---
title: 'Bitcoin after a 10% weekly drop: buy-the-dip data'
url: https://comparand.xyz/blog/bitcoin-after-10-percent-weekly-drop
description: After 77 weekly drops of 10% or more on Coinbase since 2015, Bitcoin was higher 30 days later 57.14% of the time, about the same as any day (58.04%).
updated: '2026-10-21'
category: Data
---

# Bitcoin after a 10% weekly drop: buy-the-dip data

After 77 weekly drops of 10% or more on Coinbase since 2015, Bitcoin was higher 30 days later 57.14% of the time, about the same as any day (58.04%).

On Coinbase, Bitcoin's price was higher 30 days later after 57.14% of the 77 times it fell 10% or more in a week, since July 2015. On all days it was higher 58.04% of the time, so a weekly drop did not change the 30-day odds in this sample. The middle (median) 30-day move was +2.1%, against +3.3% on all days.

What did differ was the road. In 44.16% of cases price closed at least 10% lower again within 30 days, against 35.13% on all days. Only 29 of the 77 drops were at least 90 days apart, so the sample is small. These are past cases, not a forecast.

For how often these drops happen, year by year, see [How often does Bitcoin drop 10% in a week?](https://comparand.xyz/blog/how-often-bitcoin-drops-10-percent-week).

## What counts as a 10% weekly drop?

We used daily closes, the last price of each UTC day. A 7-day change is today's close divided by the close 7 days earlier, minus 1. We call a day a **drop day** when that change is −10% or worse.

Drop days come in clumps. After a fall of 10%, the next day often still shows a fall of 10%. So we also count **episodes**: the first drop day after at least 7 days without one. We found 77 episodes on Coinbase BTC-USD, from July 20, 2015 to October 19, 2026. Then we kept only episodes at least 90 days apart, which left 29. The 29 are the fairer count of separate cases.

*Diagram: Illustration of one weekly-drop episode: a made-up price line with the 7-day change under it and a -10% line, the first drop day marked, the 7, 30 and 90 day windows that follow, and bars for the measured counts of drop days, episodes and episodes 90 days apart.*
## How often does Bitcoin fall 10% in a week?

On 345 of 4,103 days (8.41%), the 7-day change was −10% or worse. Deeper drops were rare: a 20% weekly fall happened on 60 days (1.46%) in 23 episodes, and a 30% weekly fall on 12 days in 4 episodes.

It depends a lot on the year:

| Year | Drop days | Days counted | Share of days |
| --- | --- | --- | --- |
| 2015 (from July 20) | 13 | 158 | 8.23% |
| 2016 | 14 | 366 | 3.83% |
| 2017 | 35 | 365 | 9.59% |
| 2018 | 90 | 365 | 24.66% |
| 2019 | 29 | 365 | 7.95% |
| 2020 | 18 | 366 | 4.92% |
| 2021 | 52 | 365 | 14.25% |
| 2022 | 45 | 365 | 12.33% |
| 2023 | 7 | 365 | 1.92% |
| 2024 | 12 | 366 | 3.28% |
| 2025 | 14 | 365 | 3.84% |
| 2026 (to Oct 19) | 16 | 292 | 5.48% |

In 2018, almost one day in four was a drop day. In 2023, about one in fifty. From 2023 to 2025, drop days were under 4% of days. In 2021 and 2022 they were above 12%. For the same count on Binance and other coins, see [how often Bitcoin drops 10% in a week](https://comparand.xyz/blog/how-often-bitcoin-drops-10-percent-week).

## What did Bitcoin do in the next 7, 30 and 90 days?

For each drop-day episode we looked at the price move from that day's close. We compare three groups: every day in the history, all 77 episodes, and the 29 episodes at least 90 days apart. Each cell shows the share of cases where price was higher, and the median move.

| Window after the drop day | All days | All 77 episodes | 29 spaced-out episodes |
| --- | --- | --- | --- |
| 7 days | 54.59% higher, median +0.7% | 58.44% higher, median +1.0% | 62.07% higher, median +3.2% |
| 30 days | 58.04% higher, median +3.3% | 57.14% higher, median +2.1% | 58.62% higher, median +2.0% |
| 90 days | 60.63% higher, median +10.9% | 54.55% higher, median +4.5% | 65.52% higher, median +11.1% |

Look at the 90-day row. The 77 episodes did worse than all days (median +4.5% against +10.9%), and the 29 spaced-out ones did the same as all days (+11.1%). Groups that small can swing a lot. We see no steady difference after a drop, up or down.

The spread is wide too. After the 77 episodes, the 10th percentile 30-day move was −22.28% and the 90th was +34.04%. A percentile cutoff leaves about a tenth of cases below the lower number and a tenth above the upper one. The best tenth of cases gained 34.04% or more, and the worst tenth lost 22.28% or more.

