.png){kind=logo}
A stablecoin depeg is not a single kind of event. Two of the largest ones in crypto history had almost nothing in common: one was a design that collapsed exactly as its critics predicted, the other was a fully backed coin that broke because a bank failed. Telling them apart matters more than memorising the prices.
This article covers what a depeg is, the three stablecoin designs and how each fails, the historical record with the venue for every price, and how to check a stablecoin's backing yourself. It is a reference piece, not investment advice. For the wider role these instruments play, see our overview of stablecoins in global digital payments.
What does a depeg actually mean?
A depeg is when a stablecoin trades away from the value it is designed to track, usually one US dollar. That definition is easy. The problem is the number people attach to it.
A stablecoin does not have one price. It has a price on every venue and in every trading pair, and during a panic those prices diverge sharply. On 15 October 2018, USDT traded at $0.51 against USDC on Poloniex. That figure is real and it gets quoted often. It also describes one pair on one exchange with a thin order book on that day, not what USDT was worth across the market.
So a claim like "USDT fell to 51 cents" is not wrong so much as incomplete. Without the venue and the pair it cannot be checked, compared, or reasoned about. Every price in this article names where it came from, and where we could not establish the venue we say so.
There is a second distinction worth making early. A stablecoin trading below peg on an exchange is a market event. A stablecoin that cannot be redeemed at par by the people entitled to redeem it is a solvency event. These often happen at the same time, which is why they get conflated, but they are different failures with different consequences.
What are the three designs, and how does each one break?
Almost every stablecoin falls into one of three categories, and the category tells you what its failure will look like.
| Design | How the peg is held | How it breaks | Examples |
|---|---|---|---|
| Fiat-backed | Issuer holds cash and short-term instruments, redeems at par | The reserves become unavailable or their quality is doubted. The design is sound; the custodian is the weak point | USDC, USDT |
| Crypto-collateralised | Over-collateralised with on-chain assets, liquidated automatically if coverage falls | Collateral falls faster than liquidations can clear, or the collateral is itself a stablecoin that broke | DAI |
| Algorithmic | Mint and burn against a paired volatile token, arbitrage restores the peg | Confidence in the paired token falls, so the arbitrage that is supposed to fix the peg destroys the token instead | TerraUSD |
Newer designs sit awkwardly across these categories. Synthetic dollars backed by hedged derivative positions, discussed in our piece on Ethena and whether it repeats Terra's mistakes, are neither straightforwardly fiat-backed nor algorithmic in the Terra sense, and their failure modes have not yet been tested by a full crisis.
The important consequence: in the first two designs a depeg is a liquidity or custody problem that can resolve. In the third, a deep enough depeg is self-reinforcing, because the mechanism that restores the peg works by printing the asset whose price is collapsing.
What is the historical record?
The dates and lows below are the ones we could source. Where a low is venue-specific, the venue is named, because as explained above that is the only form in which the number means anything.
| Date | Coin | Low | Trigger | Outcome |
|---|---|---|---|---|
| 15 Oct 2018 | USDT | $0.51 against USDC on Poloniex | Doubts about Tether's banking relationships | Recovered; reserves not shown to be impaired |
| 7 to 12 May 2022 | UST | Fell to a fraction of a cent | Large UST outflows as Anchor yields were being reduced | Never recovered. Around $50bn of value erased |
| 12 May 2022 | USDT | $0.941 | Contagion from the UST collapse | Recovered within days; reserves unaffected |
| 11 Mar 2023 | USDC | $0.87 on Kraken, roughly 13% below peg | $3.3bn of reserves trapped at Silicon Valley Bank | Recovered in about three days after regulators guaranteed SVB deposits |
| 11 Mar 2023 | DAI | Below $0.90 | Held USDC as collateral, so it inherited the depeg | Recovered alongside USDC |
Case study: TerraUSD, May 2022
TerraUSD is the case where the design itself failed, and it is worth understanding in mechanical detail because the failure was not an accident.
At its April 2022 peak UST was the third largest stablecoin with a market capitalisation of about $17.5bn. Roughly 75% of it sat in Anchor Protocol, which paid depositors close to 20%. That rate was not generated by lending activity. It was subsidised by Terraform Labs, and by April 2022 the subsidy was running at around $6m per day. The Terra community had already voted to reduce the rate, with the taper starting on 1 May 2022.
