# Peg defence without a custodial reserve

Endogenous collateral and synthetic-dollar stress, January 2023–October 29, 2026.

**Research/read date: October 29, 2026.** Prepared for designers of a small Ethereum stablecoin adding collateral branches, borrower-set rates and collateral redemptions. No imdUSD parameters are supplied or inferred.

Evidence convention: **Fact** means a documented mechanism or attributed observation, including an explicitly identified issuer or analyst report. It does not imply independent verification. **Inference** marks analysis and proposed design implications. **Uncertain** marks a conflict or missing evidence. Claim identifiers map directly to `data.json`; figures there retain units, dates, chains/venues and sources. Sources were consulted on the research date; inaccessible originals and indirect access are identified.

Primary sources take precedence. Contemporary secondary reporting and mirrored issuer statements are marked where originals could not be read. Mutable documentation is the version retrieved on the read date, unless a dated historical proposal or incident report is specified. Governance proposals are not silently treated as executed votes. Most on-chain figures below are source-reported, not independently replayed; the USDD API snapshot is a direct, non-atomic retrieval. Reported prices are scoped to the instrument, venue and sampling stated. The earlier 2020–21 campaign is outside this review.


## 1. Findings and boundaries


**Inference [C001].** A small stablecoin can defend its peg without warehousing another issuer’s dollar token by letting holders cancel debt in exchange for saleable collateral. Interest rates, protocol-owned liquidity and swap fees can support that exit. They cannot replace its economic value. The cases below support combining a credible contraction path with strict issuance limits; they do not support treating a local AMM quote, a high accounting collateral ratio, or a transaction surcharge as proof of dollar redeemability. [SIP-421: sUSD](https://sips.synthetix.io/sips/sip-421/), [LlamaRisk: crvUSD](https://www.llamarisk.com/research/crvusd-incident-report-20240612), [Resolv official](https://resolv.xyz/blog/resolv-postmortem-march-22-2026-incident), [Liquity v2](https://docs.liquity.org/v2-faq/redemptions-and-delegation).


**Inference [C002].** “No custodial stablecoin reserve” needs three separate tests: whether the stablecoin holds such tokens directly, whether its peg liquidity depends on them, and whether its collateral can be liquidated without them. Curve’s keepers pass the first test narrowly but depend on paired stablecoins inside LP positions. GHO’s GSM fails that test by design. Frax’s AMOs often deploy stablecoin collateral, while frxUSD explicitly uses custodians. These are useful comparators, not interchangeable examples of reserve independence. [Curve PegKeepers:](https://docs.curve.finance/developer/crvusd/pegkeepers/overview/), [Aave GSM](https://aave.com/help/gho-stablecoin/stability-module), [Frax Curve](https://docs.frax.finance/amo/curve), [Frax USD:](https://docs.frax.com/protocol/assets/frxusd/frxusd).


**Inference [C003].** Endogeneity also has degrees. SNX is issued by the same protocol whose liabilities it backs, creating direct exposure to that protocol’s reputation and revenue. ADA and TRX are network tokens with uses beyond any one stablecoin. They meet the question’s wider “own ecosystem” category, but their examples cannot establish the safety of a thin governance token. SHEN is a junior claim on ADA reserves, rather than an independently valuable second reserve asset. [SIP-420: Protocol](https://sips.synthetix.io/sips/sip-420/), [COTI: clarification](https://medium.com/cotinetwork/answers-and-clarification-to-recent-djed-community-questions-95176450db09), [Djed Alliance:](https://docs.stability.nexus/stablecoins/djed-overview/how-the-protocol-works), [USDD legacy](https://legacy-docs.usdd.io/usdd-issuance/usdd-issuance).


## 2. Peg keepers, protocol-owned liquidity and hooks


## 2.1 crvUSD: the contraction inventory matters as much as the mint ceiling


**Fact [C004].** On Ethereum, PegKeeper v1 adds allocated crvUSD to a paired stablecoin pool when crvUSD is scarce and receives LP tokens. Below peg it can remove crvUSD through those LP tokens and reduce recorded debt. The factory pre-mints the allocation: “mint and burn” in economic descriptions means moving tokens into and out of circulation, not unrestricted token issuance by the keeper. Governance sets the factory debt ceiling. Keeper LP positions do not earn gauge CRV emissions; trading profits fund the caller incentive. [Curve PegKeeperV1](https://docs.curve.finance/developer/crvusd/pegkeepers/peg-keeper-v1/).


**Fact [C005].** Curve documents two v1 weaknesses. An adversary could consume the 15-minute action window with a tiny update near parity, delaying useful intervention. More seriously, a paired stablecoin depeg could induce issuance and then prevent the keeper from recovering its crvUSD debt. Its apparent backing was the pool claim. It did not have a separate asset that guaranteed recovery if the paired token lost value. The published weaknesses are design risks, not evidence that every attack described was executed. [Curve PegKeepers:](https://docs.curve.finance/developer/crvusd/pegkeepers/overview/).


**Fact [C006].** A concrete v1 exposure episode was TUSD in January–February 2024. Curve’s February 12 proposal records an earlier ceiling reduction from 25 million to 5 million crvUSD, TUSD trading around $0.98–0.99 over the preceding month, and keeper deposits reaching the cap during January 18–21. It proposed a zero ceiling, explicitly warning that this would not immediately retire outstanding debt: the pool first had to rebalance. Stopping new risk and recovering old risk were different operations. [Reduce TUSD](https://gov.curve.finance/t/reduce-tusd-pegkeeper-debt-ceiling-to-0/9952).


**Fact [C007].** V2 retains one-sided LP intervention but adds a Regulator. Expansion requires the aggregate price above $1, an acceptable spot-versus-EMA deviation, acceptable relative paired-asset prices, and allowance under debt-distribution rules. Available crvUSD, the factory ceiling, action delay, positive execution profit and pool imbalance also constrain an update. The embedded update code operates on one fifth of the imbalance, subject to allowance. Withdrawal still consumes previously acquired LP backing; a large expansion cap is not an equally large unconditional buyback budget. [Curve PegKeeperV2](https://docs.curve.finance/developer/crvusd/pegkeepers/peg-keeper-v2/), [Curve PegKeeperRegulator](https://docs.curve.finance/developer/crvusd/pegkeepers/peg-keeper-regulator/).


**Fact [C008].** The May 30, 2024 Ethereum migration proposal set old keeper ceilings to zero and specified v2 ceilings of 25 million crvUSD each for USDC and USDT, 15 million for PYUSD and 10 million for TUSD. Those are dated proposed allocations, not October 2026 live limits. A March 21, 2025 proposal sought 50 million each for USDC and USDT. Both documents explain that capacity responds to demand for crvUSD and the composition of its liquidity, rather than being a universal reserve ratio. [PegKeeper v2](https://gov.curve.finance/t/pegkeeper-v2-upgrade-add-pkv2-usdc-25m-usdt-25m-pyusd-15m-tusd-10m/10127), [Increase USDC](https://gov.curve.finance/t/increase-usdc-and-usdt-pegkeeper-debt-ceiling-to-50m-crvusd-each/10546).


**Fact [C009].** V2’s safety gate failed operationally during the June 12, 2024 upward depeg. LlamaRisk’s contract-linked investigation found the spot/EMA deviation exceeded the permitted ±0.0005, preventing supply precisely during liquidation demand. The first successful updates in the three examined pools came roughly 525 blocks, or 45 minutes, after onset. The stronger crvUSD quote also made collateral appear cheaper in crvUSD-denominated lending markets, causing spillover liquidations. This is measured intervention latency; the report does not supply a single reliable consolidated peak price to reproduce here. [LlamaRisk: crvUSD](https://www.llamarisk.com/research/crvusd-incident-report-20240612).


