Key takeaways
- Bitcoin's deepest fall on Coinbase daily prices was 83.8%, between 16 December 2017 and 15 December 2018. The Nasdaq Composite's deepest since 1971 was 77.9%.
- Bitcoin recorded 17 falls of 20% or more in the 11.8 years to 2 September 2026, about 14.5 a decade. The Nasdaq Composite averaged 2.3 a decade.
- Bitcoin's median fall of 20% or more was back to its old high in 86 days. The Nasdaq Composite's median took 524 days, and its dot-com fall took 5,522.
- A 5% bitcoin sleeve moved a Nasdaq Composite portfolio's worst drawdown from 36.4% to 38.6% since December 2014. A 10% sleeve moved it to 41.1%.
- Bitcoin spent 65.3% of its days at least 20% below a previous peak, and 31.6% of its days at least 50% below one.
Crypto drawdowns run deeper and arrive far more often, but the median one ends sooner
You'd like to know whether crypto drawdowns are equity drawdowns with the volume turned up, or something different in kind. On the daily price series, they're different in frequency, close in depth, and shorter in the middle of the distribution.
Bitcoin's worst fall on Coinbase's daily closes was 83.8%, from 19,650.01 dollars on 16 December 2017 to 3,183.00 dollars on 15 December 2018. The Nasdaq Composite's worst since February 1971 was 77.9%, from 5,048.62 on 10 March 2000 to 1,114.11 on 9 October 2002. That's six percentage points apart. It's a lot closer than the reputation gap suggests.
The separation shows up in how often, not how far. Bitcoin produced 17 separate falls of 20% or more in 11.8 years of data, roughly 14.5 a decade. The Nasdaq Composite produced 13 in 55.6 years, about 2.3 a decade. Then it reverses on duration. Bitcoin's median 20% fall was fully recovered 86 days after the peak. The Nasdaq Composite's median took 524.
Depth: 94.0% for ether, 83.8% for bitcoin, 82.9% for the Nasdaq-100
Here's the maximum drawdown on each daily series, measured from a running peak in US dollars. Every figure below runs to the close of 2 September 2026.
- Ether, 94.0%. Peak 1,386.02 dollars on 13 January 2018, trough 83.00 dollars on 14 December 2018.
- Bitcoin, 83.8%. Peak 19,650.01 on 16 December 2017, trough 3,183.00 on 15 December 2018.
- Nasdaq-100, 82.9%. Peak 4,704.73 on 27 March 2000, trough 804.64 on 7 October 2002.
- Nasdaq Composite, 77.9%. Peak 5,048.62 on 10 March 2000, trough 1,114.11 on 9 October 2002.
- Nasdaq Composite, 59.9% in the 1973 to 1974 bear market. Peak 136.84 on 11 January 1973, trough 54.87 on 3 October 1974.
- Dow Jones Industrial Average, 37.1%. Peak 29,551.42 on 12 February 2020, trough 18,591.93 on 23 March 2020.
- S&P 500, 33.9%. Peak 3,386.15 on 19 February 2020, trough 2,237.40 on 23 March 2020.
The last two rows are shallow for a reason, and it isn't that large-cap equities are safe. The Federal Reserve Bank of St Louis publishes only 10 years of daily history for Standard and Poor's and Dow Jones series under its licence. Those two lines start in September 2016, so they can't reach 2008 or 2000 at all.
The comparison that carries weight is bitcoin against the Nasdaq-100, because both are concentrated and high-beta. 83.8% against 82.9%. On depth alone, holding bitcoin through 2018 was about as painful as holding the Nasdaq-100 through the dot-com bust. If you've read our piece on volatility vs drawdown, this is the metric that tracks what a holder actually feels, and on that metric the two assets are close.
Frequency is the gap that doesn't close
Depth is one draw from a distribution. How often you're asked to sit through one is the part that compounds into a decision.
Over the whole of each series, bitcoin averaged 14.5 falls of 20% or more per decade, ether 9.7, the Nasdaq Composite 2.3 and the Nasdaq-100 2.0. Comparing series of different lengths is unfair, so here's the same window for both. Between 1 December 2014 and 2 September 2026, bitcoin had 17 falls of 20% or more and the Nasdaq Composite had 4.
Time spent underwater says the same thing from another angle. Bitcoin closed at least 20% below a previous peak on 65.3% of its trading days, and at least 50% below one on 31.6% of them. Ether was 20% underwater on 85.3% of days and 50% underwater on 53.8%. The Nasdaq Composite's figures since 1971 are 39.3% and 17.4%.
