Coinbase is suspending deposits and withdrawals for cbETH on OP Mainnet: Coinbase Markets 🛡️ on X: "Reminder: Coinbase will end support for cbETH deposits and withdrawals on the Arbitrum, Optimism, and Polygon networks on August 17, 2026. cbETH remains fully available to you on the Ethereum and Base networks. What does this mean for you? Read more in the thread ⬇️" / X
Indeed, I completely missed this one.
Thanks
Risk Parameter Recommendations and Conditions for Re-enabling Borrowing on Base (9/15/26)
Anthias Labs proposes the following parameter changes. For more information on current parameters, please refer to our monitoring dashboard.
Borrowing restrictions and conditions for reopening
Why borrowing remains restricted
The August 27 MAMO incident left substantial bad debt in the assets borrowed against MAMO collateral, including USDC, WETH, cbBTC and wstETH (Incident post-mortem). The incident also damaged user confidence and prompted suppliers to withdraw. As available cash was depleted, remaining suppliers faced difficulty exiting. That is the bank-run dynamic we need to resolve: people understandably want access to their funds, but the markets cannot meet all those withdrawals immediately.
All markets on base have borrow caps set to one wei, effectively preventing new borrowing. Currently, USDC and EURC had zero available liquidity. The restrictions protect liquidity that remains elsewhere in the deployment.
Why not leave interest rates extremely high until borrowers repay?
Higher borrowing costs can encourage solvent borrowers to repay, returning cash to the market. But interest also continues to accumulate on bad debt. An insolvent position cannot be made whole simply by charging it more interest. Extremely high rates can increase the amount requiring remediation without producing corresponding repayments. That is the tradeoff behind moderating the interest-rate models.
We believe borrow APYs around 20% in the affected markets offer a reasonable middle ground to evaluate: meaningful repayment pressure while limiting how quickly unrecoverable balances grow. After IRM updates apart of MIP-X66, borrow APYs are approximately 17.4% for USDC and 19.3% for EURC. These rates remain variable.
If the MAMO risk is contained, why not reopen the other markets?
Restricting MAMO does not repair the debt or withdrawal pressure already present elsewhere. There is a second risk: users unable to withdraw from one market could borrow assets from another to access liquidity.
For example, a supplier with $10,000 of USDC collateral and no existing debt could, at an 88% collateral factor, theoretically borrow up to $8,800 worth of WETH if borrowing were reopened and sufficient WETH were available. This would leave an outstanding loan against the USDC.
If ETH then rises in price, the dollar value of the WETH debt increases and the position can become liquidatable. A liquidator repays WETH and receives mUSDC receipt tokens. Those tokens can still be seized when USDC liquidity is unavailable, but redeeming them into USDC requires underlying cash in the contract.
Our concern is that liquidators may be unwilling or unable to fund that repayment while holding collateral they cannot promptly redeem. Delayed liquidation, further price moves and accumulating interest could then create additional bad debt in the borrowed asset’s market. Reopening would allow withdrawal pressure in USDC to consume liquidity in other markets and expose their suppliers to further losses.
What would we need to see before supporting reopening?
- Sustained liquidity in USDC and other connected, stressed markets. Withdrawals should function consistently, with enough cash to absorb further exits and the additional borrowing proposed. A brief positive balance after a repayment is not sufficient evidence of recovery.
- Evidence that liquidations will work under stress. The assessment should cover available collateral-market liquidity as well as executable swap liquidity.
The practical answer to “when?” is when those conditions can be demonstrated. Restoring borrowing is the goal, but reopening should help the markets recover without allowing the existing liquidity problem to spread further.
Base
Summary
Risk Parameters
| Parameters | Current Value | Recommended Value |
|---|---|---|
| WELL Collateral Factor | 65% | 55% |
| MORPHO Collateral Factor | 65% | 60% |
| AERO Collateral Factor | 65% | 60% |
Rationale
WELL
We recommend decreasing the WELL collateral factor from 65% to 55% as part of a gradual reduction in WELL collateral exposure.
