How often does Bitcoin drop 10% in a week?
Bitcoin closed 10% or more below its close a week earlier on 6.79% of Binance days and 8.41% of Coinbase days. Each year, 20% and 30% drops, other coins.
Bitcoin's daily close was 10% or more below its close a week earlier on 6.79% of days on Binance. That is 176 of 2,592 days for the BTCUSDT perp, from September 2019 to October 19, 2026. On Coinbase, BTC-USD did it on 8.41% of days (345 of 4,103, from July 2015). These days arrive in bunches. They form 46 separate episodes on Binance and 77 on Coinbase. The share changed hugely by year, from 24.66% of Coinbase days in 2018 to 1.92% in 2023.
These are past cases, not a forecast.
For what Bitcoin did in the 30 days after these drops, see Bitcoin after a 10% weekly drop.
What do we mean by a 10% drop in a week?
We compare each day's closing price with the close 7 days earlier. If today's close is 10% or more lower, that day counts. The close is the last price of a UTC day, midnight to midnight on the world clock.
This is a net change over seven days. It is not the biggest dip inside the week. A week that fell 15% and then recovered to 5% down would not count.
A perp is a futures contract that never expires. Binance's BTCUSDT is a perp priced in USDT, a token built to track the dollar. Coinbase's BTC-USD is spot: the coin itself, priced in US dollars. We never mix the two.
An episode is the first day of a drop after at least 7 days with no such drop. One crash can make many counted days in a row. Episodes show how many separate times it happened.
How often do weekly drops of 10%, 20% and 30% happen?
The 7-day rows count 4,103 days on Coinbase and 2,592 on Binance. The 30-day rows count 4,080 and 2,569.
| Move over the window | Coinbase: share of days | Coinbase: episodes | Binance: share of days | Binance: episodes |
|---|---|---|---|---|
| 7-day drop of 10% or more | 8.41% | 77 | 6.79% | 46 |
| 7-day drop of 20% or more | 1.46% | 23 | 1.27% | 10 |
| 7-day drop of 30% or more | 0.29% | 4 | 0.35% | 2 |
| 7-day rise of 10% or more | 14.23% | 92 | 11.30% | 57 |
| 7-day rise of 20% or more | 3.80% | 38 | 2.51% | 19 |
| 7-day rise of 30% or more | 0.83% | 19 | 0.35% | 6 |
| 30-day drop of 20% or more | 7.84% | 22 | 7.90% | 12 |
| 30-day drop of 30% or more | 2.57% | 8 | 1.91% | 5 |
| 30-day drop of 50% or more | 0.10% | 2 | 0.04% | 1 |
| 30-day rise of 20% or more | 22.03% | 28 | 18.84% | 16 |
Weekly rises of 10% were more common than weekly drops: 14.23% of Coinbase days against 8.41%. A 30-day rise of 20% happened on 22.03% of Coinbase days, nearly three times the share for a 30-day drop of 20%.
A 30% weekly drop is rare. Binance had 9 such days: 7 in 2020 and 2 in 2022.
Did the share change from year to year?
Yes, a great deal. This is each year's count of days with a weekly drop of 10% or more. Coinbase starts July 20, 2015, Binance September 8, 2019. Both end October 19, 2026.
| Year | Coinbase BTC-USD | Binance BTCUSDT |
|---|---|---|
| 2015 | 13 of 158 (8.23%) | n/a |
| 2016 | 14 of 366 (3.83%) | n/a |
| 2017 | 35 of 365 (9.59%) | n/a |
| 2018 | 90 of 365 (24.66%) | n/a |
| 2019 | 29 of 365 (7.95%) | 13 of 108 (12.04%) |
| 2020 | 18 of 366 (4.92%) | 18 of 366 (4.92%) |
| 2021 | 52 of 365 (14.25%) | 53 of 365 (14.52%) |
| 2022 | 45 of 365 (12.33%) | 45 of 365 (12.33%) |
| 2023 | 7 of 365 (1.92%) | 7 of 365 (1.92%) |
| 2024 | 12 of 366 (3.28%) | 12 of 366 (3.28%) |
| 2025 | 14 of 365 (3.84%) | 12 of 365 (3.29%) |
| 2026 | 16 of 292 (5.48%) | 16 of 292 (5.48%) |
In 2018, about a quarter of Coinbase days (24.66%) sat 10% or more below the close a week before. In 2023 it was 1.92%. So the all-years 8.41% describes almost no single year well. Still, every year on Coinbase had at least 7 such days. A weekly drop of 20% or more did not happen at all on Coinbase in 2023, 2024 or 2025. Big weekly rises also became rarer: a 10% weekly rise came on 36.71% of Coinbase days in 2017 and 3.84% in 2025.
Do other coins fall 10% in a week more often?
Yes. Each coin uses its own history, so start dates differ and the comparison is not perfectly even.
| Market | Share of days with a 7-day drop of 10%+ | Episodes |
|---|---|---|
| Coinbase BTC-USD | 8.41% | 77 |
| Binance BTCUSDT | 6.79% | 46 |
| Coinbase ETH-USD | 13.25% | 95 |
| Binance ETHUSDT | 10.83% | 55 |
| Binance SOLUSDT | 17.43% | 67 |
| Binance XRPUSDT | 11.57% | 64 |
| Binance DOGEUSDT | 16.49% | 72 |
| Binance ADAUSDT | 15.82% | 77 |
| Binance AVAXUSDT | 20.40% | 72 |
| Binance LINKUSDT | 16.21% | 76 |
| Binance BNBUSDT | 7.67% | 45 |
| Binance TRXUSDT | 6.66% | 37 |
TRX (6.66%) and Bitcoin on Binance (6.79%) had the fewest weekly drops. Avalanche had the most, at 20.40% of its days. These coins fell together in big sell-offs, so the rows are not 12 separate tests.
