Anyone buying Bitcoin is told to expect crashes. But how often do they come? And has that changed since ETFs and companies started buying?
Both can be counted, on every daily close since July 2011.
What counts as a 30 percent crash here
Every day counts on which Bitcoin closed 30 percent or more below its highest close of the previous 30 days. When such days follow closely on each other, they count as one crash. A new one only begins after 30 days without any such day.
This is not the only definition. Measured from the all-time high, Bitcoin is in the middle of a 30 percent decline right now: on 4 October the price was 30.6 percent below the October 2025 record. Under the monthly definition, the last crash was 234 days ago. Both numbers are correct. They answer different questions.
Anyone counting crashes first has to say: crashes from what.
This piece is about the fast crashes. How often did they come?
19 crashes since 2011, and they are getting rarer
By this definition, Bitcoin fell 30 percent within a month 19 times since July 2011. Across time:
| Period | Crashes | Per year | Volatility (median) |
|---|---|---|---|
| July 2011 to 2014 | 8 | 2.3 | 58.7 % |
| 2015 to 2017 | 4 | 1.3 | 46.7 % |
| 2018 to 2020 | 3 | 1.0 | 52.3 % |
| 2021 to 2023 | 3 | 1.0 | 50.3 % |
| 2024 to Sep 2026 | 1 | 0.4 | 36.2 % |
Volatility here is the realised swing of daily prices over 30 days, annualised. It has fallen from almost 59 to 36 percent. The crashes became rarer along with it, from more than two a year to one in almost three years.
The gaps between crashes show the same. Before 2018 the median gap was 162 days, afterwards 248. The exact figure depends somewhat on how crashes are separated from each other; the direction does not: with 60 or 90 days of separation, about 220 to 240 days become about 380.
The latest pause is the clearest. Between May 2022 and February 2026 there were 1,365 days without such a crash, longer than ever before. The longest pause before that ran from January 2015 to July 2017, 914 days.
The market has become calmer, and fast crashes have become rarer with it.
Does rarer also mean milder?
When a crash comes, it rarely stops at 30 percent
The trigger is a 30 percent fall. How deep it went in the following 60 days, measured from the high before the crash:
| Crashes | Low within the next 60 days (median) | |
|---|---|---|
| up to 2017 | 12 | −52.0 % |
| from 2018 | 7 | −46.7 % |
Since 2018 only one crash stayed close to the threshold: January 2022 at −31 percent. The deepest low was −54.3 percent (May 2022).
The big bear markets have also become shallower, measured from all-time high to bottom: −89.5 percent to January 2015, −83.2 percent to December 2018, −76.6 percent to November 2022. Since the October 2025 high, the deepest point so far was −53.0 percent, on 30 June 2026.
Crashes have become rarer. Anyone caught in one still loses almost half.
And do the crashes fit the calmer market?
In calm times, the same crash is the bigger shock
A 30 percent fall does not mean the same thing in every period. What matters is how much Bitcoin normally swung beforehand. Scaled to 30 days, volatility gives a typical monthly swing.
| Crash | Volatility before | Typical monthly swing | Crash in these units |
|---|---|---|---|
| December 2013 | 142 % | about 41 % | less than 1 times |
| November 2018 | 31 % | about 9 % | 4.1 times |
| March 2020 | 34 % | about 10 % | 3.7 times |
| February 2026 | 23 % | about 6.5 % | 5.5 times |
In 2013 a 30 percent month in Bitcoin was almost an ordinary month. In February 2026 it was five and a half times the usual swing, the largest move of this kind among all 18 crashes for which the prior volatility is available. From its high of $96,952 on 14 January, the price fell to $62,910 by 5 February, a drop of 35.1 percent.
That is the other side of calmer times. Anyone sizing a position by the swings of recent weeks expects small declines in calm phases. February 2026 came out of exactly such a phase.
Low volatility makes crashes rarer, not impossible, and it makes them more of a surprise.
The obvious objection
"After 234 days without a crash, the next one must be due."
The data offers no support for that. The pauses ranged from 60 to 1,365 days, with no visible pattern for when one ends. A counter of days since the last crash describes the past. It says nothing about the next crash. How long it has been calm does not tell you how long it will stay calm.
What to do with this
Three lessons hold up:
- Plan for the crash, not its date. Since 2018 there have been seven, and one since 2024. When the next one comes cannot be read from the gaps.
- Plan for the depth that follows. Since 2018 the low of the following two months was a median −47 percent. Ask yourself whether your position can take that without forcing you to sell.
- Distrust the calm. Low volatility lowers the frequency, not the size. The biggest shock in the series came out of the calmest period.
Rarer does not mean rare enough to size your position around it.
What remains open is whether the pauses keep getting longer. That will only show when the next crash comes or fails to, and two long pauses do not yet make a trend.
FAQ
Why 30 days and not the distance from the all-time high? The distance from the all-time high measures how long a bear market lasts and reports the same state for months. The 30-day definition measures fast crashes, the kind that hit a leveraged or oversized position within a short time.
How are the crashes counted? A crash begins on the first day the close is 30 percent or more below the highest close of the previous 30 days. It only ends once 30 days pass without such a day. Daily closes from 1 July 2011, from a public price series until August 2017 and from Binance after that.
How is volatility calculated? The standard deviation of daily log returns over 30 days, annualised. The typical monthly swing is that value times the square root of 30/365. On 4 October 2026 volatility stood at 30.9 percent, lower than on 81 percent of all days since 2011 and about the median of the last two years.
Does this apply to altcoins? Not with the same threshold. An altcoin that swings three times as much as Bitcoin reaches 30 percent far more often. There the threshold mostly measures volatility.
Not investment advice, not a recommendation, not a forecast — historical patterns are no guarantee.
Study the Past — Improve your Future 🥋