Secondary-market liquidity is not deep enough for us to rely on orderly liquidations under stress. At the current snapshot, selling 15,000,000 WELL, worth approximately $29,167, into USDC returned 27,552.37 USDC at 5.55% price impact. Reaching this impact at a relatively small sale size raises concerns about recovery proceeds if multiple positions need to be liquidated together. WELL → USDC quote.
Thin liquidity also creates potential exposure to a long attack: an attacker could push the token’s market price upward and, if that inflated price feeds into its collateral valuation, borrow more than the collateral could recover when sold.
We have therefore taken a risk-off stance by setting WELL’s supply and borrow caps to 1 wei, effectively preventing additional supply and borrowing. These restrictions limit new activity, but existing WELL positions retain collateral borrowing power. We propose gradually winding down the collateral factor, with this initial reduction to 55% balancing risk reduction against the impact on existing borrowers.
Borrower impact
At the reviewed prices and balances, reducing WELL’s collateral factor from 65% to 55% would make two accounts newly eligible for liquidation, with $34.19 in combined debt.
| Account | Outstanding debt | Health factor: 65% → 55% |
|---|---|---|
| 0x525c49bf83ce3a1aaf425ac1a463537db68c8bd7 | $20.81 — 4,792.46 WELL + 11.56 USDC | 1.0641 → 0.9004 |
| 0x89b552ca032601dc60c0c2e67278b6f4bf18d069 | $13.38 — 13.38 USDC | 1.1143 → 0.9428 |
A health factor below 1 permits liquidation.
MORPHO and AERO
We recommend decreasing the collateral factors for both MORPHO and AERO from 65% to 60%. This would require more collateral relative to borrowing, providing additional protection against sharp price declines. At the reviewed snapshot, the MORPHO change introduces no new liquidation eligibility; the AERO change newly affects one account with $150.82 of debt.
Over the 90-day period from June 16 to September 14, 2026, we measured:
| Asset | Latest 30-day annualized volatility | 90-day annualized volatility | Worst 1h decline | Worst 4h decline | Worst 24h decline | Maximum daily-sampled drawdown |
|---|---|---|---|---|---|---|
| MORPHO | 110.0% | 83.3% | 6.45% | 7.84% | 11.32% | 25.44% |
| AERO | 103.1% | 93.0% | 7.61% | 11.21% | 16.38% | 31.72% |
Both assets experienced substantial downside moves over short periods. These declines can erode the collateral cushion when the borrowed asset is not correlated alongside the collateral.
Annualized volatility measures the variability of returns, including upward moves; the observed declines show the downside risk directly. The maximum drawdown captures a peak-to-trough decline that may span several weeks. These historical observations support a more conservative collateral factor, without establishing 60% as an optimal level. We favor this measured first step given the limited immediate borrower impact.
MORPHO also faces a redemption constraint: the lending market currently holds only 34.55 MORPHO, approximately $74.05, in available cash. A liquidator receiving mMORPHO needs underlying market cash to redeem those receipt tokens before selling MORPHO.
Borrower impact
- MORPHO: reducing the factor to 60% introduces no newly liquidatable accounts at the snapshot.
- AERO: reducing the factor to 60% newly affects the account below. This borrower supplies and borrows AERO, so the change in liquidation eligibility comes from the lower collateral factor itself.
| User address | Collateral | Borrows | Health factor: current → proposed | Repayment to HF 1.05 (USD equivalent) |
|---|---|---|---|---|
| 0xedeb3b99964f0fa55b63fe9f6db2f20f02e52265 | 451.82 AERO ($248.99) | 273.68 AERO ($150.82 total) | 1.0731 → 0.9905 | $8.55 |
A health factor below 1 permits liquidation.
OP Mainnet
Summary
Risk Parameters
| Parameters | Current Value | Recommended Value |
|---|---|---|
| VELO Collateral Factor | 62% | 60% |
Rationale
VELO
We recommend decreasing the VELO collateral factor on OP Mainnet from 62% to 60% as an initial step toward reducing collateral exposure.
