A depeg is when a stablecoin's market price departs meaningfully from the value it is supposed to hold, and stays there. It is the defining failure mode of the entire instrument class — the one event that turns a boring utility into a live problem — and it is also widely misdiagnosed, because several completely different things produce the same symptom. A token trading below a dollar might be a bank failure, a reserve scandal, a broken bridge, or a thin market on one exchange, and those situations call for opposite responses. This report is about telling them apart and knowing what to do. It builds on the peg mechanics set out in how stablecoins stay pegged.
01 — What actually counts
Precision matters here because alarm is cheap. A stablecoin trading a fraction of a percent away from a dollar is not depegged — it is functioning. Its price is set by supply and demand on exchanges, and small deviations simply mean the arbitrage that restores par has not yet been worth anyone's time and transaction costs. Screenshots of a token at slightly off a dollar circulate constantly and mean nothing.
Three properties separate noise from a genuine break. Magnitude: a deviation large enough that the arbitrage would obviously be profitable, yet nobody is taking it. Persistence: minutes are noise, hours are a signal, days are a verdict. And breadth: a gap visible across many venues simultaneously reflects the token, whereas a gap on one exchange reflects that exchange. The most useful single question is not how far the price has moved but why the correction is not happening — because the absence of arbitrage, rather than the price itself, is what tells you something real is wrong.
02 — Cause one: the backing was never there
The most fundamental cause is that the assets supposedly held against the tokens do not exist, are insufficient, or cannot be liquidated at anything like the value claimed. When that becomes known or credibly suspected, holders sell, and no arbitrageur buys the discount because they can see there is nothing at par to redeem into.
This category is terminal rather than temporary. There is no resolution that restores the peg, because the peg was never supported in the first place, and the price is simply discovering what the token is actually worth. It is also the rarest cause among major tokens today, precisely because reserve composition and disclosure are now the most scrutinised aspect of the sector and, in regulated markets, a statutory requirement — the questions to ask are set out in our note on reading an issuer's reserves. A depeg from this cause is the one that does not come back.
03 — Cause two: the backing is fine, the route is blocked
The second cause is entirely different and far more common. The reserves exist and are adequate, but something has interrupted the ability to reach them — a bank holding a portion of the cash has failed, a banking partner has withdrawn, a weekend or holiday has closed the settlement window, or the issuer has paused redemptions operationally.
Here the token falls not because it is unbacked but because the arbitrage loop is broken: nobody can buy the discount and redeem at par, so nothing pushes the price back. The most instructive example in the sector's history is precisely this — a major regulated token fell sharply over a weekend when part of its cash sat in a failed bank, and recovered fully once those deposits were made whole. Nothing about the design had failed; a bank had. This category is usually survivable, frequently recovers completely, and is the one where a discount can genuinely represent an opportunity for anyone able to verify the facts. It is also, uncomfortably, more likely for tokens deeply integrated with the traditional banking system than for those outside it, as our comparison in USDT vs USDC discusses.
The price tells you a stablecoin has broken. Only the cause tells you whether it is coming back — and those two questions get answered on completely different timescales, which is why panic and patience are both sometimes correct.
04 — Cause three: the mechanism eats itself
The third cause applies to designs whose stability depends on a related token rather than on external assets. When confidence falls, the mechanism intended to restore the peg issues more of a companion token that is falling for the same reason, accelerating the decline — the reflexive failure examined in our note on algorithmic stablecoins.
Two features distinguish this category in practice. It is extraordinarily fast: where a reserve-driven depeg unfolds over days as facts emerge, a reflexive one can complete in hours, because the mechanism itself is doing the selling. And it is effectively irreversible once past a certain point, since the thing that would restore the peg is confidence in an asset that the collapse has already destroyed. For a holder, this is the category where hesitation is most expensive and where the usual advice to wait for clarity is precisely wrong.
05 — Cause four: the wrapper, not the token
Here is the cause almost nobody warns about, and it catches experienced users. Stablecoins exist on many networks, and not every version is the same instrument. A token may be issued natively by the issuer on one chain and exist elsewhere as a bridged representation — a claim on tokens locked in a bridge contract, issued by that bridge rather than by the issuer.