So the modal state of a crypto holding, over these samples, is "below where it was". That's not a claim about the next decade. It's a description of what the bitcoin drawdown history contains.
Duration: the long tail belongs to equities, not to crypto
This is where the received wisdom breaks. The median drawdown recovery time is much shorter for crypto.
Bitcoin's median 20% fall took 35 days to reach its trough and 86 days from peak to full recovery. The Nasdaq Composite's median took 117 days down and 524 days to recover. Restrict both to the same December 2014 to September 2026 window and bitcoin's median recovery is 86 days against the Nasdaq Composite's 215.
The tail is another matter, and it runs the other way. The Nasdaq Composite needed 5,522 days, just over 15 years, to erase its March 2000 peak, finally doing it on 23 April 2015. The Nasdaq-100 took 5,699 days, recovering on 3 November 2015. Bitcoin's longest completed recovery was 1,080 days, from 16 December 2017 to 30 November 2020. Ether's was 1,382 days, from 8 November 2021 to 21 August 2025.
Two crypto drawdowns were still open at the close of 2 September 2026. Bitcoin peaked at 124,720.09 dollars on 5 October 2025, bottomed 53.0% lower at 58,585.96 on 30 June 2026, and was still 38.0% below that peak at 77,330.23. Ether peaked at 4,831.24 on 21 August 2025, fell 67.6% to 1,566.40 on 28 June 2026, and was 50.4% below the peak at 2,394.40. Neither has a recovery date, and neither is in the median above.
The five worst episodes on each side, with their timings
Depth without duration is half the picture. These are the largest falls on the crypto series and the equity series, with days from peak to trough and days from peak to recovery.
- Ether, 2018. 94.0% down, 335 days to the trough, 1,107 days to recovery on 24 January 2021.
- Bitcoin, 2017 to 2018. 83.8% down, 364 days to the trough, 1,080 days to recovery on 30 November 2020.
- Ether, 2021 to 2025. 79.3% down, 222 days to the trough, 1,382 days to recovery on 21 August 2025.
- Bitcoin, 2021 to 2024. 76.7% down, 378 days to the trough, 847 days to recovery on 4 March 2024.
- Bitcoin, 2025 to date. 53.0% down at the 30 June 2026 trough, 268 days from peak to trough, not recovered.
- Nasdaq Composite, 2000 to 2015. 77.9% down, 943 days to the trough, 5,522 days to recovery.
- Nasdaq Composite, 1973 to 1978. 59.9% down, 630 days to the trough, 2,065 days to recovery on 7 September 1978.
- Nasdaq Composite, 2021 to 2024. 36.4% down, 404 days to the trough, 832 days to recovery on 29 February 2024.
- S&P 500, 2022 to 2024. 25.4% down, 282 days to the trough, 746 days to recovery on 19 January 2024.
- Nasdaq Composite, 2020. 30.1% down, 33 days to the trough, 110 days to recovery on 8 June 2020.
Look at the bitcoin and Nasdaq Composite rows for 2021 to 2024 together. They peaked 11 days apart and troughed within five weeks of each other. Crypto's calendar and the equity calendar are not always independent, which matters for the next section. Our note on bear market recovery time works the equity side of that in more detail.
What crypto sleeve size did to a portfolio's worst loss
A drawdown quoted on an asset held at 100% weight isn't the number anyone experiences. So here's the arithmetic that is. Take a daily-rebalanced blend of bitcoin and the Nasdaq Composite over the 2,936 days both traded between 1 December 2014 and 2 September 2026, and measure the blend's own maximum drawdown.
- 0% bitcoin: worst drawdown 36.4%, return 15.69% a year.
- 1% bitcoin: 36.8%, and 16.32% a year.
- 2% bitcoin: 37.2%, and 16.96% a year.
- 5% bitcoin: 38.6%, and 18.84% a year.
- 10% bitcoin: 41.1%, and 21.93% a year.
- 20% bitcoin: 45.9%, and 27.89% a year.
- 100% bitcoin: 83.2%, and 57.54% a year.
The chart above plots the drawdown column. Two things sit inside it. First, the drawdown cost of a small sleeve is sublinear in the early rows: five percentage points of bitcoin bought 2.2 points of extra drawdown, not the 4.2 points that 5% of an 83.2% fall would naively suggest. Second, and less comfortable, every blend from 0% to 20% recorded its worst moment in the same episode, the November 2021 to 28 December 2022 slide. Bitcoin's own worst year barely shows at the portfolio level, because the Nasdaq Composite closed on 14 December 2018 within 0.4% of where it closed a year earlier, at 6,910.66 against 6,936.58.
