Does Bitcoin dump at the US market open? 1,788 days measured
Binance Bitcoin's 9:30 a.m. New York hour rose on 48.10% of 1,788 days, averaging +0.7 bp. No drop pattern, but moves were 2.02x a weekend hour.
No, not on average. On Binance, the hourly Bitcoin bar that contains 9:30 a.m. New York time rose on 48.10% of 1,788 US trading days from September 2019 to October 2026, and its average move was +0.7 basis points (1 basis point, or bp, is 0.01%). That is tiny and well inside the noise.
What was different was size. That hour's average move, ignoring direction, was 58.0 bp (0.58%) on trading days versus 28.7 bp (0.29%) in the same UTC hour on weekends. The open is a busy hour. It is not a falling hour. These are past cases, not a forecast.
What did we measure at the US open?
We used hourly price bars. A bar holds the open, high, low and close price for one hour. For each US trading day, we took the bar that contains 9:30 a.m. New York time and measured its return: closing price compared with opening price.
In US summer (daylight time) 9:30 a.m. is 13:30 UTC, so the bar starts at 13:00 UTC. In winter it is 14:30 UTC, so the bar starts at 14:00 UTC. We handled that switch every year, and skipped weekends and New York Stock Exchange (NYSE) holidays.
Because bars are an hour long, this bar includes the 30 minutes before the open. Data releases at 8:30 a.m. New York time fall in the bar before it; our CPI and Fed day article measures those. Summer and winter looked alike: the bar rose on 46.63% of 1,188 daylight-time days and 51.00% of 600 winter days.
Did Bitcoin fall more often than it rose at the open?
Slightly more often, but not by enough to matter. Binance rows are perps (futures contracts that never expire) priced in USDT. Coinbase is spot priced in US dollars. We never mixed them.
| Market | Trading days | Average move | Middle (median) move | Share of days up | Average size, trading days | Average size, weekend hour | Ratio |
|---|---|---|---|---|---|---|---|
| Binance BTCUSDT | 1,788 | +0.7 bp | -2.8 bp | 48.10% | 58.0 bp | 28.7 bp | 2.02 |
| Binance ETHUSDT | 1,731 | -0.7 bp | -4.4 bp | 47.60% | 71.4 bp | 45.4 bp | 1.57 |
| Binance SOLUSDT | 1,532 | +3.4 bp | -5.2 bp | 48.17% | 102.7 bp | 70.0 bp | 1.47 |
| Coinbase BTC-USD | 2,827 | +0.7 bp | -0.5 bp | 49.35% | 57.2 bp | 35.8 bp | 1.60 |
"Average size" means the move with the plus or minus sign removed. The ratio divides the trading-day figure by the weekend figure.
All four markets were up on a little under half of days, but the average moves ranged from -0.7 bp to +3.4 bp, with different signs. For Binance Bitcoin, the margin of error on the average (its standard error) was 2.0 bp, so +0.7 bp cannot be told apart from zero.
Is the 2x bigger move really an "open effect"?
Only partly. The weekend comparison mixes two things: the open, and the fact that weekdays are busier than weekends. For Binance Bitcoin, the weekday bar three hours before the open already averaged 37 bp, against 29 bp on weekends. So some of the 2.02 ratio would show up with no open at all.
Over all days, the Binance Bitcoin bar starting at 14:00 UTC was the busiest of the 24 (average size 53.7 bp) and 04:00 UTC the quietest (29.6 bp). The US open sits inside that busy stretch. See which hours move crypto the most and our weekend volatility article for more.
Does the direction change from year to year?
Yes. Here is Binance BTCUSDT by year for the US-open bar.
| Year | Trading days | Average move | Share of days up | Average size | Weekend average size | Ratio |
|---|---|---|---|---|---|---|
| 2019 | 80 | 0.0 bp | 51.25% | 41.0 bp | 40.0 bp | 1.02 |
| 2020 | 253 | +7.9 bp | 49.01% | 51.7 bp | 35.5 bp | 1.46 |
| 2021 | 252 | +5.3 bp | 53.17% | 79.3 bp | 42.4 bp | 1.87 |
| 2022 | 251 | +1.8 bp | 47.81% | 62.1 bp | 31.8 bp | 1.95 |
| 2023 | 250 | -3.2 bp | 44.00% | 38.4 bp | 15.5 bp | 2.48 |
| 2024 | 252 | +0.2 bp | 47.22% | 67.5 bp | 25.3 bp | 2.67 |
| 2025 | 250 | -1.3 bp | 50.40% | 55.0 bp | 25.2 bp | 2.18 |
| 2026 (to Oct 19) | 200 | -6.7 bp | 43.00% | 56.9 bp | 20.1 bp | 2.84 |
The sign flips: positive in 2020 to 2022 and 2024, negative in 2023, 2025 and 2026. 2026 was the weakest year so far, up on only 43.00% of 200 days. But each year's margin of error is between 3.5 and 6.8 bp, so even 2026's -6.7 bp does not clearly stand apart from zero.