Those bridged versions can depeg on their own while the native token is entirely unaffected. If the bridge is exploited, halted, or simply doubted, the wrapped version loses its redemption route and trades at a discount, sometimes a severe one, while the original sits calmly at a dollar on its home chain. The lesson is practical: when you hold a dollar token on a smaller network, establish whether it is issued natively there or bridged, because you may be holding an obligation of a bridge rather than of the issuer whose name is on it. This is a distinct risk from anything in the issuer's reserves, and it is invisible from the ticker — a variety of the settlement complexity described in how crypto settlement works. A related and milder version occurs venue-side: a single exchange with thin books or halted withdrawals can print a discount that exists nowhere else.
06 — Depegs upward
Tokens sometimes trade above par, which sounds harmless and is worth understanding. It generally means demand to acquire the token exceeds the immediate ability to create new supply — a rush into a perceived safe asset during market stress, a squeeze on a particular chain, or minting capacity constrained by banking hours.
Two implications follow. Buying a stablecoin above a dollar is straightforwardly paying more than a dollar for a dollar, which is a poor trade unless the premium is buying you something specific such as access on a particular network. And a persistent premium is a signal in itself: it says the primary market is not keeping up with demand, which is the same plumbing weakness that produces downward depegs, observed from the other side. An upward deviation is not a crisis, but it is information about the same machinery.
07 — What to do while it is happening
The instinct to act immediately is the enemy, because the correct action differs entirely by cause — and the first thing worth doing is establishing which one you are in. Check whether the gap appears across multiple venues or only one, since a single-venue discount is usually a venue problem. Check whether you hold a native or bridged version, since that determines whose obligation you are actually holding. Look for a specific, verifiable cause rather than a rumour, and check whether redemptions are still being processed — because that, more than any statement, tells you whether the arbitrage loop is intact.
From there the responses diverge. Where the cause is reflexive, speed matters more than analysis, because the mechanism does not recover. Where the cause is a blocked route and the backing is verifiable, selling into a panic discount is how holders convert a temporary problem into a permanent loss. Where the cause is a bridge or a venue, the answer may be moving to the native chain or a different exchange rather than exiting at all. And where you genuinely cannot tell, remember that reducing an oversized position is a decision you can take without a diagnosis. Two habits are worth adopting in advance: never hold more in any single stablecoin than you would accept losing a portion of, and know your redemption or exit route before you need it, since discovering it during a depeg is discovering it too late.
08 — Watching for it
Depegs are rarely completely unheralded. The reliable precursors are a redemption window that becomes slower or narrower, a banking or custody relationship publicly ending, a reserve disclosure that becomes less specific rather than more, an unusually high yield being offered to attract or retain balances, and thinning depth in the token's own trading pairs — the last of which is observable directly, as described in market depth.
The honest closing point is one of proportion. Well-collateralised stablecoins have held through severe market stress repeatedly, and the two largest have survived events that destroyed much of the surrounding industry. Depegs are the tail risk of this instrument, not its normal behaviour. But they are a real tail, they arrive faster than most risks, and the difference between a survivable episode and a total loss is usually determined in the first hours by whether the holder correctly identified which of the four causes they were looking at — which is the entire reason to have thought about it beforehand.
"A prudent man foreseeth the evil, and hideth himself: but the simple pass on, and are punished." — Proverbs 22:3
Methodology & Sources
This report describes a failure mode and how to assess it; it contains no token names, issuer names, price levels, deviation percentages, or dated incident details beyond what is necessary to describe categories of cause — readers should consult primary accounts of specific events and current issuer disclosures. The threshold criteria offered for distinguishing ordinary deviation from a genuine depeg — magnitude sufficient to make arbitrage obviously profitable, persistence beyond short intervals, and breadth across multiple venues — are analytical guidance rather than a standard definition; no numerical threshold is asserted because none is universally accepted. The four causes described are an organising taxonomy: insufficient or unverifiable backing; adequate backing with an interrupted redemption or settlement route; reflexive collapse in designs dependent on a companion token; and failures affecting bridged or venue-specific representations rather than the natively issued token. The reference to a major regulated token falling and recovering after a portion of its cash reserves was held at a failed bank reflects a well-documented 2023 episode. The distinction between natively issued and bridged stablecoin representations, and the observation that bridged versions can trade at a discount independently of the native token, reflects standard cross-chain architecture. Guidance on holder responses is general risk-management commentary, not advice, and cannot account for individual circumstances; anyone acting during a live depeg should verify facts independently and in real time. Nothing here is a recommendation regarding any token, issuer, venue, or trade.