly gross revenue Compound would have earned with and without RociFi NFCS; assuming an APR of 8%.

Using RociFi NFCS would have generated revenue of ~$493,615,642, while without only earned ~$208,607,361. By being able to offer higher LTVs, Compound would have generated an additional $285,008,281 in value (net of liquidations) with a negligible increase in insolvency risk.

Tokens Considered : [‘AAVE’, ‘BAT’, ‘COMP’, ‘DAI’, ‘ETH’, ‘FEI’, ‘LINK’, ‘MKR’, ‘SUSHI’, ‘TUSD’, ‘UNI’, ‘USDC’, ‘USDP’, ‘USDT’, ‘WBTC’, ‘YFI’, ‘ZRX’]. The base LTV used for simulating Compound revenue for each token was 80%. Revenue simulations assume no default

By scoring borrowers using the NFCS, Compound is able to achieve better capital efficiency by not penalizing responsible borrowers, enabling higher loan allocation to this subgroup, i.e. give good borrowers more loans and less to worse borrowers. This simple heuristic is illustrated in the distribution of loan amount by score chart below.

Using RociFi NFCS, despite higher LTVs, a heavy concentration of Compound loan volume went to the very best borrowers, i.e. credit scores 1-5, comprising 99% of total loan volume issued. Thus, increased revenue with minimal increase in insolvency risk.

Integrating onchain credit scores from RociFi NFCS allows existing lending protocols to offer more competitive LTVs, thus higher risk-adjusted revenues. Additional positive byproducts of adopting RociFi NFCS by protocols are (1) incremental revenue can be passed back to the community and (2) better product attractiveness via higher LTV for borrowers with lower rates, and higher supply rates to lenders.


While I don’t see Compound moving into under-collateralized lending anytime soon, if ever, I do think it’s important that folks are exploring this idea and welcome your reaching out to gauge interest from this community.

How does the scoring system work? If it’s proprietary, you’re probably better off spinning up and running your own lending markets on it (which it looks like you’re already doing on Polygon? great!). I think open-sourcing the credit system so that users can understand and model the risks would be essential for this particular community.

I took a quick look at the protocol docs and learned that the NFCS scores are generated off-chain and fed to the protocol via Chainlink. This introduces some additional oracle risk which the community would need to examine more closely, e.g. what happens if an issue with the oracle causes a borrower with a high score to suddenly no longer qualify for their high LTV and experiences a sudden liquidation. But that seems minor in comparison to the general risks of undercollateralized borrowing.

Hey @allthecolors!

Thanks for the response. Let me answer of few of your comments below.

For clarification, we’re not suggesting that Compound get into the under-collateralized lending business. We are saying that there is quality information that Compound could be using to both improve capital efficiency, i.e. offering higher LTV loans (not under-collateralized) for greater revenue, and micro-level risk assessment of individual users on the platform for liquidation risk mitigation.

Re: Scoring Openness
In a few of our blogs, Managing Default Risk. “The essence of investment management… | by RociFi - Under-Collateralized DeFi Credit Protocol | Medium, we provide an intuitive overview of how we score credit risk. Unfortunately, we do not provide specifics given bad actors could easily game the system. However, our goal is to open source the scoring process at a later date once appropriate.

Re: NFCS scores are generated off-chain and fed to the protocol via Chainlink
Apologies but this document is outdated :disappointed: There is no Chainlink dependency any longer because of exactly this issue. Scores are simply pushed into our smart contract and verified to make sure the score is coming from our certified API.

To implement, Compound can simply query our API and use it as they deem fit for capital efficiency and risk analysis, RociFi launches NFCS API. RociFi’s NFCS (Non-Fungible Credit… | by RociFi - Under-Collateralized DeFi Credit Protocol | Medium.

Thanks for your comments!

1 Like