**Fact [C010].** The later record includes a successful full monetary cycle. Curve’s August 2026 recap reports keeper debt rising from $33.75 million on August 18 to $108.2 million on August 24, then falling to $42.2 million at month-end as borrower supply expanded. Reported hourly crvUSD prices averaged $0.99991 and ranged $0.99966–$1.00061. This is issuer-published Ethereum system evidence with hourly sampling, not a tick-by-tick audit or proof that keepers alone caused the result. [Curve monthly](https://news.curve.finance/curve-monthly-recap-august-2026/).


**Fact [C085].** Pangea’s December 30, 2025 study gives a less favorable interval: TVL-weighted mean absolute pool-price deviation rose from 0.000186 to 0.000309 across equal pre/post-YieldBasis periods. It expressly does not isolate causality from market volatility. Incentivized PYUSD pool TVL rose from about $2 million to $30 million, then returned just above $2 million as rewards tapered. Its end-of-block Ethereum analysis supplies measured evidence that incentives rented liquidity; it does not identify a surcharge effect or contradict the differently sampled August 2026 result. [Pangea: crvUSD](https://blog.pangea.foundation/does-yieldbasis-hurt-the-crvusd-peg/).


**Fact [C011].** The September 2026 v3 discussion is still a proposal in the retrieved version and explicitly says V3.4 is not deployed. It would add independent backing-asset oracle guards, separate entry/exit economics, velocity limits and better yield-token accounting. Its existence is evidence of remaining v2 design tradeoffs, not evidence that the proposed protections had defended a live peg. [Migrate PegKeepers](https://gov.curve.finance/t/migrate-pegkeepers-to-a-new-version-3/11166/1).


## 2.2 GHO: overcollateralization did not create a redemption bid


**Fact [C012].** GHO launched without its GSM. The amended genesis proposal considered a $50 million facilitator bucket, a 2% borrow rate and a 25% stkAAVE borrowing-rate discount. That discount is distinct from the market discount of GHO below a dollar. Borrower collateral constrained issuance, but holders could not demand $1 of borrower collateral simply because the token traded cheaply. The genesis document expressly excluded the not-yet-live GSM from launch assumptions. [GHO genesis](https://governance.aave.com/t/arfc-gho-genesis-parameters/12151).


**Fact [C013].** Chaos Labs records GHO falling from $0.962 to $0.946 during October 26, 2023, 09:15–10:15 UTC, before recovering to its earlier discounted level. Its main Ethereum Uniswap v3 GHO/USDC pool had lost about $2.5 million of USDC between September 23 and October 6; USDC inventory was below 100,000 by October 17. Gauntlet separately observed that a September rate increase briefly helped before price decay resumed. Recovery from a liquidity shock was not restoration of the dollar peg. [Chaos Labs:](https://governance.aave.com/t/gho-depeg-and-liquidity-analysis/15250), [Gauntlet: community](https://governance.aave.com/t/temp-check-community-plan-for-gho-stability-and-peg/15252).


**Fact [C014].** The documented response combined rate increases, demand incentives and module construction. A December 2023 proposal sought a quoted 6.35% annual borrow rate, while January 30, 2024 governance discussion attributed significant price improvement to stkGHO and described GHO as at peg. Aave’s changelog dates the GSM introduction to January 29. These dates establish overlapping interventions; they do not identify a clean causal contribution for each one, or show that a rate hike alone restored par. [Increase GHO](https://governance.aave.com/t/arfc-increase-gho-borrow-rate-100-bps-to-6-35-on-aave-v3/15744), [GHO Stewards](https://governance.aave.com/t/arfc-gho-stewards/16466), [Aave changelog:](https://www.aave.com/docs/resources/changelog).


**Fact [C015].** The GSM gives an actual stablecoin exit: below peg, buy GHO and burn it for an approved reserve asset, less fees, while inventory lasts; above peg, deposit that asset, mint GHO and sell it. Exposure caps limit external-asset holdings; facilitator buckets constrain issuance; oracle freezes can disable conversion. Funding comes from stablecoins deposited by minters and potentially DAO capital. Early risk recommendations proposed $3.5 million per USDC/USDT module and 0.2% fees, conditioned on an already much stronger peg. These recommendations are not deployed limits. [Aave GSM](https://aave.com/help/gho-stablecoin/stability-module), [GSM development](https://governance.aave.com/t/gho-stability-module-update/14442).


**Fact [C016].** Actual May 2, 2024 funding guidance described caps of only 500,000 USDC and 500,000 USDT and proposed DAO allowances of one million of each, drawn from Aave v3 as needed. January 2025 records show the USDT module holding its full $8 million exposure cap and a proposal to raise it to $16 million, with a separate 24 million GHO bucket. The same thread reports USDC-module emptying coincident with July and November 2024 downward peg events. Capacity exhaustion remained a relevant limit after the launch discount had improved. [Fund USDC](https://governance.aave.com/t/arfc-fund-usdc-usdt-gsm/17566), [USDT GSM](https://governance.aave.com/t/arfc-usdt-gsm-bucket-and-exposure-cap-increase/20645).


**Fact [C017].** The February 2026 onboarding review provides a bounded peg measurement: in its preceding twelve months of hourly Ethereum Uniswap v3 GHO/USDC observations, maximum negative deviation was 37 basis points and positive deviation 18.5 basis points. With ten basis points as the recovery threshold, 105 episodes had a three-hour median recovery, but the two longest mild drifts lasted 328 and 379 hours. This narrower later peg is real reported evidence; median recovery should not conceal long tails or be extrapolated to every venue. [LlamaRisk and](https://llamarisk.com/research/pegkeeper-onboarding-review-aave-s-gho).


**Fact [C018].** Later reserve capacity should not be read from the old module addresses. The September 2026 Ethena committee assessment tracks successive GSM migrations and reports roughly $68 million redeemable across five modules, about $27 million on Ethereum and $41 million on Plasma, versus approximately $354 million of mint capacity. Those are the authors’ dated, partly independently checked measurements, not this report’s chain replay. The practical lesson is in the asymmetry: mint headroom, wrapped reserve NAV, bridge throughput and immediately withdrawable dollars are different quantities. [Ethena Risk](https://gov.ethenafoundation.com/t/gho-and-sgho-as-eligible-backing-assets-for-usde/847).


## 2.3 Frax: monetary operations survived by changing the balance sheet


**Fact [C019].** Frax’s Curve AMO deploys existing collateral and newly created FRAX into liquidity, retaining the LP claim; the Uniswap v3 AMO similarly concentrates liquidity near the desired rate. Above peg, adding FRAX expands available supply. Below peg, the AMO can withdraw and burn FRAX or deploy balance-sheet assets to purchase it. Trading fees and external LP incentives provide income. This is funded market making: the counterasset and the recoverable LP value are real resources, not resources created by the word “algorithmic.” The original Curve AMO page now marks that implementation deprecated. [Frax Curve](https://docs.frax.finance/amo/curve), [Frax Uniswap](https://docs.frax.finance/amo/uniswap-v3).


**Fact [C020].** The v3 AMO documentation retains Curve supply contraction/expansion, Fraxlend issuance against approved borrowers, and TWAMM purchases or sales of collateral. It also describes an Aave AMO able to borrow USDC to buy discounted FRAX, and says many Curve operations moved to the comptroller multisig. Caps are strategy- and governance-specific. For a dated example, FIP-415 proposed a $50 million total borrowing limit in November 2024. That proposal is not a universal AMO mint cap, and borrowed defence adds a liability with its own unwind requirements. [Frax v3](https://docs.frax.finance/frax-v3-100-cr-and-more/amos), [FIP-415: Creating](https://gov.frax.finance/t/fip-415-creating-frax-borrow-amo/3513).


**Fact [C021].** FIP-188, posted February 15, 2023, reported a 92% collateral ratio and proposed permanently targeting 100% external collateral, retiring decollateralization, retaining earnings and deferring FXS buybacks. It explicitly rejected minting new FXS to reach that target. Thus the approximately 8% algorithmic portion at that date was not a vault containing 8% FXS cash backing. The proposed repair reduced dependence on future native-token liquidity; it did not demonstrate that native-token backing had become robust. [FIP-188: Increase](https://gov.frax.finance/t/fip-188-increase-cr-to-100/2147).


