On most days a big stablecoin trades within a fraction of a cent of $1. This study counts the days when one did not, using one DefiLlama price per UTC day for 16 stablecoins pegged to the US dollar, from December 30, 2020 to November 1, 2026: the days each spent more than 0.5% and more than 2% from $1, its largest gap and every break beyond 2%. Because one price a day can sleep through a crisis, it also sets the daily figures against intraday lows recorded by Federal Reserve and Treasury researchers. For how stablecoins fit into crypto, start at the crypto hub.
| Measure | Value |
|---|---|
| Data window | Dec 30, 2020 to Nov 1, 2026 |
| USDT days more than 0.5% from $1 | 48 of 2127 |
| USDC days more than 0.5% from $1 | 53 of 2127 |
| USDC largest daily deviation | -3.89% on Mar 12, 2023 |
| Episodes more than 2% from $1 (all coins) | 19 |
Key findings
- The largest coins stayed close. USDT was more than 0.5% from $1 on 48 of 2,127 days and never beyond 2%; USDC was more than 0.5% off on 53 days and DAI on 48, each with one day beyond 2%.
- Daily data hides the lows. USDC's lowest daily reading, 3.89% below $1 on March 12, 2023, sits about 10 cents above the 86-cent trough that Federal Reserve researchers reported.
- Big breaks were brief. Of 19 episodes beyond 2%, 11 had a single day past the line, and only TerraUSD's lasted more than two weeks.
- One outright failure. TerraUSD's last break began on May 9, 2022, and its lowest reading was 99.36% below $1.
- Thin and retired coins drift. LUSD was more than 0.5% off on 41% of its days, FRAX on 26.3% and BUSD on 15.4%.
- A quiet stretch. USDT, USDC and DAI had no day more than 0.5% off in 2025 or in 2026 through November 1.
How were depegs measured?
Each coin gets one DefiLlama price per UTC day, and a day counts when that price sits more than 0.5% or more than 2% from $1.
| Measure | How it is calculated |
|---|---|
| Daily reading | DefiLlama's price for the coin, one per UTC day, timestamped at midnight UTC; the unfinished current day is left out |
| Deviation | (price − $1) ÷ $1, in percent; positive means a premium above $1, negative a discount |
| Days off by more than 0.5% or 2% | Daily readings whose deviation, in either direction, is larger than the threshold |
| Share of days | Days off by more than 0.5%, divided by all of that coin's daily readings |
| Largest deviation | The single reading furthest from $1, above or below, with its date |
| Longest run beyond 0.5% | The most consecutive calendar days beyond 0.5%; a missing day ends the run |
| Episode beyond 2% | A coin's days beyond 2%, joined into one episode when no more than 3 days separate them; the day count includes only days actually beyond 2% |
| Days by year | Days more than 0.5% from $1 in each calendar year, for USDT, USDC and DAI only |
The window starts on December 30, 2020, the first day of DefiLlama's stablecoin price history, and gives 2,127 readings for coins priced throughout; coins launched later enter on their first priced day. GUSD, FRAX and LUSD trade too thinly for their swings to say much about reserves, so they stay out of the episode list, as do BUSD's episodes from January 1, 2024, after its wind-down. Tables round deviations to one decimal place, and the text quotes two, as the data does. The glossary entry on a stablecoin depeg covers the term itself.
The feed-error filter
Price feeds glitch, so the study drops readings that look like faults rather than trades: one or two days in a row more than 10% from $1, with a reading within 3% of $1 on each side, or nothing after them at the end of a coin's data. Six readings were removed, all from BUSD, GUSD, FRAX or LUSD. Four fell on January 2 and 3, 2025, when three thinly traded coins showed prices far below $1 at once and were back near $1 a day or two later.
Worked example: How the filter treats two made-up price runs
- Run A: $0.999, then $0.48, then $0.998. The middle reading is 52% below $1, stands alone and has calm neighbors 0.1% and 0.2% from $1, so it is dropped.
- Run B: $0.95, then $0.85, then $0.95. The middle reading is 15% below $1, but both neighbors are 5% below, too far from $1 to count as calm, so it stays in as a real slide. Three or more days in a row beyond 10% would also stay in.
The filter can only make the record look milder: a genuine crash that fully reversed within a day or two would be dropped too. No USDT, USDC or DAI reading was removed.
How often did each stablecoin trade away from $1?
