Worked example · 2024-04

Renzo ezETH liquid-restaking-token depeg and looped-leverage liquidation cascade — Ethereum L1 + multi-chain — 2024-04-24

Loss: approximately $56–60M extracted from over 250 looped-leverage borrowers in a coordinated DeFi liquidation cascade on 2024-04-24, when Renzo Protocol's liquid-restaking-token (LRT) ezETH depegged on Uniswap and dependent venues, falling from approximately $3,200 to as low as $688–$750 in under one hour (a ~79% intra-hour move at the trough) before recovering toward the implied-NAV peg later the same day. The depeg was not an insolvency event — every ezETH was backed 1:1 by ETH staked through validators and restaked through EigenLayer — but Renzo had not enabled native withdrawals at the time of the event, so the only exit path was secondary-market sale, and the looped-leverage borrower cohort relied on near-1.0 ezETH/ETH price ratios to remain above liquidation thresholds. The two largest dependent venues were Gearbox (~115 borrowers liquidated, ~10,650 ezETH valued near $33M, with leverage facilities permitting up to ~10× looping) and Morpho Blue / Morpho (~146 borrowers liquidated, ~$23M of assets), with smaller losses across additional integrating venues. Cumulative liquidation cascade is the headline figure most-cited by Protos / DL News / Cointelegraph / Blockworks at ~$56M+ [protosrenzodepeg2024] [dlnewsrenzo2024] [cointelegraphrenzo2024] [blockworksrenzo2024]. The event remains the canonical 2024 worked example of a liquid-restaking-token depeg cascade driven by withdrawal illiquidity rather than underlying-asset insolvency.
Recovery: none for the liquidated cohort on the on-chain layer — the liquidations were final by block resolution and the borrowers absorbed the realised loss. ezETH itself recovered toward NAV within hours as ETH was redeployed into the Curve pool by arbitrage and as the airdrop-allocation panic-sell pressure dissipated. Renzo subsequently revised the airdrop allocation upward in response to community criticism (~10% additional allocation to the broader user base) [theblockrenzoairdrop2024], but did not compensate the liquidated borrower cohort. The recovery primitive was arbitrage-and-time rather than coordinated-rescue or operator-make-whole; the structural lesson — that LRTs without native-redemption paths are downstream of secondary-market liquidity for any de-risk path — has propagated into LRT-issuer methodology since.
OAK Techniques observed: OAK-T14.003.001(ii) (LST/LRT Depeg-Cascade — Operator-Blocked Redemption — primary; Renzo's restakeMint deposit-side function was active while the withdrawal-side primitive was disabled, producing a one-way valve that forced all sell pressure to clear via secondary-market liquidity) + OAK-T14.003 (Restaking Cascading Risk) — sub-case (b) Liquid-restaking-token (LRT) depeg cascade is the load-bearing classification. The proximate trigger was airdrop-allocation disappointment + Renzo's blocked native withdrawal path, but the propagation channel — secondary-DEX price → lending-market oracle → liquidation engine on dependent looped-leverage positions — is structurally identical to Steakhouse Financial's [steakhouselrt2024] and Carol Alexander et al.'s [alexanderleveragedrestaking2024] LRT-depeg-cascade frameworks. The leverage-amplification path through Gearbox (~10× looping facility) and Morpho (looped vault strategies) is the canonical T14.003 (b) downstream surface. OAK-T9.001 (oracle / pricing-feed dependency) is a contributing failure mode in that the affected lending markets used DEX-spot or DEX-anchored ezETH oracles rather than redemption-rate oracles with slashing-event-delay buffer; this is the specific oracle-methodology mistake Steakhouse Financial's [steakhouselrt2024] LRT methodology had documented in advance. OAK-T14.004 (Liquid Restaking Token Pricing Manipulation) — the incident is the canonical operational anchor for the LRT DEX-price-discovery and oracle-update-cadence sub-primitive: Renzo's ezETH NAV oracle updated on a cadence that lagged the DEX price, creating the stale-NAV window that looped-leverage borrowers were exposed to; the DEX price was the margin-of-solvency determinant for the Gearbox/Morpho borrower cohort, and the NAV-to-DEX-price divergence during the event (NAV ~$3,200 vs. DEX trough ~$688–$750) is the largest documented LRT depeg amplitude at v0.1. OAK-T17.002 (Liquidation-Cascade Engineering) — the propagation channel (secondary-DEX price drop → lending-market oracle → cascading liquidations across Gearbox (~115 borrowers, ~$33M) and Morpho Blue (~146 borrowers, ~$23M)) is the canonical T17.002 surface: a looped-leverage borrower cohort liquidated in cascade when the DEX-spot oracle price crossed below the looped positions' liquidation thresholds; the cascade harvested ~$56M+ at distressed prices. The incident does not classify under T9.005 (no rounding-error / arithmetic bug), under T10 (no bridge involvement), or under T11 (no key compromise). It is one of the cleanest empirical anchors for both T14.003 sub-case (b) and T17.002 at v0.1.
Attribution: pseudonymous — no external attacker; no DOJ / regulator action; no named-individual claim. The depegging force was a normal market response to airdrop-allocation disappointment combined with Renzo's structural absence of a native-redemption path. The liquidation cohort was harmed by a mechanism-design property (looped-leverage on a withdrawal-blocked LRT), not by an attacker. The case is included because the loss was real (~$56M cohort), the cascade structure is the canonical T14.003 sub-case (b) shape, and OAK's framework treats mechanism-design-driven cohort harm as a first-class T14 surface (consistent with the EigenLayer airdrop-and-AVS cohort framing at /examples/2024-05-eigenlayer-restaking-airdrop.md).
Key teaching point: a liquid-restaking-token without a native-redemption path is a forced-exit-via-secondary-market design — and looped-leverage on top of it is an LRT-depeg-cascade waiting to be triggered, with airdrop-allocation panic, oracle-feed lag, or any other liquidity shock as the proximate cause. The April 2024 Renzo case is the empirical anchor that the cascade-amplification surface is operational, not theoretical. The mitigation primitives — (1) LRT issuers shipping native-redemption paths (or transparent withdrawal-queue depths) before integrating with looped-leverage venues; (2) lending-market oracle methodology that incorporates redemption-rate / slashing-event delay rather than pure DEX spot; (3) lending-market LTV ceilings calibrated to a configurable depeg-tail-risk buffer rather than to underlying-LST parameters; (4) borrower-side awareness that the LRT-loop's liquidation threshold is the secondary-DEX price floor, not the implied NAV — are deployable now and were articulated in [steakhouselrt2024] / [gauntletrestaking2024] / [alexanderleveragedrestaking2024] in advance of the April 2024 cascade. The structural OAK lesson is that for LRTs, mechanism-design risk is not theoretical — the April 2024 cohort lost $56M to a known, pre-documented cascade pattern.