**Fact [C022].** During the March 2023 USDC shock, CryptoCompare’s market-data report records FRAX reaching $0.920 on March 11 and returning toward parity in the subsequent recovery. This is an aggregated market observation, not an identified Ethereum trade. It contradicts the idea that protocol-owned liquidity isolated FRAX from its reserve assets. The later v3 documentation explicitly distinguishes a USD reference target from direct holder redemption: it describes FRAX as non-redeemable for any specified asset at any specified time. [CryptoCompare March](https://www.cryptocompare.com/media/44081939/stablecoin-report-march-2023.pdf), [Frax v3](https://docs.frax.finance/frax-v3-100-cr-and-more/overview).


**Fact [C023].** frxUSD changes that proposition. Current official documentation describes one-for-one cash-equivalent backing, custodian-specific minting caps and redemption against available custodian assets. BUIDL’s December 2024 onboarding proposal identifies regulated fund, transfer-agent and custody arrangements, including investor restrictions. Above peg, eligible users can create frxUSD against reserves; below peg, eligible redemption and arbitrage can retire supply, subject to asset availability and access requirements. The cash/Treasury capital comes from reserve subscriptions, not from selling an endogenous governance token. It is a custodial-reserve comparator. [Frax USD:](https://docs.frax.com/protocol/assets/frxusd/frxusd), [FIP-418: BUIDL](https://gov.frax.finance/t/fip-418-buidl-as-a-backing-asset-for-frax-usd/3533).


**Fact [C084].** LlamaRisk’s December 13, 2025 Ethereum review reports frxUSD briefly reaching $0.9026 on Curve on June 26 as sell-side liquidity depleted; the Chainlink low was only $0.990154. It reports December 12 capacity to swap $26.3 million frxUSD into USDC at 7.5% impact, not at par. Its custodian table lists mint caps of 50 million frxUSD for USTB, 200 million for USDC, and 0.1 million each for WTGXX, BUIDL and USDB. These are dated reported limits. Thus custodial backing and access to primary redemption did not prevent a severe local liquidity dislocation. [LlamaRisk: frxUSD](https://governance.aave.com/t/arfc-onboard-frxusd-to-aave-v3-core-instance-on-ethereum/23513/9).


**Fact [C024].** The Frax Facts extract last updated October 14, 2026, 02:23 UTC reports $107,403,807 of frxUSD assets against $104,417,171 of circulating liabilities on Ethereum and Fraxtal, or 102.9% coverage. It lists tokenized government-money assets and cash. This is a protocol balance-sheet figure retrieved through the indexed dashboard, not an independent reserve audit or a price measurement. Independent price-series replay and AMO-specific causal attribution remain unavailable. [Frax Facts](https://facts.frax.finance/frxusd/balance-sheet).


## 2.4 Uniswap v4: a real deployed fee design, with a different objective


**Fact [C025].** Uniswap v4 supports per-swap LP-fee overrides in dynamic-fee pools. StablePair, launched September 10, 2026, supplies a concrete Ethereum implementation for USDC/USDT and USDC/USDG. Inside its reference band it adjusts fees to quote a consistent bid and ask before price impact. Outside the band it charges zero on trades moving farther from reference and auctions the corrective trade through a decaying fee. Below and above reference the directions reverse. This deliberately captures arbitrage value for LPs; it is not a surcharge on every depegging sale. [Uniswap v4](https://developers.uniswap.org/docs/protocols/v4/concepts/dynamic-fees), [Uniswap Labs](https://blog.uniswap.org/stablepair-hook-a-fee-that-moves-with-the-market), [StablePair dynamic](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/concepts/dynamic-fees).


**Fact [C026].** StablePair is funded by LP deposits and swap payments. Its documentation describes governance-controlled fee configurations and an upgradeable implementation, with permanent hook permissions that cannot block LP withdrawals. The February 2026 fee-mechanism audit found a swap-splitting advantage; the response was to cache price once per block. That introduces accepted intra-block staleness and possible direction inversion when trades cross reference. The audit covered a non-upgradeable predecessor’s fee mechanism, not every later role and upgradeability decision. [StablePair deployments](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/deployments), [StablePair audit](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/security).


**Uncertain [C027].** I found no published, reproducible StablePair event study showing fewer out-of-band minutes, a smaller maximum depeg, or aggregate cross-venue retention caused by its fee logic through October 29, 2026. The documentation specifies a configured band, decay and fee floor, but the deployment page does not provide their numerical live values. A fee-only hook has no independent mint or redemption budget; LP inventory bounds what can be bought. Claims of rapid correction in launch material are design claims, not measured peg guarantees. [Uniswap Labs:](https://blog.uniswap.org/keeping-value-in-the-pool-how-stablepair-hook-works), [StablePair deployments](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/deployments), [StablePair dynamic](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/concepts/dynamic-fees).


## 3. Own-token and own-network collateral since 2023


## 3.1 sUSD: the 2025 transition, temporary support and 2026 retirement


**Fact [C028].** SIP-420 proposed delegating SNX staking and transferring debt to a protocol-managed pool over twelve months. Its specified pooled issuance ratio was 200%, with solo staking pushed toward 1,000%; Chaos Labs describes the operational transition as a reduction from 500% to 200%, alongside removal of the liquidation process. These are different categories of ratio, not a historical proof that aggregate backing instantly became exactly 200%. The pooled collateral was SNX, so this compartment’s pledged-token composition was entirely endogenous by design. [SIP-420: Protocol](https://sips.synthetix.io/sips/sip-420/), [Chaos Labs:](https://governance.aave.com/t/arfc-further-deprecate-susd-on-aave-v3-optimism/21770).


**Fact [C029].** The April 11 risk review reported 147 million migrated SNX, valued at $92 million, and approximately 7.1 million sUSD of newly protocol-managed debt. It separately reported 53 million sUSD of migrated debt, including five million already forgiven, and a recent sUSD low of $0.83. These separate debt categories prevent treating $92 million divided by $7.1 million as the whole system’s true coverage. They also locate the depeg while substantial marked SNX collateral still existed. [Chaos Labs:](https://governance.aave.com/t/arfc-further-deprecate-susd-on-aave-v3-optimism/21770).


**Fact [C030].** Synthetix’s own April 24 SIP-421 identifies the missing demand mechanism: debt forgiveness let participants dispose of their sUSD without buying it back to finish repaying debt. The repair offered five million SNX over twelve months, locked deposited sUSD, and required debt-jubilee participants to hold 10% of original debt in the sUSD pool. The proposal dates rewards from April 19. This was a new demand incentive and constraint on debt relief, financed in the same ecosystem token. [SIP-421: sUSD](https://sips.synthetix.io/sips/sip-421/).


**Fact [C031].** The May 30 issuer update acknowledged selling toward $0.70. It reported the initial holding requirement bringing sUSD back to $0.96, but said $2 million of treasury buying and extended campaigns had not completed the move to par. It increased the required sUSD deposit to 20% effective June 2. The same page contains optimistic language about a restored peg alongside its discussion of an unfinished recovery; the dated figures support a partial repair, not a durable success finding. [Synthetix: The](https://blog.synthetix.io/the-repeggening/).


**Fact [C082].** The temporary 2025 repair did not establish durable recovery. On February 12, 2026, Synthetix itself said sUSD was below $0.70 and had depegged for months. It raised the staking requirement to 50% of original jubilee debt, with planned ten-percentage-point fortnightly increases until 100% or a price above $0.98. Its forecast of a mid-2026 sustained peg was a forecast; the later retirement superseded that plan. [Synthetix: Rebuilding](https://blog.synthetix.io/rebuilding-susd/).


