Skip to main content

Spot Price Forecast smart home integration, free and open source

Spot Price Forecast integration for Gladys Assistant

Forecast Nord Pool spot electricity prices for Finland, a week ahead, to plan appliances.

Forecast spot electricity prices for the days after the published ones, to plan the washing machine, the water heater or the car charging a week ahead.

The forecast comes from nordpool-predict-fi by vividfog (MIT license), an open machine-learning model that predicts the Nord Pool Finland day-ahead prices hour by hour, about a week ahead, from weather forecasts, wind power, nuclear availability and more.

Developed without a Finnish electricity contract to compare with: feedback welcome. Not affiliated with Nord Pool nor with the author of the model.

Overview​

  • One device, Spot price forecast Finland: the forecast price of the current hour, the average, lowest and highest price of the next 24 hours and of the next 7 days, the start and average price of the cheapest window of consecutive hours, when the forecast was updated and how far it goes.
  • A dashboard widget: the forecast curve of the next 2, 4 or 7 days.
  • Two scene triggers: the cheapest forecast hours begin, the forecast price drops below a threshold.
  • One scene action: find the cheapest forecast hours within a given time.

Prices are in €/kWh, Finnish VAT included (the model's unit, euro cents, divided by 100), without your supplier's margin nor the grid fees.

This is a forecast, not the published prices: Nord Pool publishes tomorrow's real prices around 14:00 (Helsinki). For them, use a Nord Pool or ENTSO-E integration next to this one.

Requirements​

  • Gladys Assistant 5.3.0 or later.
  • Internet access from Gladys: the forecast is a file on GitHub. No account, no key.

Installation​

  1. In Gladys, open Integrations → Install from GitHub.
  2. Enter https://github.com/guim31/gladys-spot-forecast.
  3. Open the Discovery tab of the integration and add the device.
  4. Add the Spot price forecast widget to a dashboard.

Configuration​

  • Length of the cheapest window (1 to 8 hours, 3 by default) and Search the cheapest window in (the next 12 hours to 7 days, 24 hours by default): the window the device shows.
  • Language of the feature names (French by default): used when the device is added; the names do not change afterwards.

The forecast device​

FeatureMeaning
Forecast price