Bitcoin 5-minute market statistics
When Bitcoin was already higher four minutes into a five-minute window, it finished that window higher 84.9% of the time (41,945 of 49,398). The sample is 100,000 complete Binance spot BTCUSDT windows opening from 20 October 2025 at 19:10 UTC through 3 October 2026 at 00:25 UTC.
Last updated October 3, 2026, 00:33 UTC
These figures are Binance spot prices. They are not Hyperprice rounds, not the published index in the app, and not a settlement record. This study is in English. Choosing another language in the menu opens that language’s Learn home.
Already higher after four minutes
Across the whole sample, a window finished higher 49.7% of the time and lower 50.0% of the time. The four-minute mark is a different question: the price is already on one side of the open, and the last minute still has to print.
| Mark | Already higher, then finished higher | Already lower, then finished lower |
|---|---|---|
| After minute 1 | 65.2% (31,532 of 48,373) | 65.4% (31,846 of 48,727) |
| After minute 2 | 71.9% (35,264 of 49,034) | 71.9% (35,678 of 49,646) |
| After minute 3 | 78.2% (38,615 of 49,365) | 78.1% (38,969 of 49,870) |
| After minute 4 | 84.9% (41,945 of 49,398) | 84.6% (42,370 of 50,093) |
A window that was flat at that minute is left out of both columns. “Finished higher” means the fifth-minute close was above the window’s open, not equal to it.
How the windows finished
| Finish | Windows | Share |
|---|---|---|
| Higher than the open | 49,748 | 49.7% |
| Lower than the open | 49,964 | 50.0% |
| Unchanged | 288 | 0.3% |
The average absolute move was 0.089%. The median was 0.056%. Absolute move is the size of the change, ignoring direction.
| Absolute move | Windows | Share |
|---|---|---|
| More than 0.10% | 29,749 | 29.7% |
| More than 0.25% | 6,420 | 6.4% |
| More than 0.50% | 1,037 | 1.0% |
| More than 1.00% | 97 | 0.1% |
Hour and weekday
The largest average absolute moves were at 14:00, 15:00, 16:00 UTC. Bar height is that average. The table is the same data.
| Hour (UTC) | Windows | Average absolute move |
|---|---|---|
| 00:00 | 4,170 | 0.093% |
| 01:00 | 4,164 | 0.094% |
| 02:00 | 4,164 | 0.086% |
| 03:00 | 4,164 | 0.075% |
| 04:00 | 4,164 | 0.071% |
| 05:00 | 4,164 | 0.072% |
| 06:00 | 4,164 | 0.070% |
| 07:00 | 4,164 | 0.071% |
| 08:00 | 4,164 | 0.078% |
| 09:00 | 4,164 | 0.074% |
| 10:00 | 4,164 | 0.067% |
| 11:00 | 4,164 | 0.071% |
| 12:00 | 4,164 | 0.087% |
| 13:00 | 4,164 | 0.112% |
| 14:00 | 4,164 | 0.140% |
| 15:00 | 4,164 | 0.133% |
| 16:00 | 4,164 | 0.115% |
| 17:00 | 4,164 | 0.105% |
| 18:00 | 4,164 | 0.097% |
| 19:00 | 4,174 | 0.088% |
| 20:00 | 4,176 | 0.083% |
| 21:00 | 4,176 | 0.079% |
| 22:00 | 4,176 | 0.090% |
| 23:00 | 4,176 | 0.078% |
| Weekday (UTC) | Windows | Average absolute move |
|---|---|---|
| Monday | 14,170 | 0.100% |
| Tuesday | 14,400 | 0.098% |
| Wednesday | 14,400 | 0.098% |
| Thursday | 14,400 | 0.101% |
| Friday | 14,400 | 0.102% |
| Saturday | 14,118 | 0.054% |
| Sunday | 14,112 | 0.066% |
Consecutive finishes
A streak is a run of higher finishes, or a run of lower finishes. An unchanged window ends the run. The longest higher streak was 15 windows. The longest lower streak was 15. The counts below are streaks, not windows.
| Length | Higher streaks | Lower streaks |
|---|---|---|
| 1 | 12,959 | 12,921 |
| 2 | 6,489 | 6,518 |
| 3 | 3,241 | 3,238 |
| 4 | 1,571 | 1,518 |
| 5 | 698 | 696 |
| 6 | 346 | 344 |
| 7 | 156 | 195 |
| 8 | 79 | 83 |
| 9 | 24 | 31 |
| 10 | 12 | 18 |
| 11 or more | 15 | 16 |
The first minute
When the first minute and the final close were both away from the open, they pointed the same way 65.4% of the time (63,378 of 96,923). The first minute was unchanged in 2,900 windows, and those are not in that rate.
After a rising or falling 30 days
Each window is tagged before its own result is used. The tag is the change from the close of the 1-minute candle that opened exactly 30 days earlier to this window’s open. A higher open than that older close is the rising group. A lower open is the falling group.
| Prior 30 days | Windows | Finished higher | Average absolute move |
|---|---|---|---|
| Price was higher than 30 days earlier | 50,165 | 49.5% | 0.074% |
| Price was lower than 30 days earlier | 49,835 | 50.0% | 0.103% |
0 windows had no candle exactly 30 days earlier, and 0 had the same price. Those windows stay in the rest of the study and are left out of this table.
How the sample was built
- Source: Binance spot BTCUSDT 1-minute candles from the public market-data API (https://data-api.binance.vision/api/v3/klines).
- A window opens on a UTC 5-minute boundary. Its open is the first minute’s open. Its close is the fifth minute’s close.
- The mark after minute 1, 2, 3, or 4 is that minute’s close.
- Higher means the close is above the open. Lower means it is below. Equal prices are unchanged.
- 1 aligned window in this span was missing a minute and was dropped. No price was interpolated.
- The dataset is rebuilt once a day from the latest 100,000 complete windows. The last-updated time above is that rebuild.
Download the window CSV. One row per window: open time, open, close, the four minute marks, and higher, lower, or unchanged.
What this does not say
- One exchange, spot BTCUSDT only. Other venues and other pairs can differ.
- Past windows are not a forecast of the next window.
- This series is not the Hyperprice published index, and it does not tell you whether a Hyperprice round settled UP or DOWN.
- A five-minute move of this size is not a statement about holding Bitcoin, and it is not a reason to trade.
Questions
Is this Hyperprice’s settlement price?
No. Every figure on this page uses Binance spot BTCUSDT 1-minute candles. Hyperprice rounds settle on the published index shown in the app, which is a different series.
Does a higher finish mean the UP side won?
No. On Hyperprice, UP wins when a round closes at or above its open. This study counts a higher finish only when the Binance close is strictly above the open. An unchanged window is its own row.
Can this predict the next five minutes?
No. These are past windows. A rate from the last year is not a forecast, and it is not a reason to buy a side.
What happened to minutes Binance did not print?
That five-minute window was dropped. 1 aligned window in the span was incomplete and is not in the 100,000. Nothing was filled in.