**Fact [C032].** The outcome changed decisively in 2026. SIP-423, created June 12 and now marked Implemented, describes retiring legacy sUSD on Ethereum and Optimism, exchanging each dollar-valued sUSD claim for four SNX under a one-year lock and one-year vest. It also explicitly defers some implementation phases. The official retirement dashboard dates the snapshot/freeze June 26, 2026, lock expiry June 26, 2027 and vesting expiry June 26, 2028. A deferred native-token reimbursement is not continued dollar convertibility. [SIP-423: sUSD](https://sips.synthetix.io/sips/sip-423/), [Synthetix sUSD](https://susd.synthetix.io/).


**Uncertain [C033].** I could not reconstruct a block-pinned, netted percentage of all legacy sUSD backing that was SNX at the April 2025 trough, nor an executable SNX-to-external-dollar depth curve for that date. SNX quantities and market capitalization do not answer either question. Aave’s July 2025 SNX deprecation proposal, reducing its isolation debt ceiling from $4 million to zero, is evidence of another venue’s risk response, not a measurement of SNX liquidation capacity or a safe ceiling for sUSD. [Aave long-tail](https://governance.aave.com/t/direct-to-aip-deprecation-of-long-tail-assets/22592), [Chaos Labs:](https://governance.aave.com/t/arfc-further-deprecate-susd-on-aave-v3-optimism/21770).


**Inference [C034].** The sequence supports a demand-and-exit failure before any demonstrated exhaustion of marked collateral: change debt ownership and repayment incentives, remove routine borrower buybacks, create selling pressure, subsidize a lockup, then ultimately retire the liability. Cutting a collateral ratio increases potential issuance and reduces loss tolerance, but the evidence does not identify that numerical cut alone as the depeg’s cause. A design can remain overcollateralized on an oracle screen while its holders lack an attractive route to realize a dollar. [SIP-420: Protocol](https://sips.synthetix.io/sips/sip-420/), [SIP-421: sUSD](https://sips.synthetix.io/sips/sip-421/), [SIP-423: sUSD](https://sips.synthetix.io/sips/sip-423/).


## 3.2 Djed and Indigo: ADA backing with different contraction constraints


**Fact [C035].** For Cardano’s Djed deployment, the backing asset is ADA: effectively 100% network-token reserve composition by design, not 100% issuer-governance-token backing. DJED holders have a senior claim; SHEN holders finance and absorb volatility in the surplus ADA reserve. The operator specifies a 400–800% target reserve range. Below 400%, new DJED minting and SHEN redemption stop, while DJED redemption into ADA remains available. Below peg this preserves a contraction route; above peg a mint stop can prevent supply arbitrage from eliminating a premium. [COTI: clarification](https://medium.com/cotinetwork/answers-and-clarification-to-recent-djed-community-questions-95176450db09), [Djed Alliance:](https://docs.stability.nexus/stablecoins/djed-overview/how-the-protocol-works).


**Fact [C036].** COTI’s June 14, 2023 account confirms that falling ADA pushed the reserve ratio below 400% and triggered those restrictions. It reports 255,000 DJED burned during the discussed volatility period against 172,000 minted, and roughly 54,000 ADA of transaction fees. Those are operator-reported flows over an incompletely specified window, not a continuous peg measurement. The generic Djed description also qualifies the redemption promise: once reserve value falls below total stablecoin liabilities, payment becomes pro rata rather than a guaranteed full dollar. [COTI: Djed](https://medium.com/cotinetwork/djed-shen-weathering-the-storms-of-market-volatility-while-creating-new-opportunities-a5b320241e3b), [Djed Alliance:](https://docs.stability.nexus/stablecoins/djed-overview/how-the-protocol-works).


**Uncertain [C037].** Djed’s secondary-market peg should not be inferred from the operator saying the contract worked. A later incident catalogue links historical observations of a 346% reserve ratio and approximately $1.078 DEX DJED in early July 2023 while minting was restricted. I did not independently replay those Cardano trades or obtain synchronized ADA bid depth. The defensible survival finding is that the documented contraction mechanism remained available through the June ratio breach; an uninterrupted one-dollar price across venues through October 2026 is not established here. [Djed reserve-ratio](https://github.com/ptrdsh/cardano-incident-catalogue/blob/main/CID-202302/README.md), [COTI: Djed](https://medium.com/cotinetwork/djed-shen-weathering-the-storms-of-market-volatility-while-creating-new-opportunities-a5b320241e3b).


**Fact [C038].** Indigo’s 2024 iUSD provides another ADA-backed CDP comparison: the v2 design returns ADA collateral on redemption rather than promising liquidation into the INDY governance token. The June 21 governance discussion says an approximately 21% interest rate had not sufficiently improved the peg and proposed 32%, plus a higher redemption-margin ratio. Its July 2 edit reduced the proposed margin from 300% to 250%. This was a proposal to change which debt was redeemable; it was not a change in a dollar reserve balance. [Indigo v2](https://indigoprotocol1.medium.com/indigo-protocol-v2-dual-peg-mechanisms-7965ff8a38e2), [Indigo: iUSD](https://forum.indigoprotocol.io/t/proposal-for-iusd-parameters-change/14081).


**Fact [C039].** That discussion estimated about 180,000 iUSD of net buying would balance the three named DEX markets, and listed 2,343,710 iUSD redeemable at a 300% margin using June 21 positions. These are scenario estimates for stablecoin demand and redemption eligibility, not measured ADA liquidation depth. Other passages use different position snapshots, so they should not be combined into a precise coverage ratio. The useful evidence is that admitting more collateral to redemption was considered separately from changing borrowing cost. [Indigo: iUSD](https://forum.indigoprotocol.io/t/proposal-for-iusd-parameters-change/14081).


**Fact [C040].** By the October 2026 read date, Indigo documentation describes v3 collateral beyond ADA, a stable-asset PSM such as iUSD/USDCx, and a configurable loan-redemption signer controlled by the Foundation at launch. Consequently, an ADA-only historical design must not be relabeled the current backing mix. I did not establish the current endogenous share or a clean post-change peg outcome. Survival of the protocol alongside diversification and a PSM is not evidence that the original ADA-only configuration held a tight peg unaided. [Indigo current](https://docs.indigoprotocol.io/readme/about-indigo), [Indigo current](https://docs.indigoprotocol.io/readme/redemptions).


## 3.3 USDD: current native-token exposure, with an external reserve alongside it


**Fact [C041].** USDD’s legacy documentation distinguishes whitelisted TRX-based issuance from stablecoin PSM issuance. Contemporary August 2024 reporting records removal of 12,000 BTC from disclosed reserves, leaving mainly TRX and USDT backing. That historical report is secondary, and I did not reconstruct the exact post-removal native-token share from archived balances. The legacy system and the later vault-based USDD version should not be merged into one invariant mechanism or one continuous collateral series. [USDD legacy](https://legacy-docs.usdd.io/usdd-issuance/usdd-issuance), [USDD removal](https://cointelegraph.com/news/justin-sun-says-removing-12k-bitcoin-usdd-defi-101).


**Fact [C042].** Direct official API reads on October 29, 2026 returned $1,164,130,388.78 of TRX/sTRX vault collateral on TRON and $2,211,827,191.06 of global collateral. Their ratio is 52.63%. TRON-only collateral was $1,872,526,785.01, making its native-token fraction 62.17%; counting only TRON loan vaults gives 99.94%. The denominators include different things. These are calculations from issuer-valued API observations, not audited asset values or executable proceeds. The requests are not pinned to a common block; the raw responses are retained in the dataset. [USDD TRON](https://openapi.usdd.io/api/v1/data-platform/latest-collateral?chain=tron), [USDD global](https://openapi.usdd.io/api/v1/market-site/overview), [USDD Ethereum](https://openapi.usdd.io/api/v1/data-platform/latest-collateral?chain=eth), [USDD BNB](https://openapi.usdd.io/api/v1/data-platform/latest-collateral?chain=bsc), [USDD public](https://docs.usdd.io/developers/usdd-public-api).


