Summary
Modify loan discounts in the following lend markets to these values (In cases where multiple markets exist with the same collateral, apply the loan discount value to all markets):
| Market | Chain | Current Loan Discount | Proposed Loan Discount |
|---|---|---|---|
| CRV-long | Arbitrum | 11 | 20 |
| CRV-long | Optimism | 15.5 | 20 |
| CRV-long | Fraxtal | 15.5 | 20 |
| ARB-long | Arbitrum | 11 | 21 |
| OP-long | Optimism | 15.5 | 17.5 |
| FXS-long | Fraxtal | 15.5 | 21 |
A subsequent proposal will be made to modify liquidation discount in these markets.
Analysis
See our previous analysis on CRV-long mainnet market with A=30. We recommend adjusting loan and liquidation discount parameters and standardising parameters by collateral type per chain. The following markets are included in this proposal:
We can see that Fraxtal and OP have a market with A=22 for the CRV-long markets, respectively; hence, we provide in this analysis a revised parameter recommendation.
CRV-long (A=22)
Referencing the market’s A=22, the ES at 1% reads 0.1124. So we set liquidation_discount = 12% (rounded up).
Looking at how much worse the tail can get from ES at 1% → ES at 0.25%, we measure +8.36 percentage points. Consequently, we recommend loan_discount ≈ 20% (0.196 = 0.1124 + 0.0836).
| Parameter | Current (CRV-long) | Proposed | Change |
|---|---|---|---|
| A | 22 | 22 | — |
| Liquidation discount | 12.5% | 12% | −0.5 pp |
| Loan discount | 15.5% | 20% | +4.5 pp |
Compared with the current CRV-long settings (A=22, loan=15.5%, liq=12.5%), this implies moving to loan≈20% and liq≈12% under the same fee/oracle configuration.
ARB-long
From our run, the orange-curve minimum (A*) sits substantially above the live setting, at A≈134. Holding the market’s A=30, the blue ES@1% reads 0.0913, so we set liquidation_discount = 10% (rounded up).
Looking at how much worse the tail can get from Top 1% to Top 0.25% worst days, we measure +11.772 percentage points. Consequently, we recommend **loan_discount** = 21% (0.20902 = 0.0913 + 0.11772).
| Parameter | Current (ARB2-long) | Proposed | Change |
|---|---|---|---|
| A | 30 | 30 | — |
| Liquidation discount | 8% | 10% | +2 pp |
| Loan discount | 11% | 21% | +10 pp |
Compared with current ARB2-long settings (A=30, loan=11%, liq=8%), this implies moving to loan=21% and liq=10% to align with observed one-day tail behaviour under the present fee/oracle configuration.
FXS-long
From our run, the orange-curve minimum (A*) sits well above the live setting, at A≈68. Holding the market’s A=22, the ES at 1% reads 0.1048, so we set liquidation_discount = 11% (rounded up).
Comparing ES at 0.25% vs ES at 1% shows an increment of +9.89 percentage points (20.36% − 10.48%). Consequently, we recommend loan_discount ≈ 21% (rounding [email protected]% ≈ 21% to the nearest 1%).
| Parameter | Current (FXS-long) | Proposed | Change |
|---|---|---|---|
| A | 22 | 22 | — |
| Liquidation discount | 12.5% | 11% | −1.5 pp |
| Loan discount | 15.5% | 21% | +5.5 pp |
Live is A=22, loan=15.5%, liq=12.5%. We keep A=22, raise loan → 20.5% (+5.0 pp, tighter borrowing), and trim liq → 11% (−1.5 pp, slightly more lenient liquidation trigger). Net effect: max LTV drops 84.5% → 79.5%.
OP-long
From our run, the orange-curve minimum (A*) sits well above the live setting of A=22 with A≈68. Holding the market’s A=22, the ES at 1% reads 0.1131, so we set liquidation_discount = 12% (rounded up).
Comparing ES at 0.25% vs ES at 1% shows an increment of +6.04 percentage points (17.36% − 11.31%). Consequently, we recommend loan_discount ≈ 18% (rounding [email protected]% ≈ 17.36% to the nearest 1%).
| Parameter | Current (OP-long) | Proposed | Change |
|---|---|---|---|
| A | 22 | 22 | — |
| Liquidation discount | 12.5% | 12% | −0.5 pp |
| Loan discount | 15.5% | 17.5% | +2.0 pp |
Live is A=22, loan=15.5%, liq=12.5%. Our proposal keeps A=22, moves loan → 17.5% (+2 pp, tighter borrowing), and liq → 12% (−0.5 pp, slightly more lenient trigger).
Note on the oracles: The OP-long and FXS-long market uses a Chainlink oracle as its price feed. The number of observations to calculate the EMA time is 20. The update frequency seems to be between 1 min to 3 min, yet it varies. Hence, to approximate the simulation, we use a Texp of (20*2*60) = 2400. We recommend, if the market should see an increase in TVL, re-running using the on-chain oracle price feed with the respective 20 observation EMA as specified in the Oracle contract for a more accurate picture. This is feasible for the OP-long market given that the oracle contract has sufficient price history, having been deployed 3+ years ago. FXS-long has, however, only close to 2y worth of data and may require argumentation.