The peg itself relied on a mint-and-burn arbitrage with LUNA. If UST traded below a dollar, anyone could burn UST to mint a dollar's worth of LUNA and sell it, which was supposed to remove UST from supply and push the price back up. This works when LUNA has a deep and confident market. It does the opposite when it does not.
On 7 May 2022 a large swap, widely reported as 85m UST into USDC, moved the price. UST slipped below a dollar that evening and lost the peg decisively on 9 May. Holders then did what the mechanism invited them to do: they redeemed UST for LUNA and sold. Meeting that demand required minting enormous quantities of new LUNA, which crashed its price from above $80 to a fraction of a cent inside a week. The lower LUNA went, the more of it each redemption produced, and the more selling pressure it created.
That is the part worth carrying away. The arbitrage did not fail to operate. It operated exactly as designed, and operating as designed is what destroyed it. Academic post-mortems, including work published through Harvard Law School's corporate governance forum and MIT's Center for Finance and Policy, treat the episode as a classic run: the yield attracted deposits that were never sticky, and the redemption mechanism converted an exit into hyperinflation of the collateral.
Case study: USDC, March 2023
USDC is the opposite case, and that is exactly why it is instructive. Nothing about its design failed.
On 9 March 2023 the California Department of Financial Protection and Innovation closed Silicon Valley Bank. Circle then disclosed that about $3.3bn of the cash reserves backing USDC, close to 8% of them, were held at SVB. USDC was fully backed the whole time. The question was not whether the assets existed but whether they could be reached.
By the early hours of 11 March, USDC had fallen to $0.87 on Kraken, its lowest price to date. DAI followed it down below $0.90 for the simple reason that DAI held USDC as collateral. A fully backed stablecoin and an over-collateralised one broke together, through the same single point of failure.
The recovery was equally instructive. It did not come from arbitrage or from the market working things out. On the Sunday, regulators announced that all SVB depositors would be made whole, waiving the standard $250,000 insurance limit. The peg returned within roughly three days of breaking. What restored it was a policy decision about a bank, not anything that happened on-chain. That dependence on the traditional banking perimeter is a large part of why regulators moved on the sector afterwards, as covered in our reporting on US Treasury stablecoin guidelines and the MAS framework in Singapore.
What do the two cases share, and where do they diverge?
| TerraUSD | USDC | |
|---|---|---|
| Were the assets there? | No. Backing was a token it printed itself | Yes, in full |
| What triggered it | Large outflows as the subsidised yield was withdrawn | A bank failure |
| Did the mechanism help? | It accelerated the collapse | Not relevant; the constraint was access to reserves |
| What ended it | Nothing. The system reached zero | A government guarantee of bank deposits |
| Time to resolution | Never | About three days |
Both events began the same way, with holders trying to leave faster than the system could accommodate. That is a run, and runs are a property of the liability structure rather than of the technology. What differed was what sat underneath. One had reserves that were temporarily unreachable. The other had no external reserves at all.
What actually restores a peg?
Three things can pull a stablecoin back, and they are not equally reliable.
- Redemption at par. The strongest. If holders who are entitled to redeem can do so for a dollar, the market price cannot stay far below a dollar for long, because buying below par and redeeming is close to riskless. This depends on redemption being open and operational, which is not always the case during a crisis.
- Arbitrage against collateral. Works when the collateral is independent of the stablecoin and holds its value. It is precisely the case where the collateral is the paired token that this breaks down, as TerraUSD demonstrated.
- External intervention. What actually resolved March 2023. It is also the one nobody can count on in advance, and treating it as part of a stablecoin's risk profile would be a mistake.
Which warning signs are real?
Some commonly cited red flags carry information and some do not.
| Signal | Worth |
|---|---|
| Yield far above prevailing short-term rates, paid by the issuer | Real. A subsidy has to end, and deposits attracted by a subsidy leave when it does. This was explicit in Terra's case |
| Reserves concentrated in one counterparty | Real. USDC was fully backed and still broke, because 8% of reserves sat in a single failing bank |
| Reserve composition not disclosed, or disclosed rarely | Real, though absence of disclosure is not evidence of absence of reserves |
| Holding another stablecoin as collateral | Real. DAI inherited USDC's depeg directly |
| A brief price wobble on one exchange | Weak. Thin books produce prints that do not reflect redeemability |
| Large market capitalisation | Not protective. UST was the third largest stablecoin weeks before it reached zero |
How can you check a stablecoin's backing yourself?