What happened in the 30 days after a weekly drop of 10%?
We looked at the first day of each episode, then the price 30 days later. This table compares those days with all days. "Spaced" means only episodes at least 90 days apart.
| Market and group | Cases | Higher 30 days later | Median 30-day move |
|---|---|---|---|
| Coinbase BTC-USD, all days | 4,073 | 58.04% | +3.26% |
| Coinbase BTC-USD, after a drop | 77 | 57.14% | +2.10% |
| Coinbase BTC-USD, after a drop, spaced | 29 | 58.62% | +2.05% |
| Binance BTCUSDT, all days | 2,562 | 55.82% | +2.04% |
| Binance BTCUSDT, after a drop | 46 | 50.00% | −0.11% |
| Binance BTCUSDT, after a drop, spaced | 18 | 55.56% | +1.05% |
The past cases do not show a difference we can trust. Coinbase after a drop looks like Coinbase on any day. Binance after a drop looks lower, but its 18 spaced episodes give 55.56%, close to the 55.82% for all days.
We also looked at the lowest close in the next 30 days. On Coinbase its median was 7.5% below the drop day's close, against 5.8% after all days (9.2% for the 29 spaced episodes). On Binance it was 6.4% against 6.3%. These are small samples, and a deeper dip does not say where the 30 days ended.
We look at these cases in more detail in what happened after Bitcoin fell 10% in a week. For how long Bitcoin took to get back after big falls, see crypto drawdown recovery times.
What does this data not show?
It does not show what happened inside the week. For how often price touched a level below at any point, see how wide a stop loss should be.
It is not a forecast. A busy year like 2018 did not mean a busy year followed.
The coins are a short list of big survivors. We looked at 12 markets and many cut-offs, so one odd-looking cell could be chance. For the size of an ordinary day, see how much Bitcoin moves in a day.
How we measured this
We used public daily closing prices for Coinbase BTC-USD (July 20, 2015 to October 19, 2026) and Binance BTCUSDT (September 8, 2019 to October 19, 2026), plus nine more Binance perps and Coinbase ETH-USD for the coin table. For every day with a close 7 or 30 days earlier, we computed the percent change and counted days beyond each cut-off. Spaced episodes are at least 90 days apart. Forward 30-day moves start at the drop day's close, and windows that ran past October 19, 2026 were dropped. We tested no trading rule.
Try this in Comparand
Ask your connected AI agent: "Get daily bars for Coinbase BTC-USD over my chosen dates. Count the days where the close is 10% or more below the close 7 days earlier, and list the first day of each separate episode."
- venue_bars returns open, high, low and close prices plus volume for completed bars only. Use the 1d interval. It is free on every key.
- Your agent counts the drops. The tool only returns bars.
- Ask venue_coverage (also free) how far back each exchange goes before you compare with our counts. Our test used Coinbase from July 2015, and the chart search covers Coinbase from January 2017.
To see how earlier falling weeks continued, ask: "Find the 10 closest past daily charts for Binance BTCUSDT using the last 28 completed daily bars. Show each case's path and what happened 7, 14 and 28 days later."
- shape_analogs with timeframe 1d searches only that coin's own past on that exchange, one match per episode, none overlapping your window. You can ask for 1 to 20 matches and a window of 20 to 200 bars.
- Each case comes with its own path and later moves in basis points, at a quarter, a half and the full window length: 7, 14 and 28 days for a 28-bar window. 100 basis points equals 1%.
- It does not return an average or a hit rate. A search costs $0.01 from credits.
See the docs to connect your agent, or try 3 coins free. These are past cases, not a forecast.
The short version
- Bitcoin's close was 10% or more below its close a week earlier on 6.79% of Binance days and 8.41% of Coinbase days.
- Those days form 46 episodes on Binance and 77 on Coinbase, and only 18 and 29 of them were 90 days or more apart.
- The share swung from 24.66% of Coinbase days in 2018 to 1.92% in 2023.
- Weekly drops of 30% were rare: 9 Binance days and 12 Coinbase days.
- Thirty days after a weekly drop, the cases were not clearly different from all days.
Questions people ask
How often does Bitcoin drop 10% in a week?
On 6.79% of Binance BTCUSDT days (176 of 2,592, September 2019 to October 19, 2026) the daily close was 10% or more below the close 7 days earlier. On Coinbase BTC-USD it was 8.41% of days (345 of 4,103, July 2015 on). Those days come in bunches.
How many separate times did Bitcoin fall 10% in a week?
Binance BTCUSDT had 46 separate episodes, and 18 of them were at least 90 days apart. Coinbase BTC-USD had 77 episodes, with 29 at least 90 days apart. An episode starts on the first day of a drop after at least 7 days without one.
How often does Bitcoin drop 20% or 30% in a week?
A 20% weekly drop happened on 1.27% of Binance days and 1.46% of Coinbase days. A 30% weekly drop was rare: 0.35% of Binance days (9 days, 7 of them in 2020) and 0.29% of Coinbase days (12 days).
Does Bitcoin usually bounce after a 10% weekly drop?
The past cases do not show a reliable difference. Thirty days after a 10% weekly drop, Coinbase BTC-USD was higher in 57.14% of 77 cases, against 58.04% after all days. Binance BTCUSDT was higher in 50.00% of 46 cases, but in 55.56% of the 18 cases spaced 90 days apart, close to the 55.82% for all days.