Algorithmic Stablecoin Failures — Deep Catalogue
How Terra UST, Iron Finance, USDR, and various other algorithmic stablecoins failed — the specific mechanisms in each case, and why all unbacked algorithmic stablecoins face the same structural risk.
The basics
What Makes a Stablecoin Algorithmic
A stablecoin's design determines what backs its peg. Fiat-collateralised (USDT, USDC, USDS): backed 1:1 by dollars or equivalents held by an issuer. Crypto-collateralised (DAI/USDS via Maker): backed by over-collateralised crypto positions. Algorithmic: not backed by any real asset; the protocol uses incentives and mechanisms to maintain the peg through demand-side or supply-side adjustments. Algorithmic stablecoins are the most theoretically elegant — no backing required — and also the most prone to catastrophic failure when the mechanism breaks.
The Pattern of Failure
Most major algorithmic stablecoins have eventually failed in a depeg cascade. The pattern is consistent: peg holds during calm conditions because arbitrageurs find the mechanism profitable; under stress, arbitrage stops working; the peg breaks; the same mechanism that maintained the peg now amplifies the failure. Terra/Luna (May 2022, $40B+ destroyed in days) is the canonical case but is far from unique.
The Catalogue You Should Know
Major algorithmic stablecoin failures: Terra/UST (May 2022, $40B+ destroyed); Iron Finance (June 2021, an estimated $2B in losses); Empty Set Dollar; USDR (Real Estate USD); Beanstalk (multiple incidents); Fei Protocol (significant depeg). Each has its own specific mechanism story, but the broader pattern — algorithmic stability mechanisms that work in calm but fail in stress — is a recurring one. Basis failed differently: the algorithmic stablecoin project, which had raised $133 million from venture capital, shut down in December 2018 over concerns about US regulation, not in a depeg cascade.
- Algorithmic stablecoins are unbacked — relying on mechanism, not collateral
- Most major algorithmic stablecoins have eventually failed in a depeg cascade
- The mechanism that maintains the peg becomes the cascade amplifier under stress
- The failure history is consistent across diverse designs
Key Takeaways
- Algorithmic stablecoins use mechanism rather than backing to maintain peg
- Failure pattern is consistent: peg holds in calm, breaks under stress
- Major failures: Terra/UST ($40B), Iron Finance ($2B), USDR, ESD, others; Basis shut down in December 2018 over US regulatory concerns rather than a depeg
- Understanding why they failed is the load-bearing knowledge
Going deeper
Terra/UST — The Twin-Token Mint-Burn Model
UST was backed by Luna via a mint-burn mechanism: $1 of UST could always be exchanged for $1 of Luna at protocol rate. When UST traded above $1, arbitrageurs burned Luna to mint UST and sold it for profit. When UST traded below $1, arbitrageurs burned UST to mint Luna and sold Luna for profit. Anchor Protocol offered 19.5% yield on UST deposits, driving demand. The system worked beautifully — until May 7, 2022, when sustained selling pressure pushed UST below the peg, and the resulting Luna minting tanked Luna's price, which made arbitrage impossible, which let UST fall further, which crashed Luna further. Within about a week, Luna had fallen to virtually zero and UST to about $0.10.
Iron Finance — Fractional Algorithmic
Iron Finance's IRON stablecoin was 'fractionally algorithmic' — partially backed by USDC, partially by TITAN (the protocol's volatile token). Holders could redeem IRON for some USDC + some TITAN. When TITAN's price fell, the protocol minted more TITAN to maintain redemptions, tanking TITAN's price further. In June 2021, after large holders began selling TITAN, it went into a death spiral and fell to near zero, with an estimated $2 billion in losses. Iron Finance, whose early investors included Mark Cuban, denied the collapse was a rug pull and blamed a bank run triggered by the protocol's algorithmic design — the design itself was the failure mode.
USDR — Real Estate Backed (Or Was It)
USDR (Real USD) claimed backing by tokenised real estate plus DAI plus the protocol's TNGBL token. The 'algorithmic' part: when redemptions exceeded liquid DAI reserves, TNGBL was minted to backfill. When a wave of redemptions exhausted DAI, TNGBL minting collapsed TNGBL's price and USDR lost its peg. The real-estate 'backing' was illiquid and couldn't be sold fast enough to defend the peg. This is a useful case study because it shows that 'backed by real assets' doesn't help if the assets aren't liquid in stress conditions.
