What happens after Bitcoin drops suddenly in one hour?
After big 1-hour drops, Binance Bitcoin was up a day later 56.16% of the time vs 52.04% for all hours, and 53.77% with drops spaced 48 hours apart.
Bitcoin ended the next 24 hours higher in 56.16% of 3,846 large-drop cases we measured. We used hourly prices for Binance BTCUSDT, a Bitcoin contract priced in USDT, from 8 September 2019 to 20 October 2026. Across the same history, price rose in 52.04% of 61,614 day-ahead cases in general. When we counted only drops at least 48 hours apart, the share fell to 53.77% of 876 cases.
For drops of 10% over a whole week, see Bitcoin after a 10% weekly drop. For the opposite move, see What happens after Bitcoin's biggest 1-hour pumps?.
What counts as a big one-hour drop?
We let the cutoff change with the market. A fall counted when it was larger than three times that market's typical hourly move over the previous 30 days. This finds unusually large falls for the conditions at the time, including falls outside famous crashes.
A price bar records the opening, highest, lowest, and closing price during a set period. We used one-hour bars. Each change ran from the previous hour's closing price to the current hour's closing price. A sharp fall that fully recovered before the hour ended could go uncounted.
Here, “typical” means the median size of the previous 720 hourly changes. We ignored whether those moves went up or down. The median is the middle value after sorting the moves by size. We left the drop hour out of that calculation.
This matters because a fall can be unusual during a quiet month but ordinary during a rough one. Our test compares each market with its own recent history.
These drops are not rare. Our rule flagged 3,847 Binance Bitcoin drops, between 468 and 580 in each full year from 2020 to 2025. That is more than one a day on average, so “big” here means big for its month, not a famous crash.
What did Bitcoin do over the next day or two?
The table below uses the same Binance BTCUSDT hourly history, from 8 September 2019 to 20 October 2026. USDT is a token designed to track the US dollar. This market is a perpetual futures contract, or perp: a contract tied to Bitcoin's price that never expires.
Each row includes only drops with no missing hourly prices through the end of the wait. “Up” means price ended above the drop hour's closing price, before fees. The lower cutoff marks the boundary of the worst tenth of results. The upper cutoff marks the boundary of the best tenth.
| Time after drop | Complete cases | Share up | Median move | Lower cutoff | Upper cutoff |
|---|---|---|---|---|---|
| 12 hours | 3,846 | 54.86% | +0.19% | −2.60% | +2.85% |
| 24 hours | 3,846 | 56.16% | +0.32% | −3.53% | +4.22% |
| 48 hours | 3,844 | 55.54% | +0.41% | −4.88% | +5.70% |
A positive middle result did not mean further falls were small. In those same 3,846 Binance day-ahead cases, the lower cutoff was −3.53%. About a tenth of cases ended at or below that further loss from the drop hour's closing price.
Also, a rise from the drop hour's closing price does not mean the whole fall was recovered. We measured what happened after the fall, not whether Bitcoin returned to its earlier price. Our guide to crypto drawdowns and recovery times covers that separate question.
How different was that from ordinary hours?
A base rate tells us how often something happened without the special rule we are testing. Here, it means how often Bitcoin rose after hours in general.
Across the full Binance BTCUSDT hourly history above, 51.88% of 62,335 complete day-ahead cases ended higher. But the drop rule needed a full earlier month of prices. Keeping only hours with that earlier history gives the 52.04% rate across 61,614 cases used in our opening.
That comparison includes all hours with enough earlier prices, including drop hours. It does not include only calm hours. For the same Binance Bitcoin history, the share that rose after drops was 4.13 percentage points higher. Percentage points describe the gap between rates.
The bad results were also worse after drops. Among those 61,614 Binance day-ahead cases, the lower cutoff was −3.07%, compared with −3.53% after drops. A higher share of rises did not mean smaller losses when price fell.
The groups can still differ in news, market mood, or how much prices were moving. See why past crypto moves are not forecasts for more on reading these comparisons. For drops tied to forced trade closures, see what happened after liquidation cascades.
Did Coinbase, Ethereum, and Solana show the same thing?
The next table uses the same hourly drop rule and measures price 24 hours later. Counts appear beside each rate. “All hours” means all hours with enough earlier prices for the rule.