Summary

Renzo Protocol launched in late 2023 / early 2024 as a leading EigenLayer-integrated liquid-restaking platform on Ethereum L1. Users deposited ETH (or LSTs) and received ezETH — a liquid-restaking token redeemable for the underlying ETH-plus-restaking-yield, with the underlying stake routed through EigenLayer-secured AVS for additional yield. By April 2024 Renzo had accumulated approximately $3B in TVL, with ezETH integrated into multiple DeFi looped-leverage facilities (Gearbox up to ~10× leverage, Morpho Blue vaults with looped strategies, Pendle yield-trading markets, Curve / Uniswap LP pools).

A critical operational property of Renzo at the time of the April 2024 event: native withdrawals were not enabled. ezETH could be minted by depositing ETH but could not be redeemed for ETH directly — the only exit path was to sell ezETH on a secondary venue (typically Uniswap V3 or Curve) for another asset. This made ezETH's price discovery pure-secondary-market: any imbalance between sell pressure and pool liquidity had to clear through DEX price moves rather than through arbitrage against a 1:1 redemption.

On 2024-04-24, Renzo announced the tokenomics for its upcoming REZ governance token airdrop. The community reaction was overwhelmingly negative: the announced allocation was perceived as substantially smaller than the holder cohort had expected, and a significant portion of the airdrop was structured under transferability restrictions / vesting that further reduced its perceived value. A wave of holders rushed to exit their ezETH positions to capture whatever liquidity remained. Because native redemption was disabled, every exit was a sell on Uniswap or Curve.

