Tag: Bitcoin-Backed Loans

Proposal Details: Higher Leverage, Tighter Liquidation Cushion

Aave governance is considering a parameter update proposed by risk service provider LlamaRisk that would materially increase leverage for Bitcoin-backed borrowers on Aave V3 Ethereum Core. The proposal would raise the maximum loan-to-value (LTV) ratio for Wrapped Bitcoin (WBTC) and Coinbase Wrapped BTC (cbBTC) from 73% to 81%, allowing users to borrow up to $0.81 per $1 of collateral instead of $0.73. Simultaneously, the liquidation threshold—the point at which a position becomes eligible for liquidation—would increase from 78% to 85%.

This change narrows the raw distance between the borrowing limit and the liquidation line from five percentage points to four. In practical terms, a maximally leveraged Bitcoin position would reach the proposed liquidation threshold after approximately a 4.7% collateral-price decline, compared with roughly 6.4% under current parameters. LlamaRisk announced on Sept. 21 that the proposal had advanced to Snapshot and voting would begin in less than 24 hours. The vote result and any subsequent implementation remained unverified at the reporting cutoff, meaning the higher limits are proposed parameters rather than live settings.

Cross-Chain Parameter Adjustments Beyond Ethereum Core

The proposal extends beyond Ethereum Core. On Arbitrum, WBTC’s ordinary LTV would rise by five percentage points. On Base, cbBTC’s LTV would increase by eight points, and a separate Base cbBTC stablecoin E-Mode would move to 82% LTV with an 85% liquidation threshold. Ethereum Core assets WETH, wstETH, and weETH would each receive a 0.5-point LTV increase. Selected liquidation thresholds would also rise, while Base cbBTC’s liquidation bonus would fall from 7.5% to 6%.

These figures describe maximum capacity per dollar of eligible collateral and do not quantify how much additional debt users would actually take on. The proposal does not disclose a complete current dataset of collateral-enabled balances, debt attributed to each affected asset, or account health-factor distributions. Total reserve supply would overstate usable collateral because some supplied tokens may be ineligible, disabled as collateral, or unconnected to debt. Historical seized volume likewise does not reveal the live distribution of positions.

Liquidation Speed Analysis Underpins Risk Model

LlamaRisk studied liquidation behavior from August 2025 through August 2026 across Ethereum Core, Arbitrum, and Base. On Ethereum Core, the analysis counted 7,206 ETH liquidations that seized $618 million and 2,621 BTC liquidations that seized $358 million. For both collateral families, the value-weighted 99th-percentile time a liquidation call spent at or below its execution price was five minutes. This statistic measures a liquidation work-off window and differs from saying 99% of transactions executed within five minutes of an oracle update; large positions can require several calls because a single liquidation generally repays only part of the debt.

The study separately measured processing after price-feed publications during the February and October 2025 stress windows. Its data reports that 100% of seized volume cleared within five minutes of the feed publication that made liquidation profitable in every listed market during those two events. February produced no recognized deficit. October produced $0.39 million of event-level bad debt against roughly $128 million, although LlamaRisk said none affected the ETH- or BTC-family collateral analyzed for this proposal. The result suggests liquidator response was not the binding constraint in those episodes, though it does not recreate those events under the proposed higher leverage.

Model Limitations and Residual Risks

The risk model combines a one-hour price excursion with each reserve’s liquidation bonus to derive a ceiling for the liquidation threshold. The 99.9th-percentile adverse one-hour move was 11.85% for ETH and about 5% for BTC. This percentile omits the most extreme 0.1% of one-hour observations in the two-year sample, which contained much larger moves: a worst one-hour ETH decline of 24.27% and a worst BTC decline of 10.72% (the proposal’s summary gives 11.15% for the BTC worst hour, an internal discrepancy). Either BTC figure is more than twice the roughly 5% percentile input.

This gap defines the residual risk. LlamaRisk’s framework assumes regular oracle publications and responsive liquidators prevent a maximally leveraged position from sitting untouched for a full hour. A move beyond the percentile can become more damaging if price feeds stall, liquidation activity slows, or market depth deteriorates simultaneously. The percentile therefore calibrates a protocol bad-debt buffer rather than a borrower protection level. The proposal leaves BTC thresholds below the model ceiling to account for depth, caps, and concentration risks that price history does not capture. ETH receives less room: WETH is set at the model ceiling, while wstETH and weETH sit one point inside their ceilings.

Governance Decision: Balancing Capital Efficiency and Protocol Safety

Aave’s governance choice is whether observed liquidation performance under current parameters justifies allowing future positions to run closer to liquidation. For Bitcoin collateral on Ethereum Core, the exchange is clear at the borrower level: maximum LTV would rise eight percentage points while the simplified collateral-price cushion at maximum leverage would shrink from about 6.4% to 4.7%. Existing borrowers would not automatically add debt, but the new limits would permit new or adjusted positions to carry more.

The protocol-level case appears more favorable than the borrower-level cushion alone suggests. In LlamaRisk’s sample, economically meaningful liquidations were processed quickly, and the two studied stress windows left no bad debt on the reviewed ETH- and BTC-family collateral. The model also incorporates the liquidation bonus and keeps recommended BTC thresholds below its calculated ceiling. However, historical execution does not measure an outage coinciding with an exceptional price move. The worst one-hour declines in the same dataset exceeded the percentile inputs by a wide margin, and the model cannot remove liquidity, concentration, or oracle risk. The aggregate credit effect also remains unknown; calculating it would require current collateral-enabled balances for every affected asset and market, the debt those positions already carry, their collateral settings, and their health-factor distribution. The published proposal supplies the parameter change, not that full position-level dataset.

Why This Matters

This proposal represents a significant test of decentralized finance (DeFi) risk management methodology. Aave, as one of the largest lending protocols, sets precedents for how on-chain lending markets balance capital efficiency against systemic risk. The LlamaRisk framework relies heavily on high-frequency liquidation performance data during stress events, arguing that rapid work-off of underwater positions justifies tighter liquidation cushions. However, the model explicitly acknowledges it cannot account for correlated failures—such as oracle delays, liquidator capital constraints, or market depth evaporation—during extreme volatility. The governance vote will signal the community’s risk appetite and could influence parameter-setting approaches across other lending protocols. If approved, the subsequent Aave Improvement Proposal (AIP) will codify the exact parameters for implementation, making the Snapshot vote a critical governance milestone.

Frequently Asked Questions

What are the exact parameter changes proposed for WBTC and cbBTC on Ethereum Core?

The proposal would raise the maximum LTV from 73% to 81% and the liquidation threshold from 78% to 85% for both WBTC and cbBTC on Aave V3 Ethereum Core.

When will the vote take place and what happens after?

LlamaRisk stated on Sept. 21 that the proposal had advanced to Snapshot and voting would begin in less than 24 hours. An implementation AIP would follow only after a positive Snapshot result. Until a vote and AIP establish final values, the 81% Bitcoin LTV remains a governance proposal.

Does the liquidation data guarantee the new parameters are safe?

No. The analysis shows that under current parameters, economically meaningful liquidations cleared within minutes during observed stress windows. However, the model uses a 99.9th-percentile one-hour price move (roughly 5% for BTC) plus the liquidation bonus to set thresholds, while the same dataset contains worst-case one-hour declines exceeding 10% for BTC. The framework assumes continuous oracle updates and responsive liquidators; it cannot eliminate risks from simultaneous oracle failure, liquidator inaction, or severe market depth deterioration during extreme moves.