That's the finding that survives: the sleeve didn't create a new worst day, it deepened the existing one. A separate piece on bitcoin allocation at 1-5% runs the same question against a 60/40 rather than an all-equity base, and the shape is similar.
The counter-case: 11.8 years is a sample, not a distribution
The strongest objection to everything above is that the crypto series is too short to have a tail yet, and too flattering while it lasts.
Bitcoin compounded at 57.54% a year over this window. No asset sustains that, and the drawdown statistics come from exactly the same 11.8 years as the return. An asset in a secular bull market recovers quickly by construction, because the trend keeps carrying it back to new highs.
The median-recovery advantage is also partly an artefact of counting. An asset that falls 20% roughly every eight months generates lots of short, shallow episodes, and those crowd the median down. An index with 13 episodes in 55 years has a median made almost entirely of genuine bear markets. Comparing medians across such different event counts flatters the noisier series.
And the equity tail has already happened, while crypto's may not have. Fifteen years to recover the dot-com peak is a real, observed outcome. Nothing in the bitcoin record rules out a 15-year recovery. The record simply isn't 15 years long on this series. If bitcoin's open drawdown, running since 5 October 2025, takes as long as the 2017 one did, it recovers around September 2028. That's an arithmetic extension of a past episode, not a projection.
What these price series can't tell you
Start with what's being measured. The St Louis Fed notes that its S&P 500 series "is a price index and not a total return index", and the same is true of the Nasdaq lines. Equity drawdowns computed on price alone are deeper than a holder who reinvested dividends experienced. Bitcoin and ether pay nothing, so no such correction applies to them. That bias runs against equities, and it isn't small over multi-year recoveries.
The crypto data is one venue. Coinbase prices are struck, in the St Louis Fed's words, "as of 5 PM PST", and the bitcoin series begins on 1 December 2014. Anything the market did before that date is outside this comparison. Crypto also trades seven days a week against an index's five, so it gets more chances to print a new low, which inflates both event counts and underwater day counts a little.
Everything here is in US dollars. A sterling holder's drawdown differs by whatever the exchange rate did over the same window, and in 2022 that was a large number. The maximum drawdown statistic itself is also a single worst path, not a probability. It says nothing about how likely a repeat is. The loss recovery arithmetic that follows from these depths is unforgiving in a way the percentages hide: a 94% fall needs a 1,567% gain to get back.
The UK regulatory position also sits outside the price series entirely. Since 8 October 2023, all cryptoasset firms marketing to UK consumers, including firms based overseas, have had to comply with the UK financial promotions regime. The Financial Conduct Authority's own consumer guidance says anyone deciding to invest in crypto ought to be "prepared to lose all the money you have invested", and that holders are unlikely to have access to the Financial Services Compensation Scheme or the Financial Ombudsman Service if something goes wrong. The Bank of England's Financial Stability Report of 2 December 2025 recorded that its Financial Policy Committee "is continuing to monitor developments in the unbacked crypto-asset sector and its interconnectedness with the financial sector and the real economy". A drawdown table measures price. It doesn't measure the chance of an exchange failing with your coins on it, and that risk has no equity-index equivalent.
One practical note, since the blend arithmetic above is the part that transfers: the number that matters for a portfolio is the blend's drawdown, not the asset's. LedgerTouch computes that continuously; a spreadsheet computes it once a quarter. Either gets you the same figure.
What would change the conclusion
If the open bitcoin drawdown runs past 1,080 days, which would take it beyond 19 September 2028, the "crypto recovers faster" half of this piece stops being true even in the median. That single episode would move both the median and the tail.
If crypto's falls stop arriving on their own schedule, the sleeve arithmetic changes shape. Every blend above took its worst loss inside the 2021 to 2022 equity bear market, which is the case where the two calendars did line up. A repeat with bitcoin 80% down at the same time as a 40% equity fall produces a portfolio number that none of these rows contain.
If the series gets long enough to have a real tail, the frequency gap could narrow on its own. Bitcoin at 25 years of daily data might show the same 2 to 3 events per decade the equity indices do, with the extra events of its first decade looking like the volatility of a young market rather than a permanent feature.
The number worth watching isn't the depth. Depth has already converged: 83.8% against 82.9% on the two most comparable series. It's whether crypto keeps taking its losses on a separate clock from equities, because that, not the size of the fall, is what decided how much a sleeve added to the drawdown of everything else you own.