In under one hour, ezETH fell from approximately $3,200 to as low as $688–$750 on Uniswap — a ~79% intra-hour drop at the trough. The price action was decoupled from any underlying-asset move: ETH itself was approximately flat. The depeg was a pure-secondary-market liquidity event, driven by sell pressure exceeding the pool's depth-at-price-tolerance.

The propagation into liquidations was immediate. Gearbox had been a major venue for looped-leverage on ezETH, with leverage facilities permitting up to ~10× looping (deposit ezETH → borrow ETH → restake to acquire more ezETH → repeat) for airdrop-points farming. Per DL News and Blockworks reporting, ~115 Gearbox borrowers were liquidated, with ~~10,650 ezETH (~~$33M at pre-depeg pricing) absorbed by liquidators [dlnewsrenzo2024] [blockworksrenzo2024]. Morpho Blue / Morpho absorbed ~146 borrower liquidations across ezETH-collateralised vaults for approximately $23M [dlnewsrenzo2024]. Smaller cohorts were liquidated on additional venues. The cumulative cascade exceeded $56M across more than 250 borrowers.

The LRT itself recovered toward NAV within hours as arbitrageurs deployed ETH into the Curve pool to capture the discount, but the liquidations were already final. Renzo subsequently increased the airdrop allocation modestly in response to community pressure [theblockrenzoairdrop2024], but the liquidated borrower cohort was not compensated.

For OAK's purposes the case is a clean T14.003 sub-case (b) — LRT depeg cascade with looped-leverage liquidation — anchored at empirical scale and with no external-attacker causation. The novel OAK contribution is documenting the cohort's losses as the v0.1 canonical empirical anchor for T14.003 (b), distinct from the design-document framing carried in [vitalikrestaking2023] / [gauntletrestaking2024] / [steakhouselrt2024] / [alexanderleveragedrestaking2024] — which had all predicted the cascade pattern in print before April 2024.

Timeline (UTC)

When	Event	OAK ref
2023-Q4 to 2024-Q1	Renzo launches as EigenLayer LRT issuer; ezETH minted against ETH and LST deposits; native withdrawals not yet enabled; protocol accumulates ~$3B TVL by 2024-04	(T14.003 (b) surface accumulates)
2024-Q1	ezETH integrated into Gearbox (up to ~10× looping facility), Morpho Blue / Morpho vaults, Pendle yield-trading markets, Curve / Uniswap V3 LPs; airdrop-points farming campaign incentivises looped-leverage positions	(looped-leverage cohort builds)
2024-Q1 onward	Steakhouse Financial publishes LRT methodology `[steakhouselrt2024]`; Gauntlet publishes restaking-economy analysis with LRT depeg-cascade framework `[gauntletrestaking2024]`; Alexander et al. publish leveraged-restaking risk paper `[alexanderleveragedrestaking2024]` — all explicitly characterise the cascade pattern that fires on 2024-04-24	(T14.003 (b) prediction-in-print)
2024-04-24 (early UTC)	Renzo announces REZ token airdrop allocation; community reaction overwhelmingly negative on size + transferability restrictions	(proximate trigger)
2024-04-24 (~within hours)	Sell-cascade on Uniswap / Curve: ezETH falls from ~$3,200 to as low as $688–$750 in under one hour; ~79% intra-hour drop at trough	T14.003 (b) cascade fires
2024-04-24 (~within same window)	Looped-leverage liquidations cascade: ~115 Gearbox borrowers liquidated, ~~10,650 ezETH (~~$33M); ~146 Morpho borrowers liquidated, ~$23M; cumulative ~$56M+ across 250+ borrowers	T14.003 (b) downstream extraction
2024-04-24 (~within hours)	Arbitrageurs deploy ETH into Curve / Uniswap pools, capturing the discount; ezETH price recovers toward implied NAV	(market-clearing recovery)
2024-04-24 onward	Detection chain: Protos `[protosrenzodepeg2024]`, DL News `[dlnewsrenzo2024]`, Cointelegraph `[cointelegraphrenzo2024]`, Blockworks `[blockworksrenzo2024]`, AMBCrypto `[ambcryptorenzo2024]`, CryptoSlate `[cryptoslaterenzo2024]` publish converging post-event analyses	(transparency / multi-investigator coverage)
2024-04 (within days)	Renzo modestly increases airdrop allocation in response to community criticism `[theblockrenzoairdrop2024]`; liquidated borrower cohort not compensated	(operator-side response)
2024-04 to 2025-04	LRT-issuer methodology updates; native-redemption paths shipped across the cohort (Renzo, Etherfi, Kelp, Mellow); EigenLayer mainnet slashing activated 2025-04-17 with Unique Stake Allocation `[eigenlabsslashinglive2025]`	(T14.003 mitigation propagation)
Continuing	T14.003 (b) cascade pattern remains operational on any LRT with looped-leverage downstream — the April 2024 Renzo case is the canonical empirical anchor for the surface	(T14.003 (b) anchored)

