The January 31st Crash
Yesterday, the crypto markets experienced their most violent stress event since the October 2025 flash crash. A convergence of geopolitical escalation in the Middle East, a temporary United States government funding lapse, and a historic drawdown in precious metals such as gold or silver cascaded into the crypto markets during thin weekend trading hours. Bitcoin fell from $84,000 to below $77,000 and Ethereum dropped to $2,300. Within hours, more than $2.2 billion in leveraged positions were liquidated across centralized and decentralized venues.
Aave v3 absorbed the shock and by any conventional risk metric, the system performed as designed. PERFECT.
Over $140 million in liquidations were processed, primarily in Stables, ETH, WBTC, and wstETH markets, without triggering protocol wide insolvency. With Net of liquidation fees, Smart Value Recapture revenue, and minor isolated deficits, the protocol ended the week profitable by repaying $147 million of user debt and distributing $7.14 million in liquidation bonuses according to Chaos Labs.
And yet, beneath this apparent success lay a more subtle pathology. Hundreds of positions did not liquidate and they did not recover either. Instead, they hovered in a narrow band just above the liquidation threshold, with health factors between 0.95 and 1.0, some for hours, some for days.
Today, I took a look at Chaos Labs amazing Dashboard and I wanted to write this small blog about it. I am arguing that this liminal state is not an edge case but a structural feature of modern DeFi lending markets, because of that, it requires its own analytical framework and its own protocol response especially as Aave is aiming at $100B+.
I call this framework “Zombie Positions”.
Chaos Labs data shows that this tail is numerically large but balance-light, suggesting the dominant unresolved risk is behavioral persistence near HF ≈ 1.0, not hidden insolvency. These positions generated no liquidation revenue, no interest upside, and no productive deployment of collateral. They demanded constant oracle updates, continuous health checks, and governance attention.
The Anatomy of a Zombie Position
Aave’s health factor is simple math: Collateral value multiplied by the liquidation threshold, divided by the debt value.
But between 0.95 and 1.0 lies a dead zone. Borrowers can’t withdraw collateral (health factor too low). They can’t add leverage (too risky). Many lack the cash to repay. So they wait, hoping prices recover.
Meanwhile, the protocol is trapped with them. Oracles must update more frequently, sometimes every 10-15 minutes instead of hourly. Liquidation bots burn gas watching positions they cannot profitably execute. Capital sits frozen, producing nothing while consuming infrastructure. Chaos Labs’ dashboards show thousands of wallets clustered between zero point nine and zero point nine nine across Ethereum, Arbitrum, and Polygon.
Why 0.97 Is More Dangerous Than 0.90
Liquidated positions are resolved risk while Zombie positions are stored risk.
When your health factor hits 0.90, bots liquidate you instantly. The loss is realized, the books close, everyone moves on. But at 0.97, you’re in the “uncanny valley” of Aave, close enough to solvency to seem saveable, close enough to liquidation to threaten cascade failure with every price tick.
During Saturday’s crash, gas fees spiked above 400 gwei. A position at 0.97 health factor with a standard liquidation bonus wasn’t worth the gas costs plus oracle uncertainty plus MEV competition. So liquidators passed and the protocol entered a high-cost holding pattern.
Zombie positions impose hidden taxes on the system. Volatile collateral requires expensive, high-frequency updates for positions generating zero revenue, Collateral sits unproductive, neither liquidated nor reinvested.
Evidence of the Zombie Zone (HF 0.90–1.00)
Chaos Labs’ wallet table shows eligible liquidation amounts per wallet even when HF > 0.9. Individually small, but collectively persistent across thousands of wallets during stress windows. These wallets are not liquidated and are one small price move away from forced liquidation. We can now frame the zombie mass without speculation. Zombie mass is the population of positions that remain solvent (HF > 1.0) but economically inert, clustered near liquidation thresholds, and persist through volatility without being liquidated or deleveraged.
Cost of the undead
Why Zombie Positions Matter More Than Liquidations
If zombie positions are structural, governance must treat them as such. The mistake is to pretend that every position should either be saved or liquidated immediately. Reality is messier and some positions deserve a brief attempt at recovery while others should be resolved quickly to conserve resources. A few may need to be maintained for non economic reasons such as legal constraints or strategic relationships.
The January 31st event showed both sides. Trend Research’s Ethereum position demonstrated that timely capital injection can preserve enormous value when the borrower is capable and responsive. At the same time, dozens of marginal long tail positions consumed resources for days only to end in partial deficits that could have been resolved earlier at lower cost.
Aave v4: Reinvestment Module as Life Support
Aave v4’s reinvestment architecture introduces a new variable into this calculus. Yield bearing collateral such as stETH can, under certain conditions, offset borrowing costs. In rare cases, a zombie position can heal itself if reinvestment yield exceeds interest accrual.
But, If borrowers believe that hovering near liquidation is survivable, initial leverage increases and risk shifts from borrower to protocol. Positions that will survived did or will do so either because they are large and well managed. Just like Trend Research did yesterday.
Aave v4: New Liquidation System
Aave v4 changes liquidations from a blunt, binary event into a controlled risk-resolution mechanism. Instead of liquidating a fixed portion of a position once health factor drops below 1, v4 liquidations are sized to restore a target health factor defined by governance. This means positions are only liquidated as much as necessary, avoiding over-liquidation while still forcing early action. At the same time, liquidation incentives become dynamic: the further a position falls below safety, the higher the bonus for liquidators.
This directly fixes v3 zombie mode where positions hovered near liquidation because gas costs and static rewards made execution unprofitable. This makes early, partial liquidations viable even during volatility, reducing the formation of zombie positions that linger at 0.95–0.99 health factors. Instead of forcing governance to babysit marginal accounts or wait for catastrophic moves, v4 resolves risk incrementally, cheaply, and deliberately.
Conclusion
The January 31st event confirmed that Aave v3 works: liquidations ran as expected and oracles performed reliably. But it also exposed a blind spot. Some positions, whether solvent or liquidated, safe or risky, can persist longer than intended.
These “zombie” positions are not bugs. They are a natural result of discrete thresholds in continuous markets. Ignoring them leads to wasted resources, stalled governance, and hidden risks.
Managing them goes beyond parameter settings. Like living systems, positions follow lifecycles: some rebound, some must be removed, and some persist beyond their risk threshold. A robust liquidation system needs to address each stage.
Thank you and have a great weekend.