**Fact [C043].** At that read, TRX-A/B/C API minimum collateral ratios were 120%, 117% and 130%, with reported debt ceilings of 400 million, 300 million and 400 million USDD. These are policy ceilings, not demonstrated safe borrowing sizes. USDD’s PSM documentation explicitly limits one-for-one stablecoin conversion by available inventory. Its Smart Allocator page says earlier yield subsidies came from TRON DAO and describes replacing them with investment returns on reserve capital. This case includes both native collateral and custodial stablecoin dependencies. [USDD TRON](https://openapi.usdd.io/api/v1/data-platform/latest-collateral?chain=tron), [USDD PSM:](https://docs.usdd.io/user-guide/psm-peg-stability-module), [USDD Smart](https://docs.usdd.io/system-architecture/smart-allocator).


**Uncertain [C044].** USDD’s live balances show outstanding liabilities and majority network-token collateral in this snapshot. They do not establish its maximum historical depeg, resilience under complete TRX liquidation, or a native-only peg defence. Across SNX, ADA and TRX, I found no comparable historical executable-depth series with venue access, slippage, timestamp and collateral available for sale all aligned. No debt-to-token-market-cap ratio in these examples can substitute for that missing liquidation evidence. [Chaos Labs:](https://governance.aave.com/t/arfc-further-deprecate-susd-on-aave-v3-optimism/21770), [Djed reserve-ratio](https://github.com/ptrdsh/cardano-incident-catalogue/blob/main/CID-202302/README.md), [USDD TRON](https://openapi.usdd.io/api/v1/data-platform/latest-collateral?chain=tron), [USDD global](https://openapi.usdd.io/api/v1/market-site/overview).


## 4. Synthetic and yield-dollar collapses and near-misses, 2024–2026


## 4.1 November 2025: Stream, Elixir and the separate USDX evidence


**Fact [C045].** On November 4, 2025, contemporary reporting of Stream’s announcement says an external manager had disclosed approximately $93 million of losses the preceding day. Deposits and withdrawals were suspended and counsel engaged. Users’ backing therefore included claims on managed strategies, rather than only assets atomically available to redeem their token. The original announcement is linked but returned an access error; this claim relies on the contemporary report, not a direct reading of the inaccessible post. Neither it nor this research establishes an adjudicated explanation for the manager’s losses. [Stream discloses](https://www.coindesk.com/markets/2025/11/04/stream-finance-faces-usd93-million-loss-launches-legal-investigation), [Stream official](https://x.com/StreamDefi/status/1985556360507822093).


**Fact [C046].** xUSD was an accumulating yield claim: a contemporary price account puts its preceding-month quote near $1.26, then approximately $0.30 on November 4. Comparing its price only with $1 understates the discount to the prior share value. Nexus Mutual’s year-end account independently describes the rapid collapse, frozen withdrawals, recursive leverage and fixed-rate oracle exposure, and reports roughly $100,000 of paid claims across affected lending/vault products. That payment is an insurer’s own loss experience, not a measurement of total ecosystem losses. [Stream halt](https://forklog.com/en/stream-finance-halts-operations-following-93-million-loss/), [Nexus Mutual](https://nexusmutual.io/reports/nexus_review_2025_q4.pdf).


**Inference [C047].** The observed order is a manager-level backing problem, loss of redemption access, then rapid repricing and downstream lending impairment. A loan market can remain fully utilized and continue accruing high interest while lenders cannot withdraw. That rate is compensation demanded for missing cash, not proof of productive yield or solvency. A universal claim that the November Balancer exploit caused Stream’s loss is not supported by the retrieved primary material; overlapping dates do not establish that causal chain. [Stream discloses](https://www.coindesk.com/markets/2025/11/04/stream-finance-faces-usd93-million-loss-launches-legal-investigation), [Nexus Mutual](https://nexusmutual.io/reports/nexus_review_2025_q4.pdf), [Summer DAO:](https://forum.summer.fi/t/rfc-arbitrum-usdc-vault-next-steps-dealing-with-usdx-bad-debt/458).


**Fact [C080].** Before the Stream exposure dominated remaining reserves, deUSD had a different backing mix. Gauntlet’s April 29, 2025 Euler assessment describes a wstETH/short-ETH basis strategy, sUSDS, USDT and tokenized real-world assets; its then-current marks were about $10 million USDT, $150 million sUSDS, $5 million ETH and $15 million RWAs. Staked sdeUSD had a seven-day redemption cooldown. These are dated Ethereum backing observations, not the November mix, and show why the initial synthetic-dollar description cannot substitute for monitoring subsequent reserve deployment. [Euler: deUSD](https://forum.euler.finance/t/integrate-elixir-staked-deusd-sdeusd-on-euler-yield-market/1377).


**Fact [C081].** Inverse’s November 13 retrospective records warnings before collapse: it paused new FiRM borrows in late May and voted to sunset DOLA-deUSD collateral in August. It describes the visible execution-buffer depletion in March, rising fixed-feed lending exposure and proposals to lend up to $120 million of reserve sUSDS through external curators. Those proposals are not proof that both allocations executed. The account shows a counterparty responding to changes in reserve accessibility months before the November loss announcement, rather than relying solely on the eventual market depeg. [Inverse Finance:](https://forum.inverse.finance/t/behind-the-scenes-risk-curation-at-inverse-finance/624).


**Fact [C083].** The same Euler discussion supplies a concrete depth warning: on May 2, Objective Labs reported that swapping $2.1 million of sdeUSD to USDC incurred 12% slippage, against roughly $65 million posted in a Morpho market. This measures the stablecoin wrapper’s exit depth, not ETH or SNX market depth. It illustrates why a large collateral balance and concentrated lending TVL can coexist with a much smaller executable exit. [Euler: deUSD](https://forum.euler.finance/t/integrate-elixir-staked-deusd-sdeusd-on-euler-yield-market/1377).


**Fact [C048].** Elixir’s November 3 statement, accessed through a mirror, asserted special one-for-one redemption rights on its Stream lending position and said it was unwinding. Its November 6 mirrored statement instead described mint/redeem shutdown and a holder snapshot. It said roughly 90% of the remaining deUSD supply, around $75 million, was held by Stream and a similar fraction of remaining backing was a Morpho loan to Stream. Those are post-run remaining-balance fractions, not the pre-crisis reserve allocation or a proof of repayment capacity. [Elixir statement](https://teapaw.org/elixir/status/1985371198210064474), [Elixir wind-down](https://teapaw.org/elixir/status/1986443495351927257).


**Fact [C049].** The November 6 contemporary report identifies a $68 million Elixir loan to Stream. Elixir’s issuer documentation now explicitly marks deUSD deprecated, states that the asset has no value, and describes one-for-one USDC claims for eligible earlier holders, including lending, AMM and Pendle positions. Thus a surviving compensation process is distinct from a surviving token peg. The November 6 statement’s report of redemptions for 80% of non-Stream holders is a holder-based statement; it must not be recast as independently verified redemption of 80% of all token units. [Elixir sunsets](https://www.theblock.co/news/markets/2025-11-06-elixir-sunsets-deusd-synthetic-stablecoin-following-stream-finance-unwinding-aims-full-redemptions-377961), [Elixir deUSD](https://docs.elixir.xyz/deusd), [Elixir wind-down](https://teapaw.org/elixir/status/1986443495351927257).


**Inference [C050].** Elixir failed a backing-diversification test even if it had a contractual claim at par: a stablecoin owed by the same borrower to which reserve capital was lent creates correlated assets and liabilities. Claims of institutional collateral or nominal redemption rights do not resolve that concentration. The transition from a reassuring November 3 message to shutdown on November 6 also shows how quickly usable redemption can disappear. Exact aggregate deUSD low prices are omitted because the retrieved records do not reconcile post-snapshot token trades with separately compensated ownership. [Elixir statement](https://teapaw.org/elixir/status/1985371198210064474), [Elixir wind-down](https://teapaw.org/elixir/status/1986443495351927257), [Elixir deUSD](https://docs.elixir.xyz/deusd).


