A depeg headline rarely gives the reader enough to decide whether a stablecoin is going to zero or trading at a discount for the weekend. The word "stablecoin" covers mechanically different designs that fail in different ways, and the difference between a 72-hour access problem and a death spiral is not a matter of sentiment. It is a matter of which collateral stack is under stress and why.

This article is Part 5 of "Reading Crypto News." It gives a framework for the next time a stablecoin wobbles: four peg-event patterns, three case studies drawn from the post-2022 record, a summary of the regulatory regime that is now converging on both sides of the Atlantic, and a checklist for the next headline.

Why "Stablecoin" Is a Category, Not a Guarantee

The label covers at least four mechanically distinct designs:

Fiat-backed: USDC, USDT, PYUSD. A claim on commercial-bank deposits and short-dated Treasuries held by an issuer.
Overcollateralized crypto-backed: DAI, sDAI. A claim on a smart contract holding volatile collateral that gets liquidated if it falls below a threshold.
Algorithmic / seigniorage: the extinct UST. A claim on the market's continued willingness to buy a paired governance token.
Delta-neutral synthetic: Ethena's USDe. A claim on a cash-and-carry trade pairing long spot with short perpetual futures on centralized exchanges.

Readers who want the first-principles mechanics should start with the primer on how stablecoins actually work. The point of the taxonomy here is narrower. When a headline says "stablecoin X depegged to $0.92," the useful question is not "is this Terra 2.0?" but "which of those four stacks is stressed, and how?"

The Four Peg-Event Patterns

Pattern 1 — Reserve doubt. The market loses confidence that the issuer's off-chain assets are real, liquid, or recoverable. USDC in March 2023 is the canonical example: reserves existed, but a meaningful slice was frozen inside a failing bank. Tether has faced recurring variants of this pattern around attestation quality.

Pattern 2 — Liquidity vacuum. Reserves are fine, but the on-chain or exchange venues where the coin trades run out of bid-side depth during a stress event. The secondary-market price detaches from net asset value for minutes or hours, and most small, short-lived depegs belong here. They look scarier on a candle chart than they are in substance.

Pattern 3 — Protocol or mechanism failure. The stabilization mechanism itself breaks. For algorithmic coins that means the mint-and-burn arbitrage enters a reflexive death spiral; for delta-neutral synthetics it means sustained negative funding exceeds the reserve fund; for overcollateralized coins it means oracle failure or cascading liquidations, a scenario laid out in the Academy pieces on oracles and price feeds and composability and systemic risk.

Pattern 4 — Regulatory shock. An issuer is told it cannot redeem, cannot operate in a jurisdiction, or must restructure its reserves. The NYDFS-ordered BUSD wind-down in February 2023 is the archetype; MiCA's 2024 rollout produced milder versions affecting non-compliant euro-denominated tokens.

A disciplined reader asks: which of these four patterns is the headline describing? That single question disciplines everything that follows, including whether to expect recovery within hours or a total loss.

Case Study A — USDC, March 2023 (Reserve Doubt That Resolved)

On March 10, 2023, the California Department of Financial Protection and Innovation closed Silicon Valley Bank. Late on March 10 and into March 11, Circle disclosed through official channels that $3.3 billion of USDC's cash reserves — roughly 8% of total backing — were held at SVB. CoinDesk reported the figure the same day, citing Circle's statement.

The market reaction was immediate. According to secondary-market data reported by Decrypt over the weekend of March 11, USDC traded as low as roughly $0.87. DAI, USDD, and USDP — stablecoins with USDC exposure inside their own reserves — wobbled alongside it, a textbook case of composability contagion.

The resolution arrived in two steps. On Sunday, March 12, U.S. federal banking regulators announced that all SVB depositors would be made whole. Circle resumed redemptions on Monday, March 13. According to CoinDesk's reporting that day, USDC's peg returned to approximately $1 within roughly 72 hours of its low as Circle processed multi-billion-dollar redemption flows.

The reading lesson is precise. This was pattern 1, and the sub-question was access rather than existence. The collateral was real; the concern was that it could not be reached on a weekend with the banking counterparty in receivership. Once access was restored, the peg returned. Headlines that compared the event to Terra that weekend were wrong about the mechanism. A reader who asked the three-part question — is the collateral real, is it liquid, is it recoverable? — converged on the correct probability faster than one who pattern-matched on the word "depeg."

Case Study B — UST/LUNA, May 2022 (Algorithmic Collapse)

UST was an algorithmic stablecoin whose peg relied on a mint-and-burn arbitrage with LUNA, Terra's governance token. Holders could swap 1 UST for $1 of newly minted LUNA at any time, regardless of LUNA's market price. Stability depended entirely on LUNA retaining enough market capitalization to absorb redemptions.

According to academic post-mortems published via the Harvard Law School Forum on Corporate Governance, the sequence was compressed into a week. On May 7, 2022, large UST withdrawals from the Anchor protocol and from Curve's 4pool produced the initial selling pressure, and UST dipped to roughly $0.985. Some recovery followed on May 8. On May 9, around 15:00 UTC, UST lost its peg in earnest.

From there, the spiral was reflexive. UST redemptions forced LUNA minting; LUNA supply hyperinflated from roughly 345 million tokens to trillions within a week; LUNA's price collapsed from approximately $80 to effectively zero; and by May 13 UST was trading below $0.20. A revived chain labelled Terra 2.0 launched on May 28 as a restart without the algorithmic stablecoin.