Limited secondary-market liquidity supports a conservative approach to VELO’s borrowing power. At the current snapshot, selling 5,000,000 VELO, worth approximately $119,387, into USDC returned 114,304.48 USDC at 4.27% price impact. Other available routes returned approximately 106,000 USDC for the same sale, demonstrating how recovery proceeds can depend on route availability. VELO → USDC quote.
If multiple positions require liquidation together, the resulting sales could push VELO’s price lower and reduce the proceeds available to repay debt. We therefore favor gradually reducing the collateral factor, beginning with a move to 60%. This limits borrowing power while giving existing borrowers time to adjust before further reductions.
Borrower impact
At the reviewed prices and balances, reducing VELO’s collateral factor from 62% to 60% would make one account newly eligible for liquidation, with $2,394.80 of debt.
| Account | Outstanding debt | Health factor: 62% → 60% |
|---|---|---|
| 0xcb6586874cc04b01cc4fdb777de502cea7b3d6c1 | $2,394.80 — 29,950.48 VELO + 16,923.03 OP | 1.0008 → 0.9912 |
A health factor below 1 permits liquidation. This borrower supplies both VELO and OP and is already close to the liquidation threshold.
Additional Links
Anthias Labs has not been compensated by any third party for any statements made. All opinions and suggestions provided are based solely on our independent analysis and are not influenced by external entities.
[Anthias Labs] Risk Parameter Recommendations (9/29/26)
Anthias Labs proposes the following parameter changes for Moonwell Core markets on Base, OP Mainnet and Ethereum. For more information on current parameters, please refer to our monitoring dashboard.
Base
Summary
Risk Parameters
| Parameters | Current Value | Recommended Value |
|---|---|---|
| WELL Collateral Factor | 55% | 45% |
| MORPHO Collateral Factor | 60% | 40% |
| VIRTUAL Collateral Factor | 65% | 55% |
| wrsETH Collateral Factor | 46% | 42% |
| cbBTC Reserve Factor | 10% | 99% |
| MAMO Collateral Factor | 50% | 0% |
Borrow caps on all 21 Base markets remain at one raw token unit. We don’t recommend reopening borrowing yet (see below).
IR Parameters
| cbBTC IR Parameters | Current Value | Recommended Value |
|---|---|---|
| Base | 0 | 0.01 |
| Kink | 0.8 | 0 |
| Multiplier | 0.0615 | 0 |
| Jump Multiplier | 5 | 0 |
The recommended cbBTC parameters match the existing flat model, which cbETH, VIRTUAL and cbXRP already use.
Estimated borrower impact
| Change | Newly liquidatable accounts | Total debt in those accounts |
|---|---|---|
| WELL 55% → 45% | 2 | $30.57 |
| MORPHO 60% → 40% | 0 | $0.00 |
| VIRTUAL 65% → 55% | 0 | $0.00 |
| wrsETH 46% → 42% | 0 | $0.00 |
Borrowing restrictions: status update
The conditions we set out on 9/15 for reopening Base borrowing haven’t been met. Liquidity has kept draining from the markets that still had it.
| Market | Available cash, 9/15 | Available cash, 9/29 | Change |
|---|---|---|---|
| USDC | 0.00005 USDC | 0 USDC | none left |
| EURC | 0 EURC | 0 EURC | none left |
| MORPHO | 34.55 MORPHO | 1.51 MORPHO | −96% |
| cbBTC | 23.09 cbBTC | 11.92 cbBTC | −48% |
| AERO | 3,742,480 AERO | 2,040,783 AERO | −45% |
| cbXRP | 1,218,783 cbXRP | 839,659 cbXRP | −31% |
| VIRTUAL | 230,831 VIRTUAL | 177,479 VIRTUAL | −23% |
| WETH | 391.24 WETH | 351.27 WETH | −10% |
| wstETH | 187.83 wstETH | 184.84 wstETH | −2% |
| All 21 markets (oracle USD) | $13.76 million | $11.43 million | −17% |
- Bad debt is unresolved. Reconstructing every borrower onchain gives $14.91 million of net shortfall across 111 accounts, compared with $13.73 million across 110 accounts on 9/15. Gross borrows in insolvent accounts, measured in tokens, are almost unchanged (for example, cbBTC 54.11 → 54.33 and WETH 746.74 → 747.11). The USD increase mostly reflects higher prices of the borrowed assets and accrued interest.