**Fact [C051].** Stables Labs’ USDX, a separate synthetic dollar, traded below $0.60 by November 6 according to contemporary reporting. An affected Summer DAO account records a November 3 Balancer pool exposure of about $1 million, then zeroed risk caps, missing liquidity and a Silo market remaining at 100% utilization. By November 10 its Arbitrum vault had approximately $1.49 million of exposure considered unrecoverable. These are dated affected-party statements, not a complete audit of USDX’s hedge book or proof that its full failure was caused by Stream. [USDX breaks](https://www.theblock.co/news/markets/2025-11-06-synthetic-stablecoin-usdx-depegs-pancakeswap-lista-377917), [Summer DAO:](https://forum.summer.fi/t/rfc-arbitrum-usdc-vault-next-steps-dealing-with-usdx-bad-debt/458).


**Fact [C052].** Lista’s own November 21 account describes an emergency vote and forced liquidation on November 6 of its Re7-curated USDX/USD1 vault. It says speed mattered because PancakeSwap liquidity was disappearing and reports 481,801.15 USD1 of liquidation profit for distribution. This is an example of preauthorized intervention recovering funds while an exit market still existed. It does not establish that the underlying USDX remained solvent, nor identify a universally safe delay for governance to act. [Lista DAO:](https://blog.lista.org/the-re7-vault-forced-liquidation-a-recap-and-moving-forward).


## 4.2 Resolv: the reserve survived while liabilities were corrupted


**Fact [C053].** Resolv’s April post-mortem describes crypto collateral including ETH, BTC, selected altcoins and tokenized real-world assets, combined with delta-neutral hedges. USR was the senior dollar claim; RLP was the junior risk-absorption layer. On March 22, 2026, unauthorized minting created 80 million USR: 50 million at 02:21:35 UTC and another 30 million at 03:41. The report puts extraction near $25 million in ETH. Relevant contracts were paused at 05:16, almost three hours after the first illicit mint. [Resolv official](https://resolv.xyz/blog/resolv-postmortem-march-22-2026-incident).


**Fact [C054].** The breached service could finalize minting without an independent on-chain maximum-output or collateral-ratio check. CertiK traces the privileged completion path; Resolv acknowledges that signing authority translated directly into unconstrained minting. Nexus Mutual reports Curve USR moving from $1 to $0.025 within 17 minutes. That is a reported pool-specific low, not a promise that every holder could trade at that price. The amount illicitly issued and the attacker’s extracted proceeds are different loss measures. [CertiK analysis](https://www.certik.com/blog/resolv-protocol-incident-analysis), [Resolv official](https://resolv.xyz/blog/resolv-postmortem-march-22-2026-incident), [Nexus Mutual](https://nexusmutual.io/blog/resolv-protocol-incident-report-by-nexus-mutual).


**Fact [C055].** The April 4 post-mortem says approximately 46 million illicit USR were neutralized by burns or blacklisting and that most pre-incident-holder redemptions had been processed. Later official recovery notices distinguish post-incident holders: their Stage 2 cash rate was 0.5 USDC per USR; some categories also received RESOLV allocations. The notice references a June 3 snapshot and June 27 vesting start. This was differentiated recovery, not restoration of universal one-for-one redemption. Completion totals and realized value of vested token allocations were not independently reconciled here. [Resolv official](https://resolv.xyz/blog/resolv-postmortem-march-22-2026-incident), [Resolv Stage](https://t.me/resolvlabs/275), [Resolv issuer](https://t.me/s/resolvlabs).


**Inference [C056].** Resolv’s junior tranche could absorb trading losses only within a system whose issuance remained honest. A privileged function creating claims without collateral bypassed that protection. A fee on market sales would have taxed only the trades passing through that pool; it would not have prevented the invalid liabilities or protected lenders accepting stale par values. Distinguish reserve custody controls, supply invariants and downstream collateral pricing: proving the first did not prove the other two. [CertiK analysis](https://www.certik.com/blog/resolv-protocol-incident-analysis), [Nexus Mutual](https://nexusmutual.io/blog/resolv-protocol-incident-report-by-nexus-mutual).


## 4.3 Ethena: survived stress, with significant reserve and access machinery


**Fact [C057].** Ethena’s February 2025 Bybit incident review describes spot collateral plus offsetting short derivatives, off-exchange custodians and whitelisted primary redemptions. Approximately 20% of backing was allocated to Bybit hedges; roughly $30 million of unsettled profit exposure was reduced to zero. The review reports over $120 million of redemptions serviced, a $1.9 billion stablecoin buffer and a $60 million reserve fund. Those resources and controlled settlement access helped distinguish this near-miss from a claim backed only by a thin native token. [Ethena risk](https://gov.ethenafoundation.com/t/bybit-incident-post-mortem-ethenas-risk-mitigation-and-resilience/518).


**Fact [C058].** Ethena’s August 2025 governance update puts backing at 54% stable assets and 46% delta-neutral positions, including BTC, ETH and ETH LSTs. It reports $2.9 billion of stable assets supplied to Aave. Thus even before the October stress, USDe was not solely an unreserved ETH basis trade. Yield-bearing stablecoin allocations added lending-market withdrawal constraints to hedge-settlement risk. Current October 2026 documentation describes an even broader strategy set, including lending and real-world assets; that description must not be projected backward onto the initial 2024 product. [Ethena August](https://gov.ethenafoundation.com/t/ethenas-august-2025-governance-update/689), [Ethena current](https://docs.ethena.fi/).


**Fact [C059].** For October 10–11, 2025, the committee’s analysis reports USDe at 0.65 USDT on Binance and 0.92 on Bybit, while major Ethereum and Plasma DEX pools stayed around $0.99. On Binance it spent 23 minutes below 0.90, 53 below 0.95 and 86 below 0.99. Binance separately acknowledged forced liquidations from the dislocations. An exchange mark-price feedback loop was therefore a severe user-loss event even though the issuer’s on-chain collateral and primary redemption machinery did not suffer the same collapse. [Ethena risk](https://gov.ethenafoundation.com/t/usde-analysis-during-the-oct-10th-11th-volatility-event/706), [Binance confirms](https://www.binance.com/en/support/announcement/detail/f3d5d97ed12c4f639019419bd891705e).


**Uncertain [C060].** The October committee article contains inconsistent redemption totals: its summary says approximately $1.2 billion against $8.6 billion supply, while its dedicated redemption section says $1.9 billion on October 10–11 against $14.6 billion, with no backlog and no reserve-fund use. I retain the latter as that section’s reported figure, not an independently reconciled total. The venue-price distinction is supported independently by Binance’s acknowledgement; exact primary flow totals need a transaction-level definition of the window and gross versus net burns. [Ethena risk](https://gov.ethenafoundation.com/t/usde-analysis-during-the-oct-10th-11th-volatility-event/706), [Binance confirms](https://www.binance.com/en/support/announcement/detail/f3d5d97ed12c4f639019419bd891705e).


**Fact [C061].** The later risk-committee discussion of Aave backing includes historical stress flows: approximately $95.8 million redeemed on August 5, 2024, and $267.9 million on March 1, 2025. The latter coincided with both the post-Bybit period and a Pendle maturity. Those are reported redemption events, not evidence that funding losses or an exploit caused either amount. They show that redemption demand can cluster around leverage unwinds and scheduled maturities as well as market shocks. [Ethena backing](https://gov.ethenafoundation.com/t/proposal-onboarding-aaves-ausdc-and-ausdt-as-usde-backing-assets/532).


## 4.4 Two further failure modes: changing exits and corrupt share arithmetic


**Fact [C062].** Usual’s January 2025 episode concerned USD0++, a locked yield instrument, rather than proof that its underlying USD0 reserves disappeared. Contemporary reporting describes a $0.87 early-exit floor and a market quote below $0.92 on January 10. The issuer’s January 14 update offered one-for-one early unstaking only with a variable USUAL contribution capped at six months of rewards. On January 23 it reported $250 million exiting through that mechanism and proposed improving liquidity and the floor. A conditional, fee-bearing bond exit differs from cash redemption. [USD0++ repricing](https://www.theblock.co/news/defi/2025-01-10-usual-money-protocol-update-333995), [Usual: early](https://usual.money/blog/usd0-upgrade-early-unstaking-vaults-liquidity), [Usual: Road](https://usual.money/blog/a-new-dawn-for-usd0-road-to-parity).


