Agile price forecast
Octopus Agile prices for South West England: the last 3 days as they really were, and our forecast for the next 7, with a range that widens as the future gets less certain.
Agile prices and forecast, South West England beta
p per kWh. left of now: real prices. right: our forecast
- Real prices (and tomorrow's once published)
- Forecast
- Likely range: 8 in 10 real prices land inside (measured)
| Day | Typical price | Likely range | Cheapest 3 hours | How good |
|---|---|---|---|---|
| Today | 40.9p | exact | 21:00–00:00 (34.9p) | Published |
| Tue 20 Oct | 38.2p | exact | 02:00–05:00 (32.6p) | Published |
| Wed 21 Oct | 37.0p | 29p–47p | 12:30–15:30 (29.8p) | Tested usually 4.8p out |
| Thu 22 Oct | 38.4p | 31p–49p | 00:00–03:00 (30.5p) | Tested usually 5.0p out |
| Fri 23 Oct | 32.9p | 25p–43p | 03:00–06:00 (25.8p) | Tested usually 5.1p out |
| Sat 24 Oct | 27.6p | 19p–38p | 03:30–06:30 (20.3p) | Tested usually 5.3p out |
| Sun 25 Oct | 22.0p | 13p–33p | 11:00–14:00 (11.5p) | Tested usually 5.8p out |
| Mon 26 Oct | 26.3p | 16p–38p | 00:00–03:00 (17.3p) | Tested usually 5.8p out |
Beta forecast by UK Electricity, not Octopus Energy, made Monday 19 October at 17:02. Built from NESO's wind and demand forecasts and Agile's recent prices. Real prices from Octopus Energy's public API. How it works and how accurate it is
Find your region
We look up your region with Octopus Energy and don't store your postcode.
How the forecast works
Agile prices follow the day-ahead wholesale market. That price is set mostly by how much electricity is still needed after the wind has done its share, by the time of day (the 4pm to 7pm peak costs most), and by the general level of prices, which tracks gas. We take NESO's forecasts of national demand and wind up to a week ahead, correct the wind forecast's known tendency to run high, and feed them into a model trained on the last 120 days of real Agile prices. Other regions follow London's price closely, so we convert using each region's own relationship with London's price.
How accurate it is, day by day
| Days ahead | Our forecast is usually out by | "Same as the last known day" | "Same as last week" | Better by | 80% range | Measured on |
|---|---|---|---|---|---|---|
| Tomorrow | 4.8p | 6.6p | 8.3p | 27% | −7p to +9p | back-test, 2026-08-20 to 2026-10-18 |
| 2 days | 4.8p | 8.4p | 8.3p | 43% | −7p to +8p | back-test, 2026-08-20 to 2026-10-18 |
| 3 days | 5.0p | 8.8p | 8.3p | 41% | −7p to +9p | back-test, 2026-08-20 to 2026-10-18 |
| 4 days | 5.1p | 8.8p | 8.3p | 39% | −7p to +9p | back-test, 2026-08-20 to 2026-10-18 |
| 5 days | 5.3p | 8.6p | 8.3p | 37% | −7p to +9p | back-test, 2026-08-20 to 2026-10-18 |
| 6 days | 5.8p | 8.7p | 8.3p | 31% | −8p to +10p | back-test, 2026-08-20 to 2026-10-18 |
| 7 days | 5.8p | 8.3p | 8.3p | 31% | −9p to +10p | back-test, 2026-08-20 to 2026-10-18 |
- "Usually out by" is the average gap between our forecast and the real price, in pence per kWh, London, before VAT. "Better by" compares it with the better of the two simple guesses.
- The shaded range on the graph is the measured 80% range for that many days ahead, so about 8 in 10 real prices land inside it, wider in the evening peak where misses are bigger.
- Each row switches from the back-test to our own live results once we have checked 300 half-hours at that distance (about a week). If a day ever stops beating the simple guesses, the graph hatches it as "outlook only".
- The back-test replays each day as it would have looked 1 to 7 days before, using ECMWF wind forecasts as they were issued at the time (from Open-Meteo's archive of past forecast runs, CC BY 4.0) and the prices known then. Demand used the real outturn, which is normally forecast within about 2%, so these figures are slightly flattering for days 2 to 7. The live forecast itself uses NESO's wind and demand forecasts.
How to use it
Treat tomorrow's line as a good guide and the days after as an outlook: which days look cheap, and which nights are worth waiting for. For the exact prices, check back after 4pm when Octopus publishes them. See today's cheapest times.
Common questions
Can you predict Agile prices?
Roughly, yes. Agile follows the day-ahead wholesale price, which depends mostly on how much demand is left after wind power, the time of day and the general price of gas. NESO publishes wind and demand forecasts up to 14 days ahead, so a model can estimate Agile before Octopus publishes it, with less certainty the further ahead you look.
How accurate is this Agile forecast?
Tested on the 60 days to 18 October 2026, predicting each day at 10am the day before using only the forecasts available then, it was 4.6p per kWh out on average, against 6.6p for simply assuming tomorrow will match today. The shaded range is set so that about 8 in 10 real prices fall inside it for tomorrow. Beyond tomorrow the range is deliberately wider and is being checked against real results as they come in.
When does Octopus publish tomorrow's Agile prices?
Usually around 4pm UK time. Once they are out, the graph shows them as real prices and the forecast starts from the day after.
Why does the shaded range get wider?
Because wind forecasts get less reliable the further ahead they look, and the general price level can drift. A wide range a week out is an honest sign that the exact price is not knowable yet; the typical price and the cheapest hours are still a useful guide for planning.
Is this from Octopus Energy?
No. The forecast is our own, made from public data. Real prices come from Octopus Energy's public API; UK Electricity is independent and not affiliated with Octopus Energy.