USDT, USDC and DAI were more than 0.5% from $1 on 2.3% to 2.5% of days, while BUSD, FRAX, LUSD and UST were off on 15.4% to 62.8%.
| Stablecoin | Data from | Days | Days off by more than 0.5% | Days off by more than 2% | Largest deviation | Date | Longest run beyond 0.5% |
|---|---|---|---|---|---|---|---|
| USDT (Tether) | Dec 30, 2020 | 2,127 | 48.00 | 0.00 | 1.1% | Mar 1, 2021 | 5 days |
| USDC (USD Coin) | Dec 30, 2020 | 2,127 | 53.00 | 1.00 | -3.9% | Mar 12, 2023 | 3 days |
| USDS (Sky USDS) | Sep 24, 2024 | 765 | 29.00 | 8.00 | -4.0% | Oct 1, 2024 | 9 days |
| DAI (Dai) | Dec 30, 2020 | 2,127 | 48.00 | 1.00 | -3.8% | Mar 12, 2023 | 5 days |
| USDe (Ethena USDe) | Dec 15, 2023 | 1,048 | 18.00 | 2.00 | 2.5% | Dec 25, 2023 | 2 days |
| FDUSD (First Digital USD) | Jul 29, 2023 | 1,187 | 39.00 | 1.00 | 2.3% | Aug 8, 2024 | 2 days |
| PYUSD (PayPal USD) | Aug 17, 2023 | 1,163 | 8.00 | 0.00 | 1.5% | Aug 28, 2023 | 2 days |
| RLUSD (Ripple USD) | Apr 2, 2025 | 578 | 0.00 | 0.00 | 0.3% | Nov 5, 2025 | 0 days |
| USD1 (World Liberty Financial USD) | Apr 28, 2025 | 552 | 3.00 | 0.00 | -0.9% | May 18, 2025 | 1 days |
| TUSD (TrueUSD) | Dec 30, 2020 | 2,127 | 131 | 18.00 | -3.6% | Feb 29, 2024 | 45 days |
| USDP (Pax Dollar) | Dec 30, 2020 | 2,127 | 74.00 | 1.00 | -2.3% | Jan 5, 2021 | 4 days |
| BUSD (Binance USD) | Dec 30, 2020 | 2,120 | 326 | 62.00 | -12.9% | Dec 8, 2024 | 15 days |
| GUSD (Gemini Dollar) | Dec 30, 2020 | 2,126 | 166 | 18.00 | -3.7% | Feb 26, 2021 | 11 days |
| FRAX (Frax) | Dec 30, 2020 | 2,126 | 560 | 7.00 | -4.0% | Mar 12, 2023 | 112 days |
| LUSD (Liquity USD) | Apr 5, 2021 | 2,030 | 832 | 225 | 6.8% | Apr 5, 2021 | 229 days |
| UST (TerraUSD) | Dec 30, 2020 | 959 | 602 | 478 | -99.4% | Jun 19, 2022 | 467 days |
The three big, long-running coins behaved alike. USDT's largest deviation in almost six years was a premium, 1.14% above $1 on March 1, 2021. USDC and DAI each had one day beyond 2%, the same one, March 12, 2023, at 3.89% and 3.75% below $1, and neither stayed off for more than 5 days in a row.
Newer coins mostly wobbled while new. All 8 of USDS's days beyond 2% came between September 28 and November 17, 2024, its first two months, peaking at 4.01% below $1 on October 1. Sky, formerly MakerDAO, began issuing USDS on September 18, 2024 as an upgraded DAI, with a converter that swaps the two one for one at no fee, and DAI had no day beyond 2% that autumn, so thin early trading is the likelier explanation. USDe's only two days beyond 2% were premiums in its first month of data. RLUSD, on exchanges since December 17, 2024, had no day more than 0.5% off in 578 days from April 2, 2025, though its record starts more than three months after launch, so its first weeks are missing.
| Stablecoin | Share of days | Most recent day |
|---|---|---|
| USDT (Tether) | 2.3% | Aug 8, 2024 |
| USDC (USD Coin) | 2.5% | Nov 11, 2024 |
| USDS (Sky USDS) | 3.8% | Feb 2, 2025 |
| DAI (Dai) | 2.3% | May 8, 2024 |
| USDe (Ethena USDe) | 1.7% | Feb 1, 2026 |
| FDUSD (First Digital USD) | 3.3% | Nov 19, 2025 |
| PYUSD (PayPal USD) | 0.7% | Aug 8, 2024 |
| RLUSD (Ripple USD) | 0.0% | n/a |
| USD1 (World Liberty Financial USD) | 0.5% | May 20, 2025 |
| TUSD (TrueUSD) | 6.2% | Aug 1, 2026 |
| USDP (Pax Dollar) | 3.5% | Oct 14, 2026 |
| BUSD (Binance USD) | 15.4% | Oct 29, 2026 |
| GUSD (Gemini Dollar) | 7.8% | Oct 29, 2026 |
| FRAX (Frax) | 26.3% | Nov 1, 2026 |
| LUSD (Liquity USD) | 41.0% | Nov 1, 2026 |
| UST (TerraUSD) | 62.8% | Aug 17, 2023 |
The share column puts records of different lengths on one scale. USDS's 29 days sound better than USDC's 53, but they come from 765 readings rather than 2,127, so its share is 3.8% against 2.5%. The last column dates each coin's most recent day more than 0.5% off: August 8, 2024 for USDT, November 11, 2024 for USDC and May 8, 2024 for DAI, while FRAX and LUSD were still off on November 1, 2026.