- Liquidations still can’t be relied on. Healthy borrowers have $2.15 million of USDC collateral, $1.79 million of cbETH, $102 thousand of EURC and $58 thousand of MORPHO posted in markets with essentially no cash. A liquidator can seize those receipt tokens but can’t redeem them.
We recommend keeping borrow caps at one raw unit. Before any reopening, supply and borrow caps should be re-derived from current liquidation liquidity.
Rationale
cbBTC
MIP-X66 moved several Base markets to lower-rate models to slow interest accrual on bad debt. cbBTC currently uses a kinked model (80% kink, 0.0615 multiplier, 5.0 jump multiplier). At that point its utilization was about 72% and its borrow APY about 4.4%. Since then, suppliers have withdrawn cbBTC. Supply fell from about 81.5 to 70.0 cbBTC, cash fell from 23.09 to 11.92 cbBTC, and utilization is now 83.1%, above the kink. The borrow APY is 22.6%.
Interest now mostly accrues on debt that can’t be repaid:
| cbBTC borrows | Tokens | USD | Accounts |
|---|---|---|---|
| In insolvent accounts (debt exceeds all collateral) | 54.33 | $4,520,720.89 | 8 |
| In healthy accounts | 3.79 | $315,625.57 | 93 |
- Using the same monthly-interest method as our earlier bad-debt reports, Base bad debt is accruing about $128,000 per month, compared with the ~$50,273 per month projected for MIP-X66 at the utilization prevailing then. cbBTC alone contributes $77,451 per month.
- Healthy cbBTC borrowers owe only 3.79 cbBTC, and their balances fell by roughly 0.5 cbBTC over the two weeks since 9/15. The higher rate is producing little repayment.
In our 9/15 post we suggested borrow APYs around 20% as a middle ground for the affected USDC and EURC markets. That approach fits where a meaningful share of the debt is healthy and repayment pressure can return cash, as in USDC (28.9% of debt insolvent) and EURC (46.9%). With 93.5% of cbBTC debt insolvent, cbBTC is closer to cbETH (98.2%), VIRTUAL (99.1%) and cbXRP (96.1%). Those markets already use a flat 1% model with a 99% reserve factor.
We therefore recommend:
- moving cbBTC to the same flat model: a 1% borrow rate at any utilization; and
- raising cbBTC’s reserve factor from 10% to 99%, so the small remaining accrual goes to reserves instead of increasing supplier balances that the market can’t pay.
This would cut cbBTC’s bad-debt accrual to about $3,769 per month, and total Base accrual to about $54,316 per month. It would also remove the ~16.5% supply APY the market currently displays, which overstates what a new cbBTC supplier can expect from a market whose supplier balances are only about 22% backed by cash and healthy loans.
Current rates (reserve factor 10%; simple annual rates)
| Utilization | Borrow APR | Supply APR |
|---|---|---|
| 0% | 0% | 0% |
| 80% (kink) | 4.92% | 3.54% |
| 100% | 104.92% | 94.43% |
Projected rates (flat model, reserve factor 99%)
| Utilization | Borrow APR | Supply APR |
|---|---|---|
| 0% | 1.00% | 0.00% |
| 80% | 1.00% | 0.01% |
| 100% | 1.00% | 0.01% |
Collateral-factor reductions
| Asset | CF: current → proposed | Collateral posted by healthy borrowers (accounts) | Market cash (redeemable) | Share of the market’s own debt in insolvent accounts | ~5% sale depth into USDC (impact) | 30d / 90d annualized vol | Worst 24h decline | Max drawdown (90d) |
|---|---|---|---|---|---|---|---|---|
| WELL | 55% → 45% | $71,392 (105) | $261,805 | 0.2% | $31,623 (5.26%) | 104.4% / 99.0% | 31.95% | 52.69% |
| MORPHO | 60% → 40% | $58,132 (48) | $4 | 85.8% | $562,341 (5.37%) | 77.5% / 84.5% | 11.25% | 27.97% |
| VIRTUAL | 65% → 55% | $827,773 (152) | $142,800 | 99.1% | $1,154,782 (5.08%) | 84.9% / 84.1% | 14.17% | 26.23% |
WELL
We recommend decreasing WELL’s collateral factor from 55% to 45%, the second step of the gradual wind-down we began on 9/15.