What defenders observed

The cascade was triggered by a non-economic shock — airdrop-allocation disappointment — but propagated through standard leverage-and-oracle plumbing. ETH itself moved minimally on 2024-04-24; the LRT depegged purely because the secondary market could not absorb the sell pressure quickly enough, and looped-leverage positions cleared through the standard liquidation engine on Gearbox and Morpho. The defender lesson is that the trigger for an LRT cascade is not load-bearing — it can be airdrop disappointment, governance announcement, oracle-feed lag, or any other sentiment-driven sell shock. The propagation channel is the load-bearing surface, and it sits at the lending-market oracle methodology + LTV calibration + LRT redemption-path design.
Native-redemption-path absence is the structural prerequisite for sub-case (b) cascade. Renzo's design choice to delay enabling native withdrawals concentrated all exit pressure on the secondary DEX market, which made the depeg amplitude purely a function of pool depth and sell volume. The defender lesson is that any LRT integration with looped-leverage venues should require native redemption to be enabled (or transparent withdrawal-queue-depth telemetry) as a precondition. Steakhouse Financial's LRT methodology [steakhouselrt2024] had explicitly framed this requirement before the April 2024 event.
Lending-market oracle methodology that anchors to DEX spot is the load-bearing T9.001 contribution to T14.003 (b). The lending markets that liquidated borrowers on 2024-04-24 used oracle methodologies that tracked the secondary-DEX ezETH price closely. A redemption-rate-aware oracle (the underlying ETH-claim per ezETH share, deferred by slashing-event delay) would have produced a much smaller depeg signal and most of the cascade would not have fired. The defender lesson is that LRT-collateralised lending oracle methodology is a first-class T14.003 (b) mitigation surface and should not be inherited from underlying-LST settings.
Looped-leverage facilities at 5×–10× on LRTs concentrate liquidation cliffs at small underlying-price-move thresholds. A position looped at 10× on ezETH/ETH liquidates on a ~5–10% move from peg, depending on LTV / liquidation-discount parameters. The 79% intra-hour Renzo move was orders of magnitude beyond that liquidation threshold, so every leveraged position above a modest leverage band cleared in the same window. The defender lesson is that looped-leverage facilities on LRTs require LTV / liquidation parameters that account explicitly for the secondary-market-only-exit property of the underlying — not parameters inherited from underlying-LST or underlying-stablecoin settings.
The recovery posture (arbitrage + time, no operator-make-whole for liquidated cohort) is the realistic 2024–2025 ceiling for LRT-depeg-cascade incidents. No regulator action; no DOJ involvement (no fraud); no operator-make-whole for the borrower cohort. The recovery primitive that operated was secondary-market arbitrage clearing the depeg over hours, but the liquidations had already finalised. The defender lesson is that contributors writing future T14.003 (b) entries should not over-claim recovery — the realistic outcome for affected borrowers is realised loss, not restitution.