Why does one price a day understate a depeg?
A depeg can open and close between two midnight readings, so a daily series can miss the low entirely.
USDC in March 2023 shows how. Silicon Valley Bank went into receivership at 11:37 a.m. Eastern time on Friday, March 10, and around 10 p.m. Circle said $3.3 billion of USDC's reserves, about 8%, was stuck there, according to Federal Reserve researchers. DefiLlama's readings are stamped at midnight UTC. The one dated March 11 corresponds to 7 p.m. Friday in New York, three hours before Circle's announcement, and showed USDC close to $1. The one dated March 12, 7 p.m. Saturday, came after the coin had climbed off its low and read 3.89% below $1. The 86-cent trough fell in between, and even hourly data blurs it: the researchers' charts of hourly opening prices bottom out near 90 cents.
| Episode | Intraday low reported | Hourly figure | Furthest point in the daily data |
|---|---|---|---|
| UST, May 2021 | Below 92 cents (US Attorney, SDNY) | Not reported | 3.79% below $1, in a 5-day episode from May 21, 2021 |
| USDT, May 11, 2022 | $0.94 (FSOC) | About $0.97 to $0.98 (FSOC) | Never more than 2% from $1; 4 days beyond 0.5% in all of 2022 |
| USDC, March 11, 2023 | 86 cents (Federal Reserve) | About 90 cents (Federal Reserve) | 3.89% below $1, on March 12, 2023 |
| USDP, March 11 and 12, 2023 | About 91 cents (Federal Reserve) | Not reported | No day beyond 2% that year; its only one was January 5, 2021 |
| FDUSD, April 2, 2025 | 0.87 against USDT on Binance (CoinDesk) | Not reported | No day beyond 2% that year; its only one was a 2.3% premium on August 8, 2024 |
Crypto trades around the clock, but issuers redeem through banks, and Circle said its issuance and redemptions were constrained by US banking hours that weekend. Sharp dips also tend to be bought within hours by traders who expect redemption at $1. The count of days beyond 2% is therefore a floor on how often pegs broke, not an estimate.
Worked example: What the daily figure missed in March 2023
- Daily reading: $1 × (1 − 0.0389) = $0.9611, against a trough of $0.86, a gap of $0.1011, or about 10 cents a coin.
- 10,000 USDC sold at the trough: 10,000 × $0.86 = $8,600. At the daily reading's price: $9,611. Held until the coin was back at $1 on March 13: $10,000.
The daily series puts that holder's worst moment $389 below $10,000; at the trough it was $1,400.
Which depegs were the largest and longest?
TerraUSD's 2022 collapse dwarfs the rest: 466 days beyond 2%, against 12 days for the next-longest episode, TrueUSD's in February 2024.
