> ## Documentation Index
> Fetch the complete documentation index at: https://docs.flatpeak.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Price forecasts
Price forecasts are only available in [Anode](/api-reference/anode/release-notes) and later versions. Check your current version in [Dashboard](https://dashboard.flatpeak.com/settings/versions).
Sometimes you need to know the energy price **before** it’s available.
In most European markets, day-ahead electricity prices are published around 11:00 a.m. This means your systems may not always have a full 24-hour price horizon available when planning runs.
Price gaps can also occur due to upstream outages or temporary issues in energy provider systems.
To ensure continuous operation, Flatpeak applies machine learning to:
* Forecast future prices
* Fill historical gaps
When this happens:
* The API returns forecasted prices with `confidence < 1.0`
* Once upstream data becomes available, Flatpeak back-filles prices and sets `confidence = 1.0`
## How tariff forecasts work
Flatpeak’s forecasting logic adapts by data source, tariff type and time horizon. Below is a simplified overview.
### Gaps in data
| Tariff Type | Handling | Confidence |
| --------------------- | ------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------- |
| `FIXED`/`TIME_OF_DAY` | Missing intervals are interpolated from adjacent prices. | The lowest confidence of the surrounding time blocks is applied. Value will not exceed `0.9` |
| `MARKET`/`DYNAMIC` | Missing intervals are reconstructed using extended historical data for the same period, including weather signals. | Confidence is derived from historical and contextual inputs. Value will not exceed `0.8` |
### Future data not yet available
| Tariff Type | Handling | Confidence |
| --------------------- | --------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `FIXED`/`TIME_OF_DAY` | Known pricing structures are extrapolated forward. | Starts at `0.8` and decays by `0.2` every 90 days, to a minimum of `0.2`. |
| `MARKET`/`DYNAMIC` | Prices are forecast using market prices, provider surcharges, weather forecasts and historical behaviour. | Forecast confidence never exceeds `0.75`.
When prices are published while FX markets are closed, such as weekends, the data is returned with a confidence of `0.85`, even if the underlying energy price has a confidence of `1`. |
## Historic performance
Expand the sections below to explore Flatpeak forecast accuracy vs actual published market prices, by country (Europe). MAE = Mean Absolute Error.
Forecasts for the US, Australia, and New Zealand are coming in Q1 2026.
Need more detail on Flatpeak’s energy price data or forecasts? [Get in touch with support](https://dashboard.flatpeak.com/#support).