The definition we use
A "depeg" in this catalogue means USDT trading more than 1% away from one US dollar on a major spot venue (Binance, Coinbase, Kraken, OKX, Bitfinex) for at least one hour. Brief sub-1% deviations happen weekly and are not interesting. The threshold is arbitrary but it captures the events that crypto-Twitter usually discusses without including the constant low-amplitude noise.
We use Coinbase and Kraken USD pairs as the reference where they exist, since those are the cleanest dollar quotes. For events where USDT does not list against fiat on a US venue, we use Binance USDC pairs or the spot USDT-USD price as published by Tether's own dashboard.
October 15, 2018 · the Bitfinex banking event
In early October 2018, Crypto Capital Corp.'s accounts — the payment processor handling Bitfinex's fiat flows — were seized in Poland, Portugal and the United States. Bitfinex customer withdrawals slowed dramatically. USDT, which was already loosely associated with Bitfinex in the market's mind, started to trade below peg as holders rotated to other tokens or to BTC.
On October 15, 2018, the price gap widened. USDT traded as low as 0.85 on Kraken's USDT-USD pair, the venue's lowest historical print. The cause was not a reserve shortfall (the 2021 NYAG settlement later established that intercompany lending was used to cover the gap, but the tokens were not unbacked from the holder's perspective at the time). The cause was an inability to redeem at scale combined with public uncertainty about the partner exchange.
Recovery took roughly three weeks. By early November, USDT was back within 1% of peg. The episode established a template: USDT depegs are usually about access to redemption, not about reserve composition.
March 12, 2020 · the COVID crash
On Black Thursday (March 12, 2020), BTC fell roughly 50% intraday. Maker's DAI experienced its first major liquidation crisis (the auction failures that produced the protocol-debt shortfall). USDT briefly traded to 0.962 on Kraken as holders rotated to USD on US venues.
The episode was shallow and short. USDT recovered within hours as the broader market stabilised. The cause was generic liquidity stress, not Tether-specific. Worth noting only because it is one of the cleanest examples of USDT depegging on the same day that DAI depegged more severely (DAI traded as low as 0.945 on the same venues).
May 12, 2022 · the Luna / UST contagion
UST collapsed beginning May 9, 2022, and was effectively at zero by May 12. The contagion to USDT was indirect: holders, having watched a 19-billion stablecoin go to zero, asked the obvious question about the next-largest stablecoin. USDT redemption requests spiked. Tether processed roughly 7 billion of redemptions within a week, all at par, all in cash — the cleanest stress test the issuer had been through to that point.
USDT traded as low as 0.9485 on Kraken USDT-USD on May 12, with deeper prints on some smaller venues. The cause was contagion fear, not a Tether-specific event. The recovery took about five days, mostly driven by the visible processing of redemptions at par. We covered the Luna postmortem in detail in why algorithmic stablecoins fail.
June 13, 2022 · the Celsius freeze
Celsius froze withdrawals on June 12, 2022, and filed for bankruptcy a few weeks later. The same week, Three Arrows Capital was facing its own liquidation crisis. The broader market dropped sharply (BTC fell from roughly 30,000 to 22,000 over five days). USDT briefly traded to 0.991 on Kraken, mostly at the bottom of a flash candle, and recovered within 24 hours.
This event is included for completeness — it sits right at the 1% threshold and lasted briefly — but it is borderline. The pattern (CeFi lender stress producing a small USDT discount as some holders rotate to cash) repeats in 2022 and 2023 in less acute forms.
March 11, 2023 · the SVB weekend (in reverse)
The one event in the catalogue that is a "depeg" in the technical sense (more than 1% from a fair-value USD price) but on the upside. As USDC fell to 0.8774 on Coinbase over the SVB weekend, USDT traded at a small premium on global venues. The Binance USDT-USDC pair priced USDC at roughly 0.90 against USDT for several hours, implying a USDT premium against notional USD of around 1.005 to 1.010.
The premium was modest and lasted hours, not days. It is worth cataloguing because it is one of the cleanest examples of USDT and USDC moving inversely under stress — which is the structural argument for holding both. We reconstructed the same weekend from the USDC side in the SVB weekend hour by hour.
June 15, 2023 · the Curve 3pool incident
A single wallet dumped roughly 31 million USDT into Curve's 3pool on June 15, 2023, shifting the pool composition to roughly 73% USDT / 27% USDC+DAI within an hour. On-chain USDT prices on Curve briefly traded as low as 0.985. Spot venues never went below 0.997.
The event is interesting because it is a venue-specific liquidity stress, not a market-wide signal. Cross-venue arbitrage rebalanced the pool within hours. Tether processed normal redemptions over the same period. We mention it because it is the kind of event that is often cited as "USDT depeg" headlines and is, in reality, a single-pool composition shift with no broader implications.