- Identify the design first. Fiat-backed, crypto-collateralised or algorithmic. Everything else follows from this, and it is usually stated in the issuer's own documentation.
- Find out who holds the reserves and where. Concentration in one bank or one custodian is the specific risk that broke USDC.
- Check what the reserves are made of. Cash and short-dated government paper behave differently from commercial paper, corporate debt or other crypto assets under stress.
- Read the redemption terms. Who may redeem, at what minimum size, and under what conditions the issuer may suspend it. Retail holders often cannot redeem directly at all.
- Ask where any yield comes from. If the answer is the issuer's own balance sheet rather than an underlying economic activity, treat the rate as temporary.
- Know where you are holding it. Custody risk sits on top of issuer risk. Our guides to crypto wallet basics and staying safe in crypto cover that layer.
- Compare prices across venues. A discount on one exchange and not others points to a local liquidity problem. A discount everywhere at once points to something about the coin.
How confident is each claim here?
| Claim | Basis | Strength |
|---|---|---|
| Circle held $3.3bn of reserves at SVB | Circle's own disclosure, reported contemporaneously | Established |
| USDC fell to $0.87 on Kraken | Exchange price data reported at the time | Established for that venue |
| UST's mint-and-burn accelerated the collapse | Mechanism plus academic post-mortems | Strong |
| Roughly 75% of UST sat in Anchor | Contemporaneous reporting of protocol data | Strong |
| USDT traded at $0.51 in Oct 2018 | One pair on one exchange | Venue-specific, not a market-wide price |
| The 85m UST swap triggered the depeg | Widely reported, but a single trade is hard to establish as a cause | Weak as causation |
Frequently asked questions
Has any stablecoin ever recovered from a total collapse?
No. TerraUSD did not. Once LUNA's supply had hyperinflated, there was no mechanism left that could restore the peg. Every stablecoin in the record above that recovered had external reserves that survived the event, whether or not they were reachable at the time.
Does a depeg mean the issuer is insolvent?
No, and conflating the two is the most common error. USDC traded at $0.87 while being fully backed. The market was pricing uncertainty about access to reserves over a weekend, not their absence.
Are algorithmic stablecoins inherently unsafe?
The design has one structural weakness: the asset backing the peg is one the system issues itself, so its value is correlated with confidence in the peg. That correlation is what turns an exit into a spiral. Whether a particular implementation can manage it is a separate question, but the weakness is a property of the category.
Why did DAI depeg when it is over-collateralised?
Because part of that collateral was USDC. Over-collateralisation protects against the collateral falling in value; it does not help when the collateral is a different stablecoin experiencing the same event.
How long do depegs usually last?
There is no useful average, because the sample is small and the outcomes are bimodal. Backed coins in the record above recovered in days. TerraUSD never did. The design tells you far more than any historical duration.
Sources
- Circle confirms $3.3bn of USDC's cash reserves held at Silicon Valley Bank. CoinDesk, 11 March 2023. coindesk.com
- USDC stablecoin regains dollar peg after Silicon Valley Bank-induced chaos. CoinDesk, 13 March 2023. coindesk.com
- Anatomy of a Run: The Terra Luna Crash. Harvard Law School Forum on Corporate Governance, 22 May 2023. corpgov.law.harvard.edu
- Anatomy of a Run: The Terra Luna Crash. MIT Sloan Center for Finance and Policy. mitsloan.mit.edu
- Tracing Stablecoin Contagion during the USDC Depeg after the Silicon Valley Bank Collapse. arXiv. arxiv.org
- Stablecoins: A Deep Dive into Valuation and Depegging. S&P Global, September 2023. spglobal.com
- History of Tether's peg: every time USDT traded above or below one dollar. Protos. protos.com
This article is a factual reference and does not constitute investment, financial or legal advice. Prices cited are historical and venue-specific. Do your own research before making any decision about digital assets.
