Trend gate RED · 7-day range $79,830–$88,585 (80%)
1 · Trend gate
RED BTC return over the last 28 days: -3.55%.
The rule is simple: GREEN when BTC is up over 28 days, RED otherwise. It passed an out-of-sample test on historical data. It is a historical rule and it can stop working, and it says nothing about where the price goes next.
2 · Expected range
Where BTC is likely to close (Binance, 00:00 UTC) at each horizon. The center is today's price: this index says how wide the outcomes are, not which way they go.
A hundred ways the next 30 days could go
Each line is one path the model finds plausible. None of them is a prediction. The shaded bands hold 50% and 80% of where they end.
| Horizon | Closes on | 5th | 10th | 25th | Median | 75th | 90th | 95th |
|---|---|---|---|---|---|---|---|---|
| 1 day | Oct 23, 2026 | $81,997 | $82,456 | $83,228 | $84,093 | $84,968 | $85,764 | $86,243 |
| 7 days | Oct 29, 2026 | $78,661 | $79,830 | $81,822 | $84,093 | $86,428 | $88,585 | $89,901 |
| 30 days | Nov 21, 2026 | $73,235 | $75,506 | $79,459 | $84,093 | $88,998 | $93,658 | $96,561 |
The same forecast as questions
The chance Bitcoin closes above each level on each date. These are the model's chances, graded on the date with the Brier score below. Nothing here is a market: there is nothing to buy or sell.
Bitcoin price on Oct 23, 2026?
Close at 00:00 UTC · 1 day out · today $84,093
- Above $92,503 <1% chance
- Above $88,298 <1% chance
- Above $79,889 >99% chance
- Above $75,684 >99% chance
Bitcoin price on Oct 29, 2026?
Close at 00:00 UTC · 7 days out · today $84,093
- Above $92,503 <1% chance
- Above $88,298 11.5% chance
- Above $79,889 89.7% chance
- Above $75,684 >99% chance
Bitcoin price on Nov 21, 2026?
Close at 00:00 UTC · 30 days out · today $84,093
- Above $92,503 12.8% chance
- Above $88,298 28.1% chance
- Above $79,889 72.9% chance
- Above $75,684 89.5% chance
3 · Old forecasts don't disappear
On Oct 22, 2026 the index said: there is an 80% chance Bitcoin closes between $75,506 and $93,658 on Nov 21, 2026. Today it is at $81,416, inside that range, with 29 days to go.
Tomorrow there will be a new forecast, but it does not erase this one. Every forecast stays here until its own date arrives. On that day we check the price: inside the range is a hit, outside is a miss, and both go to the scorecard below. The ledger proves each one was made before its date.
Each fan is one day's forecast. The black line is the real price.
Every day the index draws a range for 30 days out. Here they all are, from the day each was made to the day it gets checked. The shaded one is the newest. If the black line ends inside a fan on that fan's last day, it was a hit.
30-day forecasts still waiting for their date
| Made on | Checked on | 80% range | Range and today's price | Today's price is | Days left |
|---|---|---|---|---|---|
| Oct 22 | Nov 21 | 75,506–93,658 | Inside | 29 | |
| Oct 21 | Nov 20 | 74,471–92,569 | Inside | 28 | |
| Oct 20 | Nov 19 | 73,424–91,440 | Inside | 27 | |
| Oct 19 | Nov 18 | 73,334–91,949 | Inside | 26 | |
| Oct 18 | Nov 17 | 72,575–91,502 | Inside | 25 | |
| Oct 17 | Nov 16 | 72,424–91,973 | Inside | 24 | |
| Oct 16 | Nov 15 | 72,765–92,962 | Inside | 23 | |
| Oct 15 | Nov 14 | 72,031–92,626 | Inside | 22 | |
| Oct 14 | Nov 13 | 73,347–94,306 | Inside | 21 | |
| Oct 13 | Nov 12 | 72,125–92,886 | Inside | 20 | |
| Oct 12 | Nov 11 | 74,019–94,812 | Inside | 19 | |
| Oct 9 | Nov 8 | 71,567–93,392 | Inside | 16 | |
| Oct 8 | Nov 7 | 72,940–95,181 | Inside | 15 | |
| Oct 7 | Nov 6 | 75,195–97,331 | Inside | 14 |
Grey bar: the range that forecast gave. Black line: today's price ($81,416, Oct 23). Being outside today is not a miss yet: only the price on the "checked on" date counts.
7-day forecasts still waiting for their date
| Made on | Checked on | 80% range | Range and today's price | Today's price is | Days left |
|---|---|---|---|---|---|
| Oct 22 | Oct 29 | 79,830–88,585 | Inside | 6 | |
| Oct 21 | Oct 28 | 78,778–87,507 | Inside | 5 | |
| Oct 20 | Oct 27 | 77,709–86,398 | Inside | 4 | |
| Oct 19 | Oct 26 | 77,750–86,727 | Inside | 3 | |
| Oct 18 | Oct 25 | 77,055–86,182 | Inside | 2 | |
| Oct 17 | Oct 24 | 77,039–86,464 | Inside | 1 | |
| Oct 16 | Oct 23 | 77,521–87,258 | Inside | Checking |
Grey bar: the range that forecast gave. Black line: today's price ($81,416, Oct 23). Being outside today is not a miss yet: only the price on the "checked on" date counts.
4 · Scorecard
Our chances had 66% less error than chances taken only from Bitcoin's history. 0 means no better than history, 1 means perfect.
