Key takeaways:
– A brief 2.85% undervaluation of wstETH on Aave triggered roughly $27 million in liquidations, showing how small configuration errors can cascade in automated DeFi systems.
– Aave’s risk module temporarily priced wstETH around 1.19 ETH while market rates were near 1.23 ETH, pushing some positions below collateral requirements.
– Price oracles and risk-oracle settings are critical infrastructure: they determine collateral values, loan health and when liquidations occur.
– The root cause was a misconfiguration in Aave’s CAPO risk-oracle module: stale contract parameters created a temporary cap on the token’s exchange rate.
What happened
On March 10, 2026, Chaos Labs reported about $27 million in Aave liquidations within 24 hours. This was not the result of a broad market crash but of a short-lived 2.85% mispricing of wrapped staked ETH (wstETH) used as collateral. The event underscores how sensitive automated lending platforms are to accurate, timely oracle and parameter data.
How liquidations spiked
In lending protocols like Aave, on-chain price oracles supply external market prices used to value collateral. If a borrower’s collateral falls below required thresholds, the protocol’s automated liquidation mechanics allow third parties (often bots) to repay debt in exchange for discounted collateral. Because these processes are fast and automated, even brief price or parameter discrepancies can trigger rapid, large-scale liquidations.
About wstETH
wstETH is Lido’s wrapped version of stETH that represents staked ETH plus accrued rewards in a fixed-balance token. Because staking rewards accumulate, wstETH normally trades at a premium to ETH and is commonly used as collateral in DeFi.
The pricing mismatch
During the incident, Aave’s risk calculations valued wstETH at about 1.19 ETH while prevailing market prices were close to 1.23 ETH — roughly a 2.85% gap. That lower internal valuation caused some loans backed by wstETH to appear undercollateralized and be liquidated by on-chain actors.
Why oracles matter
Blockchains cannot natively read off-chain markets, so protocols rely on oracle services to bring price data on-chain. Those oracle feeds, plus any additional risk-layer logic that processes them, determine collateral valuations, loan health metrics and liquidation triggers. When either raw feeds or the risk-layer logic is incorrect or desynchronized, automated systems can react in ways that cause outsized losses for users.
Root cause: CAPO misconfiguration
Investigations showed primary price feeds were functioning. The issue lay in the correlated assets price oracle (CAPO) risk module, a layer that caps the exchange rate of yield-bearing tokens to mitigate sharp oracle-driven swings. CAPO used contract-stored parameters — a reference exchange rate and an associated timestamp — that were not refreshed together. Because those two values diverged, CAPO computed a temporary ceiling below market levels, which produced the ~2.85% undervaluation of wstETH.
Technical breakdown
Chaos Labs pointed to stale parameter values in the CAPO contract: the reference exchange rate and its timestamp became unsynchronized. CAPO’s cap logic treated the stale combination as the safe upper bound, so the protocol’s internal wstETH valuation stayed artificially low until the parameters were updated in tandem.
The liquidation cascade and outcomes
With affected positions pushed below safety thresholds, liquidators repaid parts of borrowers’ debts and seized discounted collateral. Approximately $27 million in borrow positions were liquidated. Liquidators extracted about 499 ETH in combined profits and liquidation bonuses during the event.
Protocol solvency and remediation
Aave reported there was no bad debt — core risk and liquidation processes executed as designed, preserving protocol solvency. Aave governance moved to explore borrower remediation measures using recoveries and DAO treasury support, reflecting an emerging DeFi practice of treating these incidents as infrastructure failures and considering compensations.
Not a Lido failure
Lido contributors confirmed that wstETH and the underlying staking mechanics functioned normally; the problem was in how Aave processed and constrained price and risk-parameter data, not in Lido’s token issuance or staking service.
Lessons for DeFi
This event emphasizes that oracle design and risk-oracle configuration are among the most sensitive components of DeFi infrastructure. Key lessons include:
– Treat dynamic, yield-bearing exchange rates as time-sensitive data that require coordinated updates.
– Implement automated sanity checks and monitoring to detect desynchronized parameters or sudden caps.
– Use robust deployment and parameter-management practices so related values are refreshed atomically or guarded by consistency checks.
– Prepare governance and remediation plans for infrastructure-driven incidents, including clear communication and potential borrower support.
As DeFi expands to include more yield-bearing assets, small misconfigurations or out-of-sync parameters can produce outsized, automated outcomes. Stronger safeguards around oracle layering, parameter synchronization and real-time monitoring will be essential to reduce similar systemic risks going forward.