The ratio grew, from 1.02 in 2019 to 2.84 in 2026. But look at the weekend column: it fell from 40.0 bp to 20.1 bp. Weekends got calmer while the US-open bar stayed in a band of 38.4 to 79.3 bp. We did not test why.
What about the hour after the open?
The next hourly bar was also flat on trading days: average -2.2 bp, up on 48.83% of days, average size 65.2 bp. The same UTC hour on weekends was up on 56.06% of 742 bars. That is one of many comparisons we ran, so we do not treat it as a pattern.
Does the 00:00 UTC hour have a pattern?
Some charts start a new daily candle at 00:00 UTC, which is 8 p.m. in New York in summer and 7 p.m. in winter. People ask if price tends to jump or drop then.
On Binance BTCUSDT, the 00:00 UTC bar averaged +1.2 bp, with an average size of 43.7 bp. It was up on 48.85% of 2,598 days, and it had the same sign as its overall average in only 5 of 8 years. By year, the average ran from -4.3 bp in 2025 to +5.7 bp in 2023. Our daily candle close article explains why that hour matters for charts.
Does any hour of the day show a direction pattern?
A little, but probably by chance. For Binance Bitcoin, 3 of 24 hours showed a statistically notable direction (a t-statistic above 2). Chance alone gives about 1 in 20, so you would expect about 1. When we correct for testing 24 hours at once, only one hour passes: the bar starting 21:00 UTC, just after US stocks close, which averaged +4.4 bp and rose on 54.33% of days. It is not the US open. Across the twelve markets we checked, at most two hours per market passed that stricter bar, so we do not treat any hour as a pattern.
Even +4.4 bp is smaller than the 0.05% (5 bp) a taker pays in fees on each side of a trade.
What does this not show?
- Early-close days stayed in, such as the day after Thanksgiving.
- One hourly bar hides the minutes. A fall and a rebound inside the hour can cancel out.
- News and weekdays overlap. Big news days can sit on top of the open.
- We did not test a trading rule. Fees beyond the note above, funding, spread and slippage were left out.
How we measured this
We used hourly bars from Binance perps (Bitcoin from September 2019) and Coinbase spot (Bitcoin from July 2015), ending October 19, 2026. Each return is closing price divided by opening price, minus one. NYSE holidays from 2015 to 2026 were removed. We converted 9:30 a.m. New York time to UTC for each date, so the daylight-time switch is handled. We grouped returns by hour, and compared the US-open bar with the same UTC hours on Saturdays and Sundays. We ran no fee-adjusted trading test.
Try this in Comparand
Ask your connected AI agent: "Get hourly Binance BTCUSDT bars for the dates I choose. For each weekday, find the bar that contains 9:30 a.m. New York time, and show how often it closed up and how big it was versus the same hour on weekends."
- venue_bars (1h) returns open, high, low, close and volume for completed hourly bars. It is free on every key.
- Your agent does the New York time conversion and the averages. The tool does not return a "US open report."
To zoom in on the minutes around the open, ask: "Show one-minute Bitcoin bars on Hyperliquid for the half hour after 9:30 New York time on a day I choose."
- hl_bars returns price bars for one coin from the Hyperliquid archive, one minute by default. It costs $0.005 per call on pay as you go and is unlimited on plans under fair use.
- Call hl_coverage first (free) to see which archive days are complete.
Open the docs to connect your agent, or try 3 coins free. Past cases, not a forecast.
The short version
- Binance Bitcoin rose in 48.10% of 1,788 US-open hourly bars; the average move was +0.7 bp.
- The hour is busy: 58.0 bp versus 28.7 bp on weekends, a ratio of 2.02.
- Part of that gap is plain weekday-versus-weekend activity.
- The direction flipped by year, from +7.9 bp in 2020 to -6.7 bp in 2026.
- We found no reliable best or worst hour, including 00:00 UTC.
Questions people ask
Does Bitcoin drop at the US stock market open?
Not on average. On Binance BTCUSDT, the hourly bar containing 9:30 a.m. New York time rose on 48.10% of 1,788 trading days from September 2019 to October 2026. The average move was +0.7 basis points, and the middle day was -2.8 basis points. Past cases, not a forecast.
Is Bitcoin more volatile at the US open?
Yes. The average size of the move in that hourly bar was 58.0 basis points (0.58%) on trading days versus 28.7 on weekends in the same UTC hour, about 2.02 times as much. Part of that gap is simply weekdays being busier than weekends.
What time is the US stock market open in UTC?
9:30 a.m. New York time is 13:30 UTC when the US is on daylight time and 14:30 UTC in winter. We used the hourly bar that starts at 13:00 UTC or 14:00 UTC, which covers the 30 minutes before the open too.
Does Bitcoin drop at 00:00 UTC?
No clear pattern. The 00:00 UTC hourly bar on Binance BTCUSDT averaged +1.2 basis points, rose 48.85% of 2,598 days, and had the same sign as its overall average in only 5 of 8 years.
Can I trade the US open pattern?
Our test found no direction pattern to trade. The average move was less than the 5 basis points a taker pays in fees on each side, and the sign flipped from year to year. This is research, not advice.