Aave Oracle Glitch Triggers $26M Liquidation Cascade in 15 Minutes After wstETH Price Feed Error

Misconfigured CAPO Oracle Temporarily Mispriced wstETH, Forcing Liquidations Across Dozens of Accounts

TL;DR

• A configuration error in Aave’s oracle system caused roughly $26.4 million in liquidations in about 15 minutes.
• The glitch incorrectly lowered the wstETH/stETH ratio, affecting about 34 accounts and liquidating 10,938 wstETH.
• Liquidation bots captured roughly 499 ETH in profits before the issue was identified and addressed.

We’ve launched the all-new COIN360 Perp DEX, built for traders who move fast!

Trade 130+ assets with up to 100× leverage, enjoy instant order placement and low-slippage swaps, and earn USDC passive yield while climbing the leaderboard. Your trades deserve more than speed — they deserve mastery.

A configuration error in decentralized lending protocol Aave triggered a sudden wave of liquidations after its oracle infrastructure briefly misreported the value of a key Ethereum staking derivative, wiping out collateral positions despite stable market conditions. The incident unfolded on March 11, 2026, when the protocol’s Correlated Asset Price Oracle system applied an incorrect price cap to the exchange ratio between wrapped staked Ether and staked Ether, causing automated risk controls to interpret the collateral value as falling below required thresholds.

Aave’s Correlated Asset Price Oracle, known as CAPO, is designed to protect lending markets by monitoring correlated assets and preventing manipulation of closely linked prices. The oracle is operated by risk-management firm Chaos Labs, which monitors asset relationships and deploys safeguards meant to defend the protocol against artificially manipulated prices that could distort borrowing markets.

Investigators later traced the malfunction to a mismatch between a stored price ratio snapshot and the timestamp associated with that data point. The discrepancy caused the system to calculate a maximum allowable price that was lower than the real market value, even though trading activity across crypto price indexes showed no abnormal movement in the underlying asset.

That mismatch caused the protocol to register the wstETH/stETH exchange rate roughly 2.85% below the actual market value. Chaos Labs later said the issue resulted from “data desynchronization between the snapshot ratio and the snapshot timestamp,” two internal variables that must remain aligned for the oracle system to calculate a correct price ceiling.

Smart-contract constraints complicated the problem. The parameter governing the oracle’s price cap could only increase by 3% every three days, preventing the system from updating the ratio to match real-time market data once the timestamp advanced ahead of the stored value. With the ratio capped while the timestamp continued updating, the oracle produced an incorrect reference price for the asset pair.

During the malfunction, the protocol recorded the exchange ratio at 1.1939 even though the market level stood near 1.228, creating an artificial drop in collateral value. Lending markets using the data feed interpreted the change as a real market event, triggering Aave’s automated liquidation engine to begin closing positions whose health factor appeared to fall below required collateralization levels.

Aave’s liquidation system operates entirely through smart contracts that automatically execute when the health factor of a loan falls under the protocol’s safety threshold. The contracts rely solely on oracle data feeds to determine collateral value and cannot distinguish between genuine price movements and erroneous inputs from a faulty data source.

Thousands of positions backed by wrapped staked Ether were flagged as undercollateralized once the inaccurate price propagated through the system. Liquidation transactions began firing across the protocol’s lending pools within minutes after the faulty price feed was applied.

The liquidation cascade unfolded rapidly, processing about $26.4 million in forced collateral sales in roughly 15 minutes. Blockchain records show the event affected approximately 34 accounts, which collectively saw 10,938 wstETH liquidated as the automated risk controls reacted to the artificial price drop.

Liquidation transactions were executed by automated on-chain bots that continuously monitor lending markets for positions eligible for liquidation. These bots purchase collateral from undercollateralized borrowers at a discount, capturing incentives built into lending protocols to maintain solvency.

Data from the transactions shows third-party liquidators extracted roughly 499 ETH in profits during the event. Despite the scale of the liquidations, the protocol itself did not accumulate any bad debt because collateral was sold quickly enough to cover the outstanding borrow positions.

Affected users were liquidated even though their positions remained sufficiently collateralized under real market prices tracked across crypto price index platforms and broader crypto price monitoring tools used by traders to follow movements in the digital asset ecosystem.

Developers and governance participants began investigating the anomaly shortly after abnormal liquidation activity appeared on-chain. Chaos Labs published a technical explanation describing the precise sequence of events that led to the incorrect price cap calculation, and said that affected users would be reimbursed.

Protocol contributors confirmed that affected users were not engaging in excessive leverage or risky trading behavior at the time their loans were closed. Governance discussions later began outlining compensation mechanisms intended to reimburse borrowers whose positions were liquidated due to the oracle error.

Recovery proposals described returning liquidation proceeds and, if necessary, drawing funds from the Aave DAO treasury to compensate affected users. Governance participants identified 141.5 ETH recovered from liquidation revenue and discussed a potential allocation of 345 ETH from treasury reserves to cover remaining losses.

The malfunction renewed debate across decentralized finance about reliance on price oracles, external data systems that supply smart contracts with real-world market information used to determine crypto price and collateral value across lending protocols.

Observers noted that the liquidation event occurred without a broader market crash or price manipulation attack, meaning the losses stemmed entirely from a configuration error in the oracle infrastructure rather than trading activity or malicious behavior in the coin market cap ecosystem tracked by industry analytics platforms.

Smart-contract automation allowed the protocol to execute risk controls almost instantly once the faulty price was reported. The same design also meant that liquidations proceeded automatically before developers or governance members could intervene to halt the cascading transactions.

This article has been refined and enhanced by ChatGPT.