**Fact [C063].** Resupply’s June 28, 2025 post-mortem records a new Ethereum crvUSD-wstUSR market deployed at 00:18:47 on June 26 with a ten-million-reUSD debt limit and an empty underlying vault. At 01:53:59 an attacker manipulated share value so reciprocal exchange-rate arithmetic rounded to zero, bypassing the solvency check and borrowing the full ceiling. The pair was paused at 02:53:23. The oracle tracked the inflated share value correctly; the fatal step was downstream arithmetic. The report records ten million reUSD of bad debt, distinct from net attacker proceeds. [Resupply official](https://paragraph.com/@resupply/xXNoMg4iByksss66igix).


## 5. Fee-based defences: what is implemented and what is measured


**Fact [C064].** Curve StableSwap-NG increases swap fees as normalized balances become more unequal. The off-peg multiplier bounds the amplification of the base fee; it is a pool-specific setting. The calculation uses balance states along the trade, so a finite trade restoring balance can pay less than one worsening it. This is liquidity-provider compensation and execution shaping inside a funded AMM. Neither the formula nor the invariant guarantees the dollar value of a constituent token when its backing fails. [Curve StableSwap-NG](https://docs.curve.finance/developer/amm/stableswap-ng/overview/), [Curve explanation](https://knowledge.curve.finance/protocol/pool/understanding-stableswap).


**Fact [C065].** DeFiChain provides an unusually explicit surcharge failure record within the period. The February 2024 governance design added a discount fee starting when DUSD reached $0.95 or lower, rising to 50% below $0.50, on top of the existing stabilization fee. The May official governance roundup discussed removing a combined 80% fee. These were user-funded burns and trading restrictions, not an independently capitalized reserve. DUSD was semi-algorithmic, so it is a fee-design comparator rather than a clean example of a fully overcollateralized own-token stablecoin. [DUSD dynamic](https://www.reddit.com/r/defiblockchain/comments/1auj0ne/dynamical_discount_dex_stabilization_fee/), [DeFiChain official](https://blog.defichain.com/2024/05/a-closer-look-at-july-dfip-cfp-voting.html), [DeFiChain official](https://blog.defichain.com/2024/11/a-successful-restart-for-defichains.html).


**Fact [C066].** DeFiChain’s November 20, 2024 account expressly says dynamic fees and interest had been ineffective against the algorithmic DUSD overhang and inadequate collateralization. Its replacement involved a snapshot and controlled release of liquidity in 100 tranches. Even the optimistic restart account put DUSD near $0.30. It does not isolate the surcharge’s causal effect from the pre-existing balance-sheet problem. [DeFiChain official](https://blog.defichain.com/2024/11/a-successful-restart-for-defichains.html).


**Fact [C067].** There is measured evidence that fees affect venue activity, but a weaker claim is warranted than “the peg was defended.” Curve’s June 2025 recap reports reducing stETH-ng’s fee from 0.04% to 0.008%, followed by increased volume and approximately $500 daily DAO revenue. That is a funded liquid-staking pair, not a dollar-stablecoin surcharge experiment, and the recap supplies no cross-venue counterfactual. StablePair’s launch design likewise explicitly prices flows to retain LP value, which differs from establishing a stablecoin’s solvency floor. [Curve June](https://news.curve.finance/curve-monthly-recap-june-2025/), [Uniswap Labs:](https://blog.uniswap.org/keeping-value-in-the-pool-how-stablepair-hook-works).


**Inference [C068].** A directional surcharge creates a wedge between a pool’s displayed price and the seller’s actual proceeds. A below-peg sale with an extra fee realizes less than the pre-fee quote; an unchanged displayed reference price can therefore coexist with a worse exit. With permissionless competing pools or bilateral trades, some users may avoid that wedge. Diversion is an economic prediction here, not a quantified finding: the retrieved evidence does not identify how much depeg-selling volume migrated because of any particular toll. DeFiChain’s reported failure also contradicts a general assertion that a sufficiently high sale toll necessarily restores par. [StablePair dynamic](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/concepts/dynamic-fees), [DeFiChain official](https://blog.defichain.com/2024/05/a-closer-look-at-july-dfip-cfp-voting.html), [Uniswap v4](https://developers.uniswap.org/docs/protocols/v4/concepts/dynamic-fees).


**Uncertain [C069].** I found no clean study in the reviewed primary evidence comparing the same stablecoin with and without a directional depeg surcharge while accounting for depth, subsidies, redemption access and cross-venue routing. Successful swaps, fee revenue, TVL and a price inside a configured pool band do not establish that effect. A useful evaluation would report fees-inclusive executable prices, time outside stated bands, net LP losses, redemption throughput and the same token’s volume on untaxed venues before and after activation. [StablePair audit](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/security), [Curve StableSwap-NG](https://docs.curve.finance/developer/amm/stableswap-ng/overview/), [Curve June](https://news.curve.finance/curve-monthly-recap-june-2025/), [DeFiChain official](https://blog.defichain.com/2024/11/a-successful-restart-for-defichains.html).


## 6. Design implications for a small, concentrated-collateral stablecoin


**Inference [C070].** Make contraction an enforceable economic operation. Allow debt cancellation through redemptions into collateral that buyers can actually sell, preserve repayment access during stress, and document shutdown behavior. Borrower-set rates can make low-rate debt the first redeemed and encourage rates to adjust when demand weakens. They do not create bids for the collateral. A staked-token redemption must be evaluated using the liquid staking share’s stressed sale value or the real unbonding delay, rather than assuming immediate access to its underlying token. [Liquity v2](https://docs.liquity.org/v2-faq/redemptions-and-delegation), [SIP-421: sUSD](https://sips.synthetix.io/sips/sip-421/), [Djed Alliance:](https://docs.stability.nexus/stablecoins/djed-overview/how-the-protocol-works).


**Fact [C071].** Liquity v2’s documented reference design routes redemptions across collateral branches according to outside debt, then starts with the lowest-rate borrowers within each branch. Redeemers receive oracle-valued collateral less a fee. Its published fee is a 0.5% floor plus a redemption-sensitive base rate, capped at 100%, with a six-hour decay half-life. This is a dated reference mechanism for the proposed direction of travel, not a claim about imdUSD’s contracts or parameters. The repository explicitly warns that a larger redemption fee permits a looser peg. [Liquity v2](https://docs.liquity.org/v2-faq/redemptions-and-delegation), [Liquity BOLD](https://github.com/liquity/bold).


**Inference [C072].** Treat collateral diversification as diversification of exit risks. Different wrappers on one token, claims on one borrower, and several stablecoins financed by the same leveraged strategy do not provide independent liquidity. Add collateral with separately funded buyers, usable price feeds and liquidation capacity, and cap each branch before allowing growth to consume that capacity. Branch isolation limits some liquidation transfers; a shared stablecoin and redemption router still transmit demand and price shocks between branches. [Elixir wind-down](https://teapaw.org/elixir/status/1986443495351927257), [LlamaRisk: crvUSD](https://www.llamarisk.com/research/crvusd-incident-report-20240612), [Liquity BOLD](https://github.com/liquity/bold), [Summer DAO:](https://forum.summer.fi/t/rfc-arbitrum-usdc-vault-next-steps-dealing-with-usdx-bad-debt/458).