What this example tells contributors writing future Technique pages

T14.003 sub-case (b) is the most empirically anchored T14.003 surface at v0.1, and the Renzo April 2024 case is its canonical worked example. Sub-cases (a) AVS-design slashing-cascade and (c) operator concentration risk and (d) slashing-as-economic-attack remain at the cohort / design-document framing level (see /examples/2024-05-eigenlayer-restaking-airdrop.md); sub-case (b) has a clean $56M+ cohort loss with a documented cascade path. Contributors writing future T14.003 (b) entries should reference Renzo April 2024 as the empirical anchor.
The pre-event prediction-in-print quality of [steakhouselrt2024] / [gauntletrestaking2024] / [alexanderleveragedrestaking2024] is itself a structural OAK datapoint. These three pieces had documented the cascade pattern before April 2024. The case is the empirical anchor that the design-document framing was correct, not over-cautious. Contributors writing future emerging-Tactic worked examples should treat the pre-event prediction-in-print pattern as a valid attribution-of-mitigation-design-credit dimension distinct from post-event forensic-analysis credit.
LRT depeg without insolvency is a distinctive 2024+ surface that should not be conflated with stablecoin depeg. The Renzo case is not an insolvency event — every ezETH had 1 ETH backing. A stablecoin depeg in 2022 (UST collapse) was a solvency event. The 2024 LRT cascade pattern is structurally different and contributors writing future T14 / T9 entries should preserve the distinction explicitly: LRTs depeg from withdrawal illiquidity under sell-pressure shocks; stablecoins depeg from reserve adequacy failures. The mitigation surfaces are different.
Oracle methodology for LRT-collateralised lending markets is a specific T9.001 sub-surface that intersects T14.003 (b). The DEX-spot vs redemption-rate oracle choice is the load-bearing parameter. Contributors writing future T9.001 entries that intersect T14 should document the oracle-methodology choice explicitly and reference the LRT-specific buffer requirements documented in [steakhouselrt2024].
Looped-leverage / yield-loop facilities on LRTs are a distinct category from LST-collateralised lending and should be treated as such. Gearbox's ~10× ezETH looping facility was the dominant liquidation venue for the April 2024 cohort. The defender lesson — and the Mitigations-layer entry — is that looped-leverage on LRTs requires its own LTV / liquidation-parameter calibration, separate from LST-collateralised lending and separate from underlying-stablecoin lending.

Public references

[protosrenzodepeg2024] — Protos, "Depeg of $3B restaking token ezETH causes over $60M in DeFi liquidations" (2024-04); primary cumulative-loss aggregation source.
[dlnewsrenzo2024] — DL News, "Renzo's ezETH falls as low as $700, leading to $56m in liquidations" (2024-04); primary breakdown source for Gearbox / Morpho liquidation distribution.
[cointelegraphrenzo2024] — Cointelegraph, "Renzo's ezETH depegs to $688 following end of airdrop farming window" (2024-04).
[blockworksrenzo2024] — Blockworks, "What was behind the run on Renzo's liquid restaked ETH?" (2024-04); structural-analysis source for the looped-leverage / withdrawal-illiquidity framing.
[ambcryptorenzo2024] — AMBCrypto, "Renzo Crypto rethinks strategy after $60M in liquidations" (2024-04); secondary cumulative-loss source.
[cryptoslaterenzo2024] — CryptoSlate, "Renzo's ezETH token depeg triggers liquidations across DeFi platforms" (2024-04).
[theblockrenzoairdrop2024] — The Block, "Renzo increases airdrop allocation slightly after community outrage and temporary ezETH depeg" (2024-04); operator-response source.
[steakhouselrt2024] — Steakhouse Financial steakLRT methodology and LRT risk-disclosure framework; pre-event documentation of the cascade pattern.
[gauntletrestaking2024] — Gauntlet 2024 restaking-economy analysis; LRT collateral-health methodology and submodular-profit framework.
[alexanderleveragedrestaking2024] — Carol Alexander et al., "Leveraged Restaking of Leveraged Staking: What are the Risks?" (SSRN 2024); pre-event formalisation of the depeg-and-liquidation cascade amplification.
[vitalikrestaking2023] — Vitalik Buterin, "Don't overload Ethereum's consensus" (2023-05); systemic-risk framing for restaking-secured services.
[eigenlabsslashinglive2025] — EigenCloud blog, "Slashing Goes Live on Mainnet" (2025-04-17); EigenLayer mainnet slashing activation and Unique Stake Allocation primitive.