Each day the index publishes chances like "11.5% that Bitcoin closes above $88,298 in 7 days". The Brier score is the average squared distance between that chance and what happened (1 if it closed above, 0 if not): 0 is perfect, lower is better. We compare it with the same score for a reader who only knows history: how often Bitcoin has ended above that level over the same number of days since 2017, counting only days before the forecast. The 1-day chances are left out: they are almost always near 0% or 100%, and would flatter the score.
| Horizon | Probabilities graded | Brier, ours | Brier, history | Skill |
|---|---|---|---|---|
| 7 days | 28 | 0.011 | 0.034 | 0.66 |
Resampling the forecasts gives a range of 0.64 to 0.68. Treat it as too narrow while the record is short: forecasts made a day apart overlap and share the same market.
Do the chances mean what they say?
| When we said | Cases | Average chance said | How often it happened |
|---|---|---|---|
| under 20% | 14 | 9% | 0% |
| 80% or more | 14 | 92% | 100% |
If the chances are honest, the last two columns match once there are enough cases.
Resolved questions
| Question | We said | Resolved |
|---|---|---|
| Above $89,850 on Oct 22? | 2.2% | No |
| Above $85,766 on Oct 22? | 15.2% | No |
| Above $77,598 on Oct 22? | 86.1% | Yes |
| Above $73,514 on Oct 22? | 98.7% | Yes |
| Above $91,486 on Oct 21? | 2.2% | No |
| Above $87,327 on Oct 21? | 15.2% | No |
| Above $79,011 on Oct 21? | 86.1% | Yes |
| Above $74,852 on Oct 21? | 98.7% | Yes |
| Above $90,035 on Oct 20? | 2.3% | No |
| Above $85,943 on Oct 20? | 15.3% | No |
| Above $77,758 on Oct 20? | 85.9% | Yes |
| Above $73,665 on Oct 20? | 98.7% | Yes |
The latest graded questions. A red row means the outcome went against the side our chance leaned to; that happens about as often as the chances say it should, if they are honest.
Ranges
Every forecast is written before the outcome is known and graded automatically against the Binance close. A well-calibrated 80% band should contain about 80% of outcomes. Small samples swing a lot; the counts are shown so you can judge.
| Horizon | Graded | Inside 80% band | Inside 50% band | Average error |
|---|---|---|---|---|
| 1 day | 13 | 85% | 46% | 0.33% |
| 7 days | 7 | 100% | 100% | 0.63% |
| 30 days | 0 | — | — | — |
Average error is the pinball loss, the standard grade for percentile forecasts: it adds up how far the real close fell from each published percentile, as a share of the price. Lower is better.
Graded 1-day forecasts
| Made on | Checked on | 80% range | Price that day | Result |
|---|---|---|---|---|
| Oct 21 | Oct 22 | 81,396–84,694 | 84,093 | In |
| Oct 20 | Oct 21 | 80,313–83,596 | 83,028 | In |
| Oct 19 | Oct 20 | 80,438–83,829 | 81,938 | In |
| Oct 18 | Oct 19 | 79,785–83,233 | 82,116 | In |
| Oct 17 | Oct 18 | 79,855–83,416 | 81,491 | In |
| Oct 16 | Oct 17 | 80,427–84,106 | 81,616 | In |
| Oct 15 | Oct 16 | 79,828–83,579 | 82,246 | In |
| Oct 14 | Oct 15 | 81,282–85,099 | 81,682 | In |
| Oct 13 | Oct 14 | 79,982–83,762 | 83,169 | In |
| Oct 12 | Oct 13 | 81,901–85,688 | 81,850 | Miss |
Graded 7-day forecasts
| Made on | Checked on | 80% range | Price that day | Result |
|---|---|---|---|---|
| Oct 15 | Oct 22 | 76,869–86,797 | 84,093 | In |
| Oct 14 | Oct 21 | 78,270–88,374 | 83,028 | In |
| Oct 13 | Oct 20 | 76,999–87,007 | 81,938 | In |
| Oct 12 | Oct 19 | 78,911–88,935 | 82,116 | In |
| Oct 9 | Oct 16 | 76,664–87,183 | 82,246 | In |
| Oct 8 | Oct 15 | 78,135–88,853 | 81,682 | In |
| Oct 7 | Oct 14 | 80,381–91,051 | 83,169 | In |
Graded 30-day forecasts
No 30-day forecast has reached its date yet; the first one is checked on Nov 6, 2026. Until then you can follow them above.
The tables show the last ten graded forecasts per horizon; the counts above include all of them. We will not claim accuracy before roughly 90 days of record.
5 · Volatility level
The level is the percentile of today's 30-day realized volatility (standard deviation of daily log returns, annualized) among all 3,324 days since 2017-09-16. 0 is the calmest month on record, 100 the most turbulent. It is a measurement, not a forecast: a calm month says nothing about which way the next move goes.
Get the weekly recap, misses included: by RSS.
What this is — and isn't
The BTC Oracle Index is the daily output of a set of statistical models that are published before the outcome is known and graded publicly afterward, including misses. It does not predict whether Bitcoin will go up or down: none of the return signals we tested held up out of sample, and this index makes no directional call. The expected range describes uncertainty, not a target. The trend gate is a historical rule that has worked in past data and may stop working. Past performance does not guarantee future results. Nothing here is a recommendation to buy, sell or hold any asset, and nothing is tailored to your situation. The author may hold crypto assets, including BTC.