| Stablecoin | Start | End | Days beyond 2% | Largest deviation |
|---|---|---|---|---|
| TUSD | Jan 5, 2021 | Jan 5, 2021 | 1.00 | -2.1% |
| USDP | Jan 5, 2021 | Jan 5, 2021 | 1.00 | -2.3% |
| UST | Jan 18, 2021 | Jan 18, 2021 | 1.00 | -2.3% |
| UST | Jan 31, 2021 | Feb 3, 2021 | 4.00 | 3.9% |
| UST | Feb 8, 2021 | Feb 9, 2021 | 2.00 | 2.3% |
| UST | May 21, 2021 | May 25, 2021 | 5.00 | -3.8% |
| UST | May 9, 2022 | Aug 17, 2023 | 466 | -99.4% |
| USDC | Mar 12, 2023 | Mar 12, 2023 | 1.00 | -3.9% |
| DAI | Mar 12, 2023 | Mar 12, 2023 | 1.00 | -3.8% |
| USDe | Dec 25, 2023 | Dec 25, 2023 | 1.00 | 2.5% |
| USDe | Jan 7, 2024 | Jan 7, 2024 | 1.00 | 2.3% |
| TUSD | Feb 17, 2024 | Feb 29, 2024 | 12.00 | -3.6% |
| TUSD | Mar 25, 2024 | Mar 27, 2024 | 3.00 | 2.6% |
| FDUSD | Aug 8, 2024 | Aug 8, 2024 | 1.00 | 2.3% |
| USDS | Sep 28, 2024 | Oct 3, 2024 | 5.00 | -4.0% |
| USDS | Oct 9, 2024 | Oct 9, 2024 | 1.00 | -3.8% |
| USDS | Oct 28, 2024 | Oct 28, 2024 | 1.00 | 2.6% |
| USDS | Nov 17, 2024 | Nov 17, 2024 | 1.00 | -2.0% |
| TUSD | Dec 8, 2024 | Dec 11, 2024 | 2.00 | 2.6% |
Eleven of the 19 episodes had just one day beyond 2% and six had two to five. In 8 the largest deviation was a premium rather than a discount, including both of USDe's and two of TrueUSD's. A premium usually means demand has outrun new issuance, and a buyer at $1.03 loses about 3% when the price returns to $1. USDC and DAI broke on the same day because DAI's peg leaned partly on a facility that took USDC one for one, a channel the Fed note traces and how stablecoins work explains.
TrueUSD's trouble ran longer: 131 days more than 0.5% off, a longest run of 45 days and a low of 3.56% below $1 on February 29, 2024. In September 2024 the SEC alleged that a substantial portion of TUSD's backing had been invested in a speculative offshore fund, and that 99% of the reserves were in it by that month. TrueCoin and TrustToken, the companies charged, settled without admitting or denying the allegations.
What happened to TerraUSD?
TerraUSD lost its peg for good in May 2022: its final break began on May 9, and its lowest reading, on June 19, was 99.36% below $1.
UST had no pool of dollars behind it. Terraform Labs promised that one UST could always be swapped for $1 of its sister token, LUNA, so the peg held only while LUNA kept its value. Cracks showed in 2021, when UST went more than 2% from $1 four times in five months, at worst 3.79% below in May. Federal prosecutors later said it had fallen below 92 cents then, and that Do Kwon, Terraform's co-founder, falsely credited the algorithm with the recovery when a trading firm had bought UST to prop up the peg.
From May 9, 2022, UST never again read within 2% of $1, for 466 days until DefiLlama's series ended on August 17, 2023; in all it spent 602 of its 959 days more than 0.5% off. The SEC's complaint says UST and its sister tokens fell close to zero in May 2022, and prosecutors put investor losses above $40 billion. Kwon pleaded guilty in August 2025 and was sentenced on December 11, 2025 to 15 years in prison. The bitcoin drawdowns study places the crash inside bitcoin's 2021 to 2022 fall.
Worked example: What 99.36% below $1 means
- Price at the lowest reading: $1 × (1 − 0.9936) = $0.0064.
- 10,000 UST bought at $1 would have been worth 10,000 × $0.0064 = $64.
No other coin had a daily reading more than 13% from $1; the next largest gap was BUSD's 12.91%.
How did BUSD's wind-down show up in the data?
As a slow drift, not a crash: BUSD was more than 0.5% from $1 on 326 of its 2,120 days, at worst 12.91% below on December 8, 2024.
- February 13, 2023. New York's financial regulator ordered Paxos, BUSD's issuer, to stop minting it. Paxos stopped issuing new BUSD on February 21 and said existing tokens would stay redeemable through at least February 2024.
- November 29, 2023. Binance, whose brand the coin carried, said it would end BUSD support on December 15 and from January 2, 2024 convert remaining balances on its platform to FDUSD.
With its main venue gone, few traders had reason to hold BUSD at exactly $1. It logged 62 days beyond 2% over its record and was still more than 0.5% off on October 29, 2026. Drift in a coin whose market is shutting down reflects vanishing buyers more than missing reserves, and the regulator said Paxos remained required to redeem BUSD at $1 for customers in good standing. That is why BUSD's episodes from 2024 on are left out of the episode list.
Why do thinly traded coins look less stable?
Because a few small trades can push their price well away from $1 even when nothing in their backing has changed.
- LUSD was more than 0.5% off on 41% of its days, with runs of up to 229 days, and its largest deviation was a 6.78% premium on its first day of data, April 5, 2021. Liquity designed it to float: redemption for $1 of ether supports a floor, and the 110% minimum collateral ratio caps it near $1.10.