Liquidity hasn’t improved. Selling 13,851,400 WELL, worth about $31,623, returned 29,929.60 USDC at 5.26% price impact, which is nearly identical to our 9/15 observation. Borrowers hold $71,392 of WELL collateral, 2.3× that sale size. WELL fell 31.95% within 24 hours and 21.29% within a single hour on August 27. Its 90-day maximum drawdown is 52.69%. These reasons, together with the long-attack concern raised on 9/15, support continuing to reduce WELL’s borrowing power.
The same two accounts identified on 9/15 become newly eligible for liquidation, now with $30.57 of combined debt.
MORPHO
We recommend decreasing MORPHO’s collateral factor from 60% to 40%.
- mMORPHO can’t be redeemed. The market holds 1.51 MORPHO ($3.70) of cash, down from 34.55 MORPHO on 9/15. A liquidator receiving mMORPHO has no cash to redeem it against.
- The collateral itself is impaired. 85.8% of MORPHO debt is insolvent. Cash and healthy loans back only about 14% of MORPHO supplier balances.
- Market risk is high. The ~5% sale size into USDC is about $562,000, and 90-day volatility is 84.5%.
The larger step is possible because no account becomes newly liquidatable at 40%. The next account to cross the threshold would do so only at 35%, with $139.11 of debt.
VIRTUAL
We recommend decreasing VIRTUAL’s collateral factor from 65% to 55%.
The VIRTUAL market is already treated as impaired: it has a flat 1% rate model, a 99% reserve factor, and 99.1% of its own debt in insolvent accounts. The token is still accepted as collateral at 65% CF.
Healthy borrowers post $827,773 of VIRTUAL collateral, but the market holds only $142,800 of redeemable cash, down 23% since 9/15. Cash and healthy loans back only about 13% of VIRTUAL supplier balances.
The step is modest relative to VIRTUAL’s 84–85% annualized volatility and 14.17% worst 24-hour decline. This change introduces no new liquidation eligibility.
wrsETH
We recommend decreasing wrsETH’s collateral factor from 46% to 42%, continuing its deprecation. The market’s mint and borrow functions are paused, and seven borrowers hold $67,603 of wrsETH collateral. No account becomes newly liquidatable.
The largest wrsETH borrower, 0x451ff8…73a3, with $4,256 of WETH debt, would move from a health factor of 1.18 to 1.08. We urge it to repay before a further step.
Borrower impact
The accounts below become newly eligible for liquidation
| User address | Collateral | Borrows | Health factor: current → proposed |
|---|---|---|---|
| 0x525c49bf83ce3a1aaf425ac1a463537db68c8bd7 | 15,012.37 WELL ($34.27) | 2,399.27 WELL ($5.48); 11.63 USDC ($11.63) ($17.10 total) | 1.1020 → 0.9017 |
| 0x89b552ca032601dc60c0c2e67278b6f4bf18d069 | 11,886.1 WELL ($27.14) | 13.46 USDC ($13.46 total) | 1.1086 → 0.9071 |
A health factor below 1 permits liquidation.
MAMO
Chainlink plans to deprecate the price oracle which the MAMO market uses. Therefore. we recommend sunsetting the MAMO market by setting Collateral Factor to 0.