**Inference [C073].** Use a hook to improve trade economics around a solvent redemption system. Distinguish a trade’s direction from whether it is helpful at the current dollar price, include its final price impact, and preserve profitable corrective arbitrage after all fees. Cap any subsidy by already realized, externally valuable fee assets. Rebate promises denominated only in the same stressed token are weak buying power. Because arbitrary venues remain available, the hook should publish an economic objective for aggregate peg quality as well as local LP protection. [StablePair dynamic](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/concepts/dynamic-fees), [StablePair audit](https://developers.uniswap.org/docs/protocols/uniswap-labs-hooks/stable-pair/security), [DeFiChain official](https://blog.defichain.com/2024/11/a-successful-restart-for-defichains.html).


**Inference [C074].** Keep safety gates from disabling the only useful response. Curve’s observed 45-minute delay shows that oracle divergence can be evidence of genuine demand, not just manipulation. Use independent asset-quality checks, explicit action/velocity caps and direction-specific emergency rules; test both insolvency and scarce-stablecoin scenarios. A mint freeze may contain additional exposure while leaving repayment and risk-reducing withdrawals possible. A fixed par quote is unsuitable as the only downstream lending valuation when redemption is unavailable. [LlamaRisk: crvUSD](https://www.llamarisk.com/research/crvusd-incident-report-20240612), [Migrate PegKeepers](https://gov.curve.finance/t/migrate-pegkeepers-to-a-new-version-3/11166/1), [Nexus Mutual](https://nexusmutual.io/blog/resolv-protocol-incident-report-by-nexus-mutual).


**Inference [C075].** Separate durable liquidity from purchased demand. LP claims owned by the protocol provide a balance sheet and reduce dependence on another LP renewing an incentive, but their counterasset can be drained or impaired. Borrowed stablecoins and emissions-funded LP incentives rent capacity; debt forgiveness and lockups may postpone selling without resolving its source. Retained external trading revenue, conservative borrower interest and genuinely independent collateral add more durable resources, although each still needs an exit. Report these funding sources separately rather than adding their headline values together. [Frax Curve](https://docs.frax.finance/amo/curve), [FIP-415: Creating](https://gov.frax.finance/t/fip-415-creating-frax-borrow-amo/3513), [FIP-188: Increase](https://gov.frax.finance/t/fip-188-increase-cr-to-100/2147), [Synthetix: The](https://blog.synthetix.io/the-repeggening/), [Pangea: crvUSD](https://blog.pangea.foundation/does-yieldbasis-hurt-the-crvusd-peg/).


**Inference [C076].** Set ceilings using executable stressed depth, not token capitalization. Define L(s,T) as the collateral’s dollar notional that independent buyers can absorb within horizon T and maximum impact s after an initial stress shock. Exclude liquidity paid for with the stablecoin itself, inaccessible venues and repeated use of the same underlying pool. If fraction r of debt might need liquidation or redemption in T, collateral incentive b is needed, and only fraction eta of measured depth is dependable, a conservative operational constraint is (1+b) × r × D <= eta × L(s,T). This is a proposed risk model, not a calibrated empirical law. [LlamaRisk: crvUSD](https://www.llamarisk.com/research/crvusd-incident-report-20240612), [Chaos Labs:](https://governance.aave.com/t/gho-depeg-and-liquidity-analysis/15250), [Chaos Labs:](https://governance.aave.com/t/arfc-further-deprecate-susd-on-aave-v3-optimism/21770), [Resupply official](https://paragraph.com/@resupply/xXNoMg4iByksss66igix).


**Inference [C077].** Apply a separate solvency test: a marked collateral ratio R must satisfy R × (1−h) × (1−s) >= 1+c, where h is the initial token-price shock and c liquidation/funding cost per dollar of debt. Do not count the initial shock again inside execution impact s. These inequalities constrain different failures: sufficient assets at stressed prices and enough willing buyers in time. They require fresh quotes and stress assumptions. A higher initial ratio helps the first constraint but cannot alone make a large sale executable. [Djed Alliance:](https://docs.stability.nexus/stablecoins/djed-overview/how-the-protocol-works), [Chaos Labs:](https://governance.aave.com/t/arfc-further-deprecate-susd-on-aave-v3-optimism/21770), [Liquity BOLD](https://github.com/liquity/bold).


**Uncertain [C078].** The evidence does not support a transferable numerical ceiling such as a fixed percentage of market capitalization or a fixed multiple of daily volume. The cases lack synchronized historical depth and comparable run fractions; observed success also involved different reserves, subsidies and market structures. For a predominantly thin own-token system, the supported direction is a ceiling below conservatively discounted, stress-tested executable capacity for the chosen run horizon. Neither the discount nor the run fraction can honestly be calibrated from the assembled cases alone. [LlamaRisk and](https://llamarisk.com/research/pegkeeper-onboarding-review-aave-s-gho), [Chaos Labs:](https://governance.aave.com/t/arfc-further-deprecate-susd-on-aave-v3-optimism/21770), [Djed reserve-ratio](https://github.com/ptrdsh/cardano-incident-catalogue/blob/main/CID-202302/README.md), [USDD global](https://openapi.usdd.io/api/v1/market-site/overview).


**Inference [C079].** Grow only after demonstrating the complete path at meaningful size: purchase discounted dollars, redeem or liquidate, sell the actual collateral, and settle outside assets under stressed gas, stale-feed, LP-withdrawal and competing-venue conditions. Publish achieved proceeds, elapsed blocks, available capacity and concentration by common counterparty. Add per-operation and per-period issuance bounds even where a signer or keeper is trusted. Resolv’s unconstrained mint and Resupply’s immediately exhausted new-market limit show that both authority compromise and valid-looking arithmetic can consume capacity before human response. [CertiK analysis](https://www.certik.com/blog/resolv-protocol-incident-analysis), [Resupply official](https://paragraph.com/@resupply/xXNoMg4iByksss66igix), [Nexus Mutual](https://nexusmutual.io/blog/resolv-protocol-incident-report-by-nexus-mutual).


## 7. Unanswered questions and limits

Each item below is **Uncertain**. These are explicit research gaps, not assumed zeros or evidence of success. Their IDs and next steps are also in `data.json`.


- **G01 — Historical executable collateral depth:** Synchronized SNX, ADA and TRX executable bid curves, common stress horizons and independently funded buyer concentration during the cited incidents.


- **G02 — Net endogenous backing shares:** A reconciled system-wide sUSD asset/debt ledger at the 2025 low and historical USDD backing composition around August 2024.


- **G03 — Curve historical execution and live caps:** Independent replay of every cited keeper operation, October 2026 active ceilings and executed status of subsequent v3 proposals.


- **G04 — GHO causality and immediately usable reserves:** A controlled attribution of 2023–24 recovery to rates, stkGHO, liquidity incentives and GSM; fully independent September 2026 cross-chain reserve replay.


- **G05 — Frax peg performance and live limits:** Independent replay of the reported frxUSD price series, AMO-isolated performance analysis, and complete October 2026 active per-strategy/custodian limits.


- **G06 — Djed and Indigo historical/current state:** Independent replay of Djed July 2023 reserve and DEX observations; continuous price survival evidence; current Indigo branch weights and usable redemption throughput.


- **G07 — USDD accounting verification:** Block-pinned on-chain reconciliation and look-through composition of Smart Allocator positions.


- **G08 — Stream and Elixir primary access and recovery:** Directly readable original Stream loss post, an authenticated archive of Elixir’s November statements, complete manager-loss root cause, pre-run reserve allocation and final aggregate recoveries.


- **G09 — USDX underlying failure:** Complete verified backing and hedge book and a post-mortem allocating Balancer, Stream or other credit losses.


- **G10 — Resolv recovery completion:** Reconciled final recovery by holder class and realized value of vested RESOLV allocations.


- **G11 — Ethena October 2025 redemption totals:** Reconciliation of conflicting $1.2 billion/$8.6 billion and $1.9 billion/$14.6 billion statements within the committee article.


- **G12 — Fees and venue diversion:** Controlled evidence of aggregate peg improvement or measured migration caused by depeg surcharges; StablePair live numerical settings and a post-launch event study.


- **G13 — Completeness and deployment provenance:** Exhaustive incident coverage, archived historical versions for every mutable documentation page, and bytecode verification for all cited deployments.