The reading lesson is that this was pattern 3, and it was pre-warned. Academic researchers had flagged the circular-collateral problem for more than a year; the unsustainable 20% Anchor yield was a visible canary. The newsworthy event was not the collapse itself but the earlier signal that LUNA's float had grown too small to absorb the UST it was supposed to backstop. For any algorithmic design, the defining question is the ratio of stabilization-mechanism capacity to outstanding stablecoin supply. Once that ratio inverts, the peg is a countdown.

Case Study C — Ethena's USDe, 2024–2025 (Delta-Neutral Under Stress)

According to Ethena Labs' own documentation, USDe launched publicly on February 19, 2024. Its backing is not fiat reserves but a cash-and-carry trade: long spot ETH, BTC, and liquid staking tokens held at off-exchange custody providers such as Copper, Ceffu, and Fireblocks, paired with an equal short perpetual-futures position on major venues including Binance, Bybit, and OKX. The short perp is supposed to exactly offset the long spot; yield comes from perpetual funding payments.

Ethena's documentation identifies two principal failure modes. The first is sustained negative funding rates that erode the reserve-fund buffer faster than it can be replenished. The second is exchange or custody counterparty failure during settlement windows when traditional markets are closed.

Both failure modes have had real-world stress tests. The February 2025 Bybit hack and the October 10, 2025 flash crash — both of which unfolded into weekends when TradFi was closed — put the design under observable pressure. According to Ethena's own disclosures and to Coin Metrics' State of the Network #335 analysis, funding remained net positive through the February 2025 event, but the October 2025 episode saw funding flip sharply negative across venues with the most severe secondary-market dislocation in USDe quotes concentrated on Binance. Neither event was a full depeg; both were localized. Ethena's Reserve Fund and its increasingly diversified holdings — with a larger allocation to liquid fiat-backed stables and Treasuries than at launch — absorbed the stress.

The reading lesson is that this is pattern 2 and pattern 3 mixed, and the distinction matters. A secondary-market dislocation during an exchange-specific liquidity vacuum is not the same event as a mechanism failure, even if the headline candle looks similar. When a USDe story runs, the useful questions are: what is the current perpetual-funding rate, what is the size of the reserve fund relative to mark-to-market losses, and which CEX is the stress localized on?

Regulatory Convergence — MiCA and the GENIUS Act

The regulatory picture has narrowed visibly since 2023. Two regimes are now doing most of the work.

European Union. According to the European Banking Authority, MiCA's stablecoin provisions went live on June 30, 2024. The regulation defines two categories. E-Money Tokens (EMTs) are single-fiat-referenced and can be issued only by authorized credit institutions or e-money institutions. Asset-Referenced Tokens (ARTs) are basket or multi-asset tokens subject to a separate authorization regime and stricter reserve rules. Purely algorithmic stablecoins with no real-asset backing are effectively shut out of the EU market.

United States. The Guiding and Establishing National Innovation for U.S. Stablecoins Act — the "GENIUS Act," S.1582 of the 119th Congress — passed the Senate 68–30 on June 17, 2025 and the House 308–122 on July 17, 2025. According to the White House and to analyses published by Latham & Watkins and Covington & Burling, President Trump signed it into law on July 18, 2025. The Act creates a federal "payment stablecoin" category requiring 1:1 backing by USD cash, short-dated Treasuries, or equivalently liquid assets; it licenses issuers via the OCC, FDIC, or approved state regulators; and it prohibits purely algorithmic stablecoins that lack real-asset backing. It also carves payment stablecoins out of the Securities Act and the Commodity Exchange Act. Operatively, the law takes effect on the earlier of 18 months after enactment or 120 days after primary regulators publish final rules, and the FDIC has already begun rulemaking for issuer application procedures.

The convergence point for a reader is that both regimes now require reserves to be high-quality liquid assets, both prohibit purely algorithmic designs, and both license issuers. This narrows the universe of patterns 1 and 4 that a reader is likely to encounter in major jurisdictions. Reserve-doubt events should become rarer and more technical; regulatory-shock events should concentrate around licensing edges rather than surprise enforcement.

A Checklist for Reading the Next Stablecoin Headline

Identify the design. Fiat-backed, crypto-backed, algorithmic, or delta-neutral synthetic? This alone determines which failure modes are in scope.
Identify the pattern. Reserve doubt, liquidity vacuum, mechanism failure, or regulatory shock? The headline usually gestures at one; the reader's job is to name it precisely.
Find the primary source. Issuer disclosure (Circle, Tether, Paxos, Ethena), regulator filing, or on-chain evidence. Secondary reporting is useful but should trace back to one of those three.
Size the problem. What fraction of reserves or backing is at risk? For USDC in March 2023 it was roughly 8%. For UST the LUNA-to-UST market-cap ratio had already inverted into dangerous territory before May 9. For Ethena the relevant number is the reserve fund versus the daily drag from negative funding.
Check the composability ring. Which protocols treat this coin as a dollar input? Lending markets, LP pools, and other stablecoins' reserves all propagate stress — see the composability and systemic risk write-up for how the transmission works.
Default to time-boxed probabilities, not verdicts. Reserve-doubt events with clean regulator resolution tend to unwind in days. Mechanism failures in algorithmic designs tend not to unwind at all. Headlines that conflate the two are usually wrong by Monday.

The headline is a prompt, not an answer. The framework above turns the prompt into a question a reader can actually resolve with the next three paragraphs of the story.