Anchor's Role in UST
Anchor Protocol paid 19.5% APY on UST. This wasn't sustainable — Anchor's actual yield generation was far below 19.5%, and the gap was subsidised by Terra's foundation reserves. The artificial yield drove UST demand beyond what the system could sustain in stress. When stress arrived, the yield couldn't compensate for the depeg risk. The lesson: any stablecoin paying yield substantially above sustainable rates is using subsidies that will eventually run out.
Key Takeaways
- UST: mint-burn arbitrage worked until selling pressure broke the arbitrage; Luna minted into hyperinflation
- IRON: fractional backing with volatile token; TITAN minting tanked TITAN, exhausted backing
- USDR: illiquid 'real asset' backing couldn't be sold fast enough during redemption stress
- Common pattern: mechanism amplifies stress rather than absorbing it
- Subsidised yield drives unsustainable demand that breaks when subsidy ends
Under the hood
Why The Pattern Repeats
Algorithmic stablecoins fail for structural reasons that any specific design replicates. Reflexivity: the mechanism's success depends on token X's price being stable; the mechanism's response to stress involves minting more X; minting more X under stress destabilises X's price further. Bank run dynamics: holders who fear depeg sell first; their selling causes depeg; selling accelerates. Subsidy dependence: artificially high yields drive adoption beyond organic demand; when yields normalise, demand collapses. These dynamics aren't bugs in specific implementations — they're features of unbacked stablecoin designs.
Why People Keep Trying
Despite the failure history, new algorithmic stablecoin designs keep appearing. Reasons: the theoretical elegance is real (a stablecoin without collateral is genuinely useful); the early adopter rewards are substantial; the failure cases are framed as 'their specific design was flawed, ours is different.' Designers underweight the recurring pattern. Investors underweight the high prior. Whether any algorithmic stablecoin design could fundamentally break the pattern is genuinely unsettled; the empirical track record is mostly failure.
What 'Delta Neutral' Stablecoins Do Differently
Ethena's sUSDe (covered in its own learn topic) uses a different approach: backed by short-perpetual positions hedging spot ETH/BTC exposure. Not algorithmic in the unbacked sense — there's real collateral, hedged via derivatives. But the hedging mechanism has its own structural risks (funding rate inversions, exchange counterparty risk, settlement risk during extreme volatility). Delta-neutral designs performed better than pure algorithmic ones through the October 2025 market turbulence but are not free of structural risk: on 10 October 2025 USDe briefly fell to $0.65 on Binance's spot market while trading close to $1 on-chain (21Shares), and its supply fell from about $14.7B on 9 October 2025 to about $6.3B two months later (ForkLog, December 2025).
Regulatory Treatment
Post-Terra, regulators have hardened against unbacked algorithmic stablecoins. The EU's MiCA does not ban algorithmic stablecoins by name: where a token falls within the definition of an asset-referenced token (ART) or e-money token (EMT), the ART or EMT rules apply however its stable value is maintained, and recital 41 of Regulation (EU) 2023/1114 says the same applies to so-called algorithmic 'stablecoins' 'that aim to maintain a stable value in relation to an official currency, or in relation to one or several assets'; EMTs must be issued at par and be redeemable at par (Art. 49). MiCA's ART and EMT rules have applied since 30 June 2024. The US GENIUS Act (Public Law 119-27, approved 18 July 2025) does not ban algorithmic stablecoins by name either: it makes it unlawful for anyone other than a permitted payment stablecoin issuer to issue a payment stablecoin in the United States (Sec. 3(a)), requires permitted issuers to 'maintain identifiable reserves backing the outstanding payment stablecoins ... on an at least 1 to 1 basis' (Sec. 4(a)(1)(A)) and orders a Treasury study of non-payment stablecoins, including 'endogenously collateralized' stablecoins (Sec. 14). The Act takes effect on the earlier of 18 months after enactment (18 January 2027) or 120 days after the federal regulators issue final implementing regulations (Sec. 20). Terraform Labs and Kwon agreed in June 2024 to pay $4.5 billion to settle the SEC's fraud case (SEC release of 13 June 2024, as cited by Wikipedia), and a bankruptcy judge approved Terraform's wind-down in September 2024 (Decrypt). Do Kwon was extradited to the US in December 2024 (Crowdfund Insider), pleaded guilty to fraud charges in August 2025 and was sentenced to 15 years in prison on 11 December 2025 (Decrypt). New algorithmic stablecoin designs face substantial regulatory friction in major jurisdictions. This doesn't prevent them from existing on-chain but does affect what user bases they can reach.