Pattern · what every depeg has in common
| Event | Underlying cause | Tether-specific or market-wide | Recovery driven by |
|---|---|---|---|
| Oct 2018 | Crypto Capital seizure | Tether-adjacent (partner exchange) | Restored redemption rails |
| Mar 2020 | COVID liquidity event | Market-wide | Broad market stabilisation |
| May 2022 | Luna / UST collapse | Contagion | Visible processing of redemptions at par |
| Jun 2022 | Celsius / 3AC stress | Contagion | Time |
| Mar 2023 | USDC SVB (premium, not discount) | Inverse contagion | USDC recovery |
| Jun 2023 | Curve 3pool dump | Venue-specific | Arbitrage |
The common feature: most USDT depegs are about access, not about reserves. The 2018 event was a partner-exchange access problem. The 2022 events were contagion-driven access problems (holders worried they would not be able to redeem). The 2023 events were venue-specific liquidity. In every documented case, Tether processed redemption requests at par when presented; the depeg was driven by perceived rather than actual reserve risk.
What would a real reserve-driven depeg look like
We have not seen one in USDT's history. The closest hypothetical is a scenario where Tether's reserve composition is found to include a major asset class that has lost value (the 2021 commercial paper book, had a major issuer in the book defaulted, would have been the candidate). The reserve composition has shifted since then toward US Treasury bills and overnight repos, which makes that vector less plausible. The remaining real-reserve vectors are: a custodian bank failure that traps the cash residue, a software vulnerability that allows unauthorised minting, or a regulatory action that freezes a portion of reserves. None of these have happened.
A reserve-driven depeg would, in our view, look different from any of the above events. The recovery would be slower because the underlying cause would need to be resolved rather than the perception. The depth could be deeper because the redemption rail itself would be impaired. The market reaction would include a sustained USDT-to-USDC rotation (rather than the brief premium of 2023) because the inverse trade would have a real basis. We have no historical example to point to. The pattern below is forward-looking inference, not retrospective.
The framework the desk uses for the next one
Step 1 · Identify the trigger category
Is the trigger market-wide (Black Thursday, Luna), partner-specific (Bitfinex / Crypto Capital, possible Binance event), reserve-specific (commercial paper default, custodian bank failure), or venue-specific (Curve pool dump)? Each category has a different recovery profile. The first two recover when the market or the partner stabilises. The third recovers when the reserve issue is resolved. The fourth recovers within hours through arbitrage.
Step 2 · Check the redemption rail
The single signal that distinguishes a real depeg from a perceived one is whether Tether is processing redemptions at par. The treasury bills, money-market funds and overnight repos that make up most of current reserves can be converted to cash in 24 to 48 hours. If Tether continues to process redemptions through a depeg, the depeg is perceived, not real. If redemptions are slowing or paused, the depeg is real. The Tether dashboard publishes mint and burn data publicly; the data are slightly delayed but visible.
Step 3 · Watch the USDC-USDT spread
A USDT-specific event produces a USDC premium (USDC trades above USDT). A market-wide event produces both tokens trading at a discount to USD. A USDC-specific event produces a USDT premium. The spread is a cleaner signal than the absolute price on either token because it removes market-wide noise. The desk runs a spread alert at 0.5% deviation from parity.
Step 4 · Hold both tokens before, not during, the stress
The diversification argument for holding USDT and USDC works only if positions are established before the stress event. Inter-token swaps during a depeg execute at the discount; rotating during the event locks in losses on whichever side is depressed. The desk's working policy is to hold both, in roughly 60% USDT / 40% USDC weights, adjusted by jurisdiction.
What this catalogue is not
It is not an argument that USDT cannot have a different kind of stress in future. The 2018 to 2023 record is six events, all of which Tether handled at par. The next event might look different. The catalogue is intended as base rate information: holders who pattern-match too literally on any single event (especially the 2018 episode) tend to overestimate or underestimate the next one. Treat the catalogue as one input, not a forecast.
If you want to act on this
The narrow action is to split stablecoin holdings between USDT and USDC. The broader action is to keep a portion of working balance off-exchange (in self-custody) so that exchange-specific or partner-specific stress does not lock the position. The desk's working venue is Binance for spot trading; the Binance referral link uses code BN16188 and does not change your fees.
Each event traced back to primary source
- Kraken historical USDT-USD ticker data, 2018-10 through 2023-12, accessed via Kraken public API and TradingView.
- Coinbase historical USDT-USD ticker (USDT listed on Coinbase from May 2021).
- Curve.fi 3pool subgraph data, queryable on TheGraph for the relevant periods.
- Tether transparency dashboard (tether.to/transparency) for mint / burn timestamps.
- 2021 NYAG settlement order, "In the Matter of iFinex Inc., Tether Holdings Ltd, et al.", for the 2018 background; covered separately in our 2021 settlement piece.
- Maker Protocol forum, Black Thursday post-mortem (forum.makerdao.com), for the March 2020 cross-event detail.
Spotted an error or a missing episode? Write to [email protected] with the line; the corrections log is on the corrections page.