Bitcoin price prediction for the next 24 hours
A model estimate for where BTC/USDT could close 24 hours from the latest completed hourly candle, with the range the model has historically missed by.
Current Bitcoin forecast
As of 30 Oct 2026, 17:00 UTC · Binance BTC/USDT · closed candles
| Horizon | Model estimate | Change | Lower | Upper | Reference close | Target (UTC) |
|---|---|---|---|---|---|---|
| 24 hours | 76,826 | −0.68% | 74,792 | 77,892 | 77,353 | 31 Oct 2026, 17:00 UTC |
| 7 days | 76,595 | −1.32% | 73,458 | 79,704 | 77,622 | 6 Nov 2026, 16:00 UTC |
| 30 days | 78,335 | +0.44% | 68,262 | 96,871 | 77,992 | 29 Nov 2026, 00:00 UTC |
Prices in USDT. Experimental estimates, not guaranteed outcomes; the actual price may fall outside the range.
When the 24-hour forecast updates
The 24-hour forecast uses 1-hour candles and projects 24 steps ahead. It refreshes after each hourly candle closes, so the target time moves forward one hour at a time. An unfinished candle never enters the calculation. If Binance data is missing, out of order or stale, the forecast is shown as unavailable rather than filled with a guess.
For traders following Polymarket's hourly and daily Bitcoin Up or Down markets, note that those markets state Binance BTC/USDT candles as their resolution source. The same exchange pair feeds this forecast, but always read the rules of the specific market before trading. See how the Up or Down markets resolve.
How to read this 24-hour forecast
The model estimate is a single end price for the close 24 hours after the reference candle. The scenario range (lower and upper) is built from the model's own historical misses: the 10th and 90th percentiles of past forecast errors for this horizon, widened if needed so the estimate sits inside it. It describes how wrong the model has been before. It is not a calibrated 80% probability, and it is a range for the end price only, not for the highest or lowest price along the way.
Every figure is in USDT from Binance spot BTC/USDT, not USD. The forecast starts from the last closed 1-hour candle, so the reference close can differ from the live price you see elsewhere on the page. Times are UTC.
What goes into it
- Trend: linear-regression slopes of log prices over three look-back windows of 12, 48 and 144 1-hour candles.
- Trend consistency: R² of those fits, so a noisy trend counts for less than a clean one.
- Momentum over the recent candles.
- Realised volatility, which scales how far the estimate is allowed to move.
- Volume is shown as context only; it is not a fitted factor.
The model is a fixed, deterministic heuristic. It is not trained on your data, it does not read news or social media, and it has not established a predictive edge. Full details are on the methodology page.
How it has done historically
Each forecast is checked walk-forward: at up to 24 past points, the model is run using only the candles available at that moment, and its 24-hour estimate is compared with what actually happened. Target windows do not overlap. The current check for this horizon covers 24 past forecasts, with an average absolute error of 1.49% against 1.39% for simply assuming the price stays unchanged. These are retrospective checks, not a recorded live track record, and they do not guarantee future accuracy.
Other horizons
For a wider view, see the 7-day Bitcoin forecast (4-hour candles) and the 30-day forecast (daily candles). Each horizon is calculated separately and can point in a different direction.