All price histories end on 20 October 2026. Binance BTC starts on 8 September 2019; Binance ETH, or Ethereum, starts on 27 November 2019; Binance SOL, or Solana, starts on 14 September 2020. Coinbase BTC-USD starts on 20 July 2015 and is a spot market, where Bitcoin itself trades for US dollars.
| Hourly market | Share up after drops (cases) | Share up after all hours (cases) | Median after drops |
|---|---|---|---|
| Binance BTCUSDT perp | 56.16% (3,846) | 52.04% (61,614) | +0.32% |
| Coinbase BTC-USD spot | 56.67% (5,414) | 53.01% (83,550) | +0.38% |
| Binance ETHUSDT perp | 56.03% (3,684) | 52.14% (59,704) | +0.43% |
| Binance SOLUSDT perp | 53.52% (2,816) | 49.77% (52,696) | +0.45% |
Each market showed a higher share of rises after drops, measured against its own past. Coinbase's history starts earlier, so the table cannot tell us which exchange rebounds more often over the same dates.
Solana's larger median rise also came with a wider range of results. In its 2,816 day-ahead drop cases over the dates above, the lower and upper cutoffs were −6.48% and +8.81%. A larger typical rise alone tells you little about the size of the bad outcomes.
What changes when nearby drops count less?
A rough stretch can contain several hours that meet our drop rule. Their later prices cover some of the same time, so counting each drop counts parts of the same stretch again.
We repeated the test by keeping a drop, then skipping further drops less than 48 hours later. In the same Binance BTCUSDT hourly history, this left 876 complete day-ahead cases. The share that rose fell to 53.77%, and the median move fell to +0.16%.
For Binance ETHUSDT over its hourly history and dates above, this left 841 complete day-ahead cases. The share that rose fell to 49.94%, with a median move just below zero. Ethereum no longer rose in most cases when we changed how we counted drops. Coinbase BTC-USD held up better: 55.08% of its 1,171 spaced day-ahead cases rose.
A check you can run in your own research: whenever a study counts “events,” count each rough stretch once and see whether the result survives. Here, Binance Bitcoin's day-ahead share up went from 56.16% to 53.77% that way, and Ethereum's lead over a coin flip disappeared.
Comparand's shape_analogs search returns one match per past episode to avoid counting nearby charts separately.
How we measured this
We used public Binance and Coinbase hourly prices for the markets and dates named above. We required a complete earlier month and no missing hourly prices through each later measurement. We measured from the drop hour's closing price, when we could first know that the drop met the rule. For the 876 spaced Binance BTCUSDT day-ahead cases above, fees of 0.05% at entry and 0.05% at exit reduced the median move to +0.06%. That fee check assumes trades at recorded closing prices and leaves out worse trade prices, a trade's effect on price, borrowing costs, and payments for keeping perp positions open. The selected coins still trade today, and cases that share the same hours or market conditions limit what we can conclude.
Try this in Comparand
Give your connected AI app the drop hour's closing time and ask: “Show completed hourly Binance BTCUSDT bars around this drop. Find similar earlier Bitcoin charts on Binance using the 48 hourly bars ending then. Show each case and what happened next.”
You get:
- Price bars: venue_bars gives completed bars with opening, highest, lowest, and closing prices, plus volume, or how much traded. It is free with a key across exchanges, including Binance and Coinbase.
- Similar charts: shape_analogs returns matches from the same coin, exchange, and bar size, one per episode. Each shows its own path and, for this chart length, moves 12, 24, and 48 hours later. A search costs $0.01 from your account credit.
The search gives no average or share that rose; we calculated these tables separately. Matching whole charts differs from testing our hourly drop cutoff. You can try 3 coins free; see the connection guide for tool access.
These are past cases, not a forecast.
The short version
- Bitcoin rose more often after unusually large hourly drops than after ordinary hours in the measured histories.
- Rising from the drop close does not mean recovering the whole fall.
- Bad outcomes still included substantial further losses.
- Counting drops farther apart weakened Bitcoin's result. Ethereum no longer rose in most day-ahead cases.
Questions people ask
Does Bitcoin usually bounce after a sudden drop?
A little more often than usual, in the past. On Binance BTCUSDT from September 2019 to October 2026, price was higher 24 hours after an unusually large hourly drop in 56.16% of 3,846 cases, versus 52.04% for all hours. Counting only drops at least 48 hours apart, it was 53.77% of 876 cases.
Does a bounce mean Bitcoin recovered the whole drop?
No. We measured price changes from the end of the hour when the drop happened. Price could rise from there while remaining below where the drop began.
What counts as a big hourly Bitcoin drop?
We compared each hourly fall with that market's typical hourly move over the preceding month. The size needed to count as a big drop changed with the market.
Can this study explain why Bitcoin dropped?
No. Hourly price records show how price changed. They cannot tell us whether news, forced trade closures, or something else caused a fall.