## How much further did price fall?

Bitcoin often kept falling before it turned. For every case we found the lowest daily close in the next 30 days, and compared it with the drop day's close.

| Next 30 days | All days | All 77 episodes | 29 spaced-out episodes |
| --- | --- | --- | --- |
| Median lowest close | −5.8% | −7.5% | −9.2% |
| Closed 10% or more below the drop-day close | 35.13% | 44.16% | 44.83% |
| Closed 20% or more below the drop-day close | 14.93% | 16.88% | 13.79% |

Even on a normal day, the next 30 days dipped a median 5.8%, so a dip is common. After a drop, it was a bit deeper, and a 10% dip was more common. The 20% row shows no clear gap. About one case in six went 20% lower or more. For the other side, see [how long Bitcoin rallies last](https://comparand.xyz/blog/how-long-do-crypto-rallies-last).

## Did it look the same in all eras?

Not quite. The 59 episodes from 2015 to 2022 and the 18 from 2023 to 2026 had different 90-day pictures.

| Period | Episodes | Higher after 30 days | Higher after 90 days | Median 90-day move |
| --- | --- | --- | --- | --- |
| 2015 to 2022 | 59 | 57.63% | 50.85% | +0.2% |
| 2023 to 2026 | 18 | 55.56% | 66.67% | +12.8% |

With 18 cases in the second row, treat this as a note, not a finding.

We repeated the work on Binance Bitcoin (a USDT perp, which is a futures contract that never expires), from September 2019. There were 46 episodes and the 30-day result was 50.00% higher with a median of −0.1%. On all days, it was 55.82% higher and +2.0%. A different exchange, a different start date, and a different answer. This is a reason to read all of these numbers with care. For more on how cases can disagree, see [when past crypto cases disagree](https://comparand.xyz/blog/when-past-crypto-cases-disagree).

## What does this data not show?

- It does not say a drop is a good or bad moment to buy. We did not test any trade.
- It does not say what the next drop will do. A 10% drop in a calm year and in a crash year are not the same situation.
- It does not use one-hour drops. Those are a different question, covered in [Bitcoin after big one-hour drops](https://comparand.xyz/blog/bitcoin-after-big-one-hour-drops).
- The episodes are few. With 29 separate cases, one or two big years can change the answer.
- We checked several markets and drop sizes. Any single cell that looks different can be luck.

## How we measured this

We used Coinbase BTC-USD daily bars from July 20, 2015 through October 19, 2026 (4,110 daily bars), with Binance BTCUSDT daily bars from September 8, 2019 as a check. A drop day is a day when the close is 10% or more below the close 7 days earlier. An episode is the first drop day after at least 7 days without one. For each episode we measured the change from its close to the close 7, 30 and 90 days later. We dropped windows that ran past the last bar. We compare with the same measure for all days in the same history. Cases overlap, so the raw count of 77 overstates how many separate cases there are. We used no fees, since this is not a trade test.

## Try this in Comparand

With Comparand connected to your AI app, you can look for other drops:

- "Get daily bars for Coinbase BTC-USD from 2017 to today. List every day when the close was 10% below the close 7 days earlier." The tool is **venue_bars**. It returns open, high, low, close and volume for completed daily bars, and it is free on every key. Your agent works out the 7-day change.
- "Find the 10 closest past daily charts to Coinbase BTC-USD now, using the last 28 days." The tool is **shape_analogs** with a daily timeframe. It compares the coin only with its own past on the same exchange. It returns each match's own path and what price did 7, 14 and 28 days later (a quarter, half and the full length of the window), in basis points, where 100 basis points is 1%. It costs $0.01 per search from credits, and it gives no average or hit rate.

Daily Coinbase chart matching starts in January 2017. The results are past cases for you to read, not a forecast. You can [try 3 coins free](https://comparand.xyz/try), read the [tool docs](https://comparand.xyz/docs), or see [pricing](https://comparand.xyz/pricing). For how long Bitcoin took to get back to a high, see [drawdowns and recovery times](https://comparand.xyz/blog/crypto-drawdowns-and-recovery-times).

## The short version

- Since 2015 on Coinbase, Bitcoin had 77 weekly drops of 10% or more (29 at least 90 days apart). 30 days later, 57.14% were higher, against 58.04% on all days.
- The 30-day picture matched all days. The road was rougher: the median lowest close in the next 30 days was −7.5%, against −5.8%.
- Drop days were 24.66% of days in 2018 and 1.92% in 2023.
- The 90-day results swung by group and by period, and Binance gave 50.00% higher at 30 days. The sample is small.
- Past cases, not a forecast, and not advice.

## Questions people ask

**Does Bitcoin usually bounce after falling 10% in a week?**

Not more than on any other day, in our test. On Coinbase BTC-USD since July 2015, price was higher 30 days after 57.14% of 77 weekly drops of 10% or more. Across all days it was higher 58.04% of the time.

**How often does Bitcoin fall 10% in a week?**

On Coinbase BTC-USD, 8.41% of 4,103 days had a 7-day fall of 10% or more. It was 24.66% of days in 2018 and 1.92% in 2023.

**Can Bitcoin keep falling after a 10% weekly drop?**

Yes. In 44.16% of the 77 cases, a close in the next 30 days was at least 10% below the drop day's close. For all days it was 35.13%.

**Does this data show that buying the dip works?**

No. The results after drops were close to the results after any day, and the sample is small. These are past cases, not a forecast, and not advice.