OP Mainnet
Summary
Risk Parameters
| Parameters | Current Value | Recommended Value |
|---|---|---|
| VELO Collateral Factor | 60% | 55% |
| OP Collateral Factor | 65% | 60% |
| wrsETH Collateral Factor | 37% | 30% |
| cbETH Collateral Factor | 1% | 0% |
| WBTC Collateral Factor | 0.1% | 0% |
Estimated borrower impact
| Change | Newly liquidatable accounts | Total debt in those accounts |
|---|---|---|
| VELO 60% → 55% | 1 | $922.28 |
| OP 65% → 60% | 1 | $12.45 |
| wrsETH 37% → 30% | 2 | $238.73 |
| cbETH 1% → 0% | 0 | $0.00 |
| WBTC 0.1% → 0% | 0 | $0.00 |
Rationale
VELO
We recommend decreasing VELO’s collateral factor from 60% to 55%, the next step of the gradual reduction begun on 9/15. The VELO market is closed to new supply and borrowing and carries a 99% reserve factor.
- Liquidity. Selling 5,080,140 VELO (about $177,828) returned 168,975 USDC at 4.98% impact. Borrowers post $593,521 of VELO collateral, 3.3× that amount.
- Price moves. VELO has risen 47.9% since 9/15. Its 30-day annualized volatility is 116.4%, and its 90-day maximum drawdown is 33.8%.
One account becomes newly eligible for liquidation: 0xcb6586874cc04b01cc4fdb777de502cea7b3d6c1, the same borrower flagged on 9/15.
The largest VELO-collateralized loan, 0x771549…6f3f, with $73,420 of USDC and WETH debt, has benefited from VELO’s rally. Its health factor is 1.49 today and remains well above 1 at 55%.
OP
We recommend decreasing OP’s collateral factor from 65% to 60%, in line with the 60% now applied to AERO and VELO.
- Liquidity: OP’s own market is closed to new supply and borrowing. OP Mainnet liquidity for the token is thin: 1,883,000 OP (about $245,824) sold into USDC at 5.28% impact.
- Volatility: Its 30-day volatility is 114.1%.
One account holding $12.45 of OP debt, against OP collateral, becomes newly eligible for liquidation.
wrsETH, cbETH and WBTC
These are deprecation steps:
- wrsETH, 37% → 30%. This affects two small accounts with $238.73 of combined debt.
- cbETH, 1% → 0%. This completes the change we recommended on 8/20. The only two accounts holding cbETH collateral are already liquidatable, so no account is newly affected.
- WBTC, 0.1% → 0%. No account is newly affected.
Borrower impact
| User address | Collateral | Borrows | Health factor: current → proposed | Repayment to HF 1.05 (USD) |
|---|---|---|---|---|
| 0xcb6586874cc04b01cc4fdb777de502cea7b3d6c1 | 41,853.77 VELO ($1,465.07); 1,048.75 OP ($136.91) | 26,347.53 VELO ($922.28 total) | 1.0496 → 0.9628 | $76.63 |
| 0xc56b6d7d14de9a38a51ff2029e8a6b1ee4ff1d68 | 0.141 wrsETH ($406.39) | 0.0503 WETH ($134.40 total) | 1.1187 → 0.9071 | $18.29 |
| 0xe80fc2700ec6faf5f0347a2e7e7798faf548e1c3 | 0.100 wrsETH ($288.85); 0.00000646 weETH ($0.02) | 0.0350 weETH ($103.26); 1.06 USD₮0 ($1.06) ($104.32 total) | 1.0246 → 0.8308 | $21.78 |
| 0x97ac41f951b6eb32f302d919fb7e44274e06ca57 | 147.71 OP ($19.28) | 95.35 OP ($12.45 total) | 1.0069 → 0.9294 | $1.43 |
Additional Links
- Monitoring Dashboard
- Anthias Labs risk parameter recommendations thread (9/15/26 post)
- Post-mortem: MAMO market incident on Base
- Anthias Labs X Account
Anthias Labs has not been compensated by any third party for any statements made. All opinions and suggestions provided are based solely on our independent analysis and are not influenced by external entities.