Algorithmic Stablecoins Carry High Failure Risk
An algorithmic stablecoin carries a high risk of failure regardless of how well-designed its mechanism appears. The history is mostly failure; the structural reasons for failure are general; new designs facing first-ever stress tests have no track record. This isn't a claim that algorithmic stablecoins are impossible; it's a claim that the prior on failure is high. Fiat-backed, crypto-collateralised and algorithmic stablecoins carry different structural risks.
Key Takeaways
- Failure pattern is structural: reflexivity, bank-run dynamics, subsidy dependence
- New algorithmic stablecoin designs keep appearing despite the failure history
- Delta-neutral designs (Ethena sUSDe) have different risk profiles but aren't free of structural risk
- Post-Terra regulation (MiCA's ART/EMT rules, which cover algorithmic stablecoins that aim to maintain a stable value in relation to an official currency or one or several assets; US prosecution of Do Kwon; the GENIUS Act's permitted-issuer rule, 1:1 reserve requirement and Treasury study) hardens against unbacked algorithmic stablecoins
- New algorithmic stablecoins carry a high failure risk until they build an empirical track record under stress
Related Content
Related Coins
Key Terms
Case Studies
More Topics
Browse all 189 topicsICP Canister Model
How Internet Computer (ICP)'s canisters fundamentally differ from Ethereum-style smart contracts — combining compute, storage, and HTTP-server-like functionality in a single on-chain primitive.
ERC-721A and Compressed NFTs
How gas-optimised NFT standards like ERC-721A made batch mints affordable, and how Solana state compression takes the same idea further by committing compressed NFTs to a Merkle tree instead of storing each one in its own account.
ERC-6551 Token-Bound Accounts
How ERC-6551 lets NFTs own smart contract wallets that can in turn own other assets — enabling 'NFTs that own NFTs,' character-bound inventories, and new composability patterns.
Ordinals, Inscriptions, and BRC-20
How Casey Rodarmor's 2023 Ordinals protocol brought NFTs and tokens to Bitcoin via inscriptions, what BRC-20 means in practice, and how Runes evolved the model.
Delta-Neutral Stablecoins (Ethena sUSDe)
How Ethena's sUSDe captures perpetual funding rates by hedging spot crypto with short perpetuals, what makes it different from algorithmic stablecoins, and the structural risks beneath the surface.
Liquid Staking Tokens and Restaking
How stETH/rETH liquid staking tokens work, why EigenLayer's restaking model created a new economic primitive, and what AVS economics mean for the broader Ethereum ecosystem.
Options Greeks On-Chain
How delta, gamma, vega, and theta translate from traditional options markets to on-chain protocols like Derive (formerly Lyra), Aevo, and Premia — and what's structurally different about decentralised options markets.
TWAP and VWAP Execution Algorithms
How time-weighted and volume-weighted average price execution algorithms split large orders to minimise market impact, and what their crypto-native equivalents look like on DEXs and CEXs.
References & further reading
- secondary
- secondary
- secondaryWikipedia — Iron Finance
The June 2021 IRON/TITAN collapse.
- secondaryWikipedia — Stablecoin
Basis, an algorithmic stablecoin that had raised $133 million, shut down in December 2018 over US regulatory concerns.
archived copy ↗ - primaryRegulation (EU) 2023/1114 (MiCA)
Recital 41 (algorithmic stablecoins) and Art. 49 (EMT issuance and redemption at par).
- secondaryDechert — MiCA phase one: issuers of ARTs and EMTs (June 2024)
MiCA's ART and EMT rules apply from 30 June 2024.
- primaryGENIUS Act, Public Law 119-27 (govinfo)
Sec. 3(a) permitted issuers; Sec. 4(a)(1)(A) 1:1 reserves; Sec. 14 Treasury study of endogenously collateralised stablecoins; Sec. 20 effective date.
- secondaryWikipedia — Terra (blockchain)
Cites the SEC's 13 June 2024 release on the $4.5 billion settlement.
- secondaryDecrypt — Terraform Labs bankruptcy wind-down approved (September 2024)
Bankruptcy judge approves Terraform's wind-down.
- secondaryDecrypt — Do Kwon sentenced to 15 years (11 December 2025)
Sentencing, August 2025 guilty plea and the SEC settlement.
- secondaryCrowdfund Insider — Do Kwon receives 15-year sentence (December 2025)
Kwon was extradited to the US in December 2024 ('December of last year' in this December 2025 report).
- secondary21Shares — Why did Ethena's stablecoin depeg on Binance? (October 2025)
USDe's 10 October 2025 dislocation to $0.65 on Binance.
- secondaryForkLog — USDe capitalisation down $8.3 billion since October (December 2025)
USDe supply about $14.7B on 9 October 2025 and $6.3B two months later.