Citations

[protosrenzodepeg2024] — Protos; primary cumulative-loss aggregation.
[dlnewsrenzo2024] — DL News; primary breakdown of Gearbox + Morpho liquidations.
[cointelegraphrenzo2024] — Cointelegraph; primary timeline source.
[blockworksrenzo2024] — Blockworks; structural-analysis source.
[ambcryptorenzo2024] — AMBCrypto; secondary cumulative-loss source.
[cryptoslaterenzo2024] — CryptoSlate; secondary post-event analysis.
[theblockrenzoairdrop2024] — The Block; operator-response source.
[steakhouselrt2024] — Steakhouse Financial LRT methodology.
[gauntletrestaking2024] — Gauntlet restaking-economy analysis.
[alexanderleveragedrestaking2024] — Alexander et al. leveraged-restaking risk paper.
[vitalikrestaking2023] — Buterin restaking-systemic-risk essay.
[eigenlabsslashinglive2025] — EigenCloud mainnet-slashing-activation blog.

Discussion

The Renzo ezETH April 2024 depeg-and-liquidation cascade is OAK's canonical T14.003 sub-case (b) worked example at v0.1 freeze and the empirical anchor for the Liquid-restaking-token (LRT) depeg cascade surface. It pairs with the EigenLayer airdrop-and-AVS cohort case at /examples/2024-05-eigenlayer-restaking-airdrop.md as the two T14.003 anchor examples in the v0.1 corpus: the EigenLayer case anchors the design-and-cohort framing across sub-cases (a)/(b)/(c)/(d); the Renzo case anchors sub-case (b) at empirical scale ($56M+ cohort loss with documented cascade structure).

The case sits at the intersection of three OAK Tactics. Primary: T14.003 (b) — LRT depeg cascade. Contributing: T9.001 — oracle / pricing-feed dependency, where DEX-spot anchored oracles propagated the depeg amplitude into liquidation triggers without a redemption-rate buffer. Adjacent: T14.001 — operator-side staking failures, in the sense that the absence of a native withdrawal path is an operator-design choice that converts an LRT into a forced-secondary-market-exit token. Contributors writing future cross-Tactic T9 ↔ T14 mappings should treat the Renzo case as a clean example of how T9.001 oracle-methodology choices become the load-bearing T14.003 amplifier.

The pre-event prediction-in-print quality of [steakhouselrt2024] / [gauntletrestaking2024] / [alexanderleveragedrestaking2024] is the analytically distinctive feature at the corpus level. All three pieces had documented the cascade pattern in print before April 2024 — in some cases by months. The Renzo April 2024 case is the empirical anchor that the design-document framing was correct, the cascade is operational, and the deployable mitigation primitives (native-redemption-path, redemption-rate-aware oracles, LRT-specific LTV ceilings) are not over-cautious. This is one of the cleanest cases in the OAK v0.1 corpus where defender-side design-document prediction preceded attacker-or-mechanism-driven realisation by a meaningful margin.

The recovery posture — arbitrage and time, no operator-make-whole, no regulator action — is the realistic 2024–2025 ceiling for LRT-depeg-cascade incidents and contributors should expect the same posture for future T14.003 (b) cohort cases. The mechanism-design surface is structural to the LRT primitive; the cascade fires on any sentiment shock + withdrawal-illiquidity combination; the loss accrues to the looped-leverage borrower cohort whose lending-venue oracles propagated the depeg into liquidation triggers.

For OAK's broader credibility, including the Renzo April 2024 case as a T14.003 anchor closes a gap: the T14.003 Technique was rated emerging at v0.1 because the canonical loss incident did not exist in the corpus. The Renzo case operationalises sub-case (b) into a live empirical anchor and provides contributors writing future T14.003 entries (Bedrock September 2024 oracle-mint exploit, KelpDAO April 2026 RPC-poisoning bridge attack, future LRT cascade events) with a tone-and-structure precedent for the cascade-framed worked-example treatment.