Aave Liquidation Risk: The Trade That Hasn't Been Tested

This isn't a story about Aave being broadly overleveraged. It's a story about one specific, repeated bet. Roughly 9% of positions on the protocol are responsible for about half of all outstanding debt, and nearly all of that debt traces back to the same mechanical loop: deposit a liquid-staking token like weETH, wstETH or rsETH as collateral, borrow WETH against it, restake that WETH into more liquid-staking tokens, and deposit those as collateral again. Repeat a few times and a modest starting stake turns into a position running roughly 10.7x debt-to-equity, all of it correlated to the same underlying asset moving the same direction.

The appeal is straightforward: liquid-staking tokens earn ETH staking yield, and Aave's E-mode lets correlated assets borrow against each other at higher loan-to-value ratios than it would allow for unrelated collateral, precisely because the two sides of the trade are supposed to move together. weETH alone makes up about 42% of this cohort's collateral. The trade works exactly as designed, right up until the correlation it depends on breaks.

The mechanism is what makes this different from ordinary leveraged exposure. A normal leveraged long gets liquidated when the underlying asset's price falls far enough. This trade doesn't need ETH to fall at all — it needs the wrapper token to stop trading at parity with the ETH it represents. Liquid-staking tokens like weETH are supposed to track ETH's value closely, adjusted for accrued staking yield, but that peg is maintained by market liquidity and redemption mechanics, not a hard guarantee.

The positions in this cohort are sitting at an average health factor of roughly 1.05 to 1.06 — Aave's liquidation threshold is 1.0, so there is almost no cushion. If wrapper tokens trade at an 8-10% discount to the ETH they're meant to represent, a large share of this cohort crosses below 1.0 at the same time. Because these positions are similar in construction and collateral makeup, they don't liquidate one at a time — they liquidate in clusters, and each forced sale of a discounted wrapper token into thin liquidity pushes the discount wider, which triggers the next round. That reflexive loop, not ETH's absolute price, is the actual liquidation trigger.

Two distinct paths get there. The more obvious one is a genuine ETH drawdown sharp enough to strain liquid-staking token liquidity — profit-taking after this week's squeeze, or a risk-off shock around next week's Jackson Hole Fed symposium (August 27-29), where markets will be parsing the tone for signs the current liquidity-driven rally is overextended. A sizable reversal doesn't need to break the wrapper peg directly; it just needs to be fast enough that liquidity providers pull back and the discount opens on its own.

The second path doesn't need a price move at all. In March 2026, a stale oracle parameter — not a real depeg, not a market crash, just a pricing feed that fell out of sync — triggered roughly $26-27 million in wstETH liquidations on Aave. Nothing about ETH's price was actually wrong; the infrastructure reporting it was. That precedent matters because it shows the mechanism can fire on a technical failure just as easily as a market one, and the current cohort is carrying meaningfully more concentrated debt than it was back then.

There's also a governance variable in play. Chaos Labs, Aave's longtime risk-parameter provider, exited earlier this year, handing primary responsibility for setting loan-to-value and liquidation-threshold parameters to LlamaRisk and Aave's internal risk stewards. They've already acted once this month — adjusting E-mode parameters on pendle-token (PT) markets in mid-August — which suggests the new stewardship is paying attention. Whether it can move fast enough during an actual liquidity event, rather than a routine parameter tweak, hasn't been tested.

Because the one thing that would stress-test this trade — a real ETH pullback — hasn't shown up. ETH's rally since Wednesday's breakout has done the opposite of what would expose the risk: it has kept climbing, past $2,000 and now above $2,500, rewarding the recursive trade rather than punishing it. In that environment, the leveraged cohort has little reason to unwind and every incentive to keep growing, since restaking yield compounds as long as the position survives. The risk here isn't shrinking during the rally — if anything, it's likely still building as more capital chases the same yield.

That's the core asymmetry of this setup. It isn't visible in daily price action because nothing about it shows up as a chart pattern or a funding-rate anomaly. It sits dormant in the health-factor distribution of a specific cohort of large accounts, invisible until the one condition that exposes it actually occurs.

While ETH holds its current uptrend or even consolidates sideways, there's no reason to expect a cascade — the wrapper-to-ETH discount that triggers mass liquidation simply doesn't open under calm conditions, and the trade likely keeps compounding rather than unwinding on its own. That's the more probable near-term path.

The scenario that breaks it is a discrete event, not a slow grind: either an ETH drawdown sharp enough to strain liquid-staking token liquidity, or another oracle or liquidity glitch along the lines of March's. Jackson Hole is the nearest calendar risk that could supply the former. If a wrapper discount does open, the scale of forced liquidations this time would likely exceed March's $26-27 million, given the cohort's larger size and higher leverage today. Whether Aave's risk stewards tighten E-mode parameters before that happens, rather than reacting after, is the main variable standing between a contained event and a genuinely disorderly one.