- FRAX was off on 26.3% of days, with a 112-day run, and was still off on November 1, 2026. By April 2025 Frax called the coin the Legacy Frax Dollar, and its proposal FIP-430, passed by May 2, 2025, ended the guaranteed one-for-one swap into the newer frxUSD, which this study does not include.
- GUSD was off on 7.8% of days, at worst 3.71% below $1 on February 26, 2021. The Fed note says it fell to about 96 cents over the March 2023 weekend as traders swapped discounted DAI for it.
Are the biggest stablecoins holding their peg more tightly?
In this data, yes: USDT, USDC and DAI logged 33 to 45 days more than 0.5% off in 2021 and none in 2025 or 2026 through November 1.
| Year | USDT | USDC | DAI |
|---|---|---|---|
| 2020 | 0.00 | 0.00 | 0.00 |
| 2021 | 33.00 | 40.00 | 45.00 |
| 2022 | 4.00 | 5.00 | 0.00 |
| 2023 | 10.00 | 6.00 | 2.00 |
| 2024 | 1.00 | 2.00 | 1.00 |
| 2025 | 0.00 | 0.00 | 0.00 |
| 2026 | 0.00 | 0.00 | 0.00 |
Most of each total came in 2021: 33 of USDT's 48 days, 40 of USDC's 53 and 45 of DAI's 48. Counts fell to between 0 and 5 in 2022, rose to 10 for USDT in 2023, the year of the Silicon Valley Bank weekend, and reached 0 by 2025. One caution: none of the three coins' 2021 readings is exactly $1.00, but from 2022 to 2025 the feed often records exactly $1.00, including a flat USDT run from May 9 to 14, 2022, the week the FSOC says USDT touched $0.94. Part of the decline may therefore come from the feed. In 2026 the readings vary from day to day again, and none of the three crossed 0.5%.
What can't daily price data tell you?
It cannot show intraday lows, prices on any one exchange, whether redemptions were open or whether a coin's reserves were sound.
- Intraday extremes. Days beyond 2% are a minimum count.
- One price source. DefiLlama gives one price per coin, while FDUSD's April 2025 low was quoted on Binance, its main market.
- Flat stretches. Readings of exactly $1.00 for days at a time can hide small or brief moves.
- Different start dates. Records begin when DefiLlama starts pricing a coin, so launch weeks can be missing.
- A softening filter. Dropping isolated readings more than 10% off would also drop a genuine crash that reversed within two days.
- Price is not redemption. A coin can trade near $1 while redemptions are limited, or below $1 while fully backed, as USDC did.
- Selection and weighting. The 16 coins are today's large coins plus notable older ones, coins that failed before December 30, 2020 are missing, and a day off for a small coin counts as much as one for USDT.
- Thresholds are choices. A 1% or 5% line would give different counts.
What does this history mean for someone holding stablecoins?
It shows how often pegs slipped and how fast they came back, not what will happen next: past performance never predicts future results.
The record has three patterns. Breaks in the big fiat-backed coins were short: USDT was back just under $1 within a day of its May 2022 low, by the FSOC's account, and USDC returned to $1 when redemptions resumed two days after its trough. The algorithmic UST never recovered. And BUSD drifted for months as its market dried up, while its issuer still redeemed at $1.
What decides how a holder fares in a break is something the data cannot show: who may redeem and how fast, what the reserve report says and where the coin can be sold. USDC vs USDT compares the two largest issuers on those points, and top stablecoins tracks supply and prices. The GENIUS Act sets a reserve rule for permitted US issuers: at least 1:1 reserves in cash or Federal Reserve balances, deposits at insured banks and credit unions, Treasuries with 93 days or less to maturity, overnight repos and reverse repos backed by Treasuries, government money market funds that hold only those assets, other liquid federal government assets a regulator approves, and tokenized versions of the non-repo assets. The GENIUS Act guide covers when it applies. Rewards for holding stablecoins carry risks of their own, covered in stablecoin yield vs bank interest, and a break mid-transfer changes what arrives, as sending money with stablecoins notes. Stablecoins are not FDIC insured, and the crypto traded against them is volatile and can lose most or all of its value.
How to cite or reuse the depeg data
The data is free to reuse under CC BY 4.0, with credit and a link to this page.
- Get the CSV file, one row per coin as in the main table; the other tables can be quoted with the same credit.
- Quote figures with their window, for example: USDC was more than 0.5% from $1 on 53 of 2,127 days from December 30, 2020 to November 1, 2026.
- Note that the figures come from one price a day, so intraday lows were deeper.
- Suggested credit: Source: Sonic Wallet analysis of DefiLlama stablecoin prices, data through November 1, 2026.
The editorial policy explains how the site's studies are sourced, checked and corrected.