RociFi Credit Scores for Capital Efficiency — Moonwell

Moonwell can increase revenue by up to 31.21% without increasing liquidation risk.

RociFi’s NFCS seeks to rank potential borrowers based on the likelihood of repayment if they were extended a loan. Using a plethora of historical data collected from major DeFi protocols from 2018–01–01 up to 2022–10–05, we have included Moonwell data** into these transactions, and have been able to refit and estimate the risk of default for borrowers extended over-collateralized loans.

Using 2022–08–05 to 2022–10–05 as an evaluation period, we define “good” accounts as those that did not default on their loans due in this period, and “bad” accounts as those that did. To better understand if more creditworthy borrowers are less likely to get liquidated, we can look at the distribution “Actual Goods” vs. Actual Bads” for the broader ecosystem first. For this analysis we exclude any borrowers who have only ever taken 1 DeFi loan, regardless of whether they have paid it back or not.

As we can see roughly 73% of “bad borrowers” get out on score 10 with 10% on score 9, meaning roughly almost 83% of bad borrowers are at least score 9. We can also see that as the score improves (moves towards score 1), the number of bad borrowers decreases to 0.

To understand the impact more specifically for Moonwell, we can look at the distribution for those borrowers who used Moonwell. Again, any borrowers who have only ever taken 1 loan ever are excluded from this analysis.

As we expect, the majority of bad debt exists on scores 9 and 10, both encompassing about 80% of bad borrowers. As the score improves, risk generally decreases.

The bad loan rates for the broader ecosystem and Moonwell are as follows

As we can see, utilizing the RociFi NFCS would allow Moonwell to reduce the risk of default at higher LTVs by limiting those LTVs to borrows with more creditworthy scores.

In order to estimate how much capital efficiency would be improved, we look at the history of Moonwell lending transactions from the Messari Subgraph, and run revenue simulations based on the lending transaction history derived from the subgraph.

The RociFi NFCS model was retrained and refitted using data collected from all covered protocols including Moonwell. An LTV curve by score is proposed below based on an assumption around liquidations slippage, implied default rates, and asset standard deviations. Furthermore, the LTVs are intentionally aggressive to generate the most conservative simulation results, i.e. maximum likelihood of bad debt from liquidation losses.

For stable collateral assets (USDC / mUSDC)

For volatile collateral assets

The formula for picking the LTV curve here is:

The idea here being assuming that we don’t want LTV’s to exceed 100%, using a 5% margin of error, we scale back the LTV by the Expected Default Rate, and 2x the daily standard deviation of price returns. The scaling factor was selected in each instance so that whenever volatile collateral is involved, in the worst case we assign an LTV of 60%, which is consistent with the current maximum LTV on Moonwell. Given default risks and price volatility, we believe this LTV curve conservatively gives sufficient runway for fast price movements that may cause problems for liquidators.

Interest Rates

Following these LTVs we simulate lending revenue’s using approximate interest rates based on what was observed on Moonwell net of rewards, those APRs used for simulation are as follows

Using the interest rates above and proposed LTVs we align them onto a dataset of Moonwell’s borrowers by score, and estimate the hypothetical revenue that could be generated by utilizing the proposed LTVs.

The simulation procedure is performed as following

To do this, we employ the following procedure

1. Isolate all borrow, repay and liquidate transactions
2. For transaction, align the appropriate score for that address
3. For each transaction, using the score computed for that address, align the proposed LTV based on score & token
4. For each transaction, using the token, align the interest rate
5. Using the ratio of the proposed LTV and Moonwell’s LTV, lever repay and borrow transaction sizes by this ratio to compute the theoretical increase in size that would be produced by the affects of leverage
6. For each address, by each token, by each date, compute the net outstanding loan owed both using the original LTV and proposed LTVs
7. Group these by address, token, and date, and compute cumulative sum of the outstanding loan by day
8. Estimate the daily revenue by scaling each day’s outstanding amount by the daily interest rate for that token. If the net outstanding on a particular day is negative, we treat this whole quantity as interest income
9. Compute the incremental revenue as the difference between the original revenue estimate and the revenue estimate produced using the enhanced LTVs

Simulation Results

We estimate if Moonwell had used the LTVs driven by RociFi’s NFCS, it could have generated $12,816,386.48 in revenue, annualized.

For comparison, we also look at the revenue we estimate Moonwell generated using the current 60% LTV and the interest rates proposed in this document. Since June, we estimate this would have been $9,809,416.29, annualized.

Obviously, increasing the leverage borrowers are allowed to take enhances the potential return given the same interest rate. The increase in revenue is 31.21% increase, which represents what the effective increase in capital efficiency to the Moonwell ecosystem could be.

We can take the difference in these quantities, and look at the incremental revenue by month.

Overall, we estimate that using the RociFi NFCs could have added an additional $3,006,970.19 in revenue.

Loan Distributions

Currently, we rank 61% of Moonwell borrowers as 10. That doesn’t necessarily mean that they’re bad users, but could have limited borrowing history to score. If these users continue to utilize Moonwell responsibly, their scores will undoubtedly increase.

The fact that 2’s are a small percentage of total loans, yet are the majority of volume tells us that the best ranked users on Moonwell borrow the most. Given these borrowers are less risky, this suggests returns are being left on the table by limiting their max LTV currently.

Using the proposed LTV curve we do not observe ANY instances where a loan that ends in liquidation would have an outstanding balance that exceeds available collateral for which that liability could be offset against.

However, there were some instances where the loan is paid off but there is lost interest revenue. These should not be confused with true losses as the lender has gotten back all the principal they lent out.

Hypothetical example, 20% APR is owed on $100 and $110 in Collateral is posted. Loan experiences liquidation and accounting for repays and liquidation proceeds, $102 is recovered at time of liquidation. If $120 is owed, then although we recovered all $100 of principal, there is $18 of un-accrued interest.

These events reduce the amount of potential revenue earned by lenders as there is unpaid interest of $1,413,680.19. We look at the unearned potential interest revenue by token

The largest unearned revenues are on WETH, xcDOT, mDot, and mETH. This suggests that on these 4 tokens we might consider a more conservative LTV curve. There are some unearned interest revenues on mUSDC and USDC, however these are far smaller than ETH & DOT pairs. This is surprising given we allow USDC loans to go up to 95%. This may be a function of scoring for those borrowers that took those loans, or total lending volumes on these tokens, thus requiring further investigation before implementation.

Looking at unearned revenues by token as percentage of borrow volume per token

As a percentage of borrow volume, mWBTC liquidations are the clear leader followed by mFRAX, mUSDC, and mDOT. This suggests that our LTVs on those four assets might need to be reduced on final implementation, although no principal losses are observed.

In conclusion, althoughno principal losses were observed, there were several instances of unearned interest revenue which diminish the potential revenue gains from increased capital efficiency. Further optimization of the LTV curves for these particular assets is required to maximize risk-adjusted revenue while maintaining principal losses at zero.

Key Assumptions

1. Increases in leverage offered to borrower does not affect the probability of default as long as the LTV does exceed the proposed LTV by score
2. For loans that get liquidated, if there are any repayments before the liquidation occurs, we assume that those repayments will also get scaled up by the leverage ratio
3. All loans are originated at the liquidation LTV
4. Constant scoring over the lifetime of the simulation
5. Constant rates over the lifetime of the simulation
6. Interest revenue can be collected daily against outstanding loan via collateral offset

For assumptions 1–4, we anticipate that actual revenue increases could be lower than what is displayed here. As stated prior, we’ve already spotted areas in our analysis which can be optimized for better risk-adjusted gains.

On the historical data, using the proposed LTV curve we do not observe any instances where a loan that ends in liquidation would have an outstanding balance that exceeds available collateral for which that liability could be offset against.

We also note that the data had excluded GLMR loans, had it included GLMR loans, potential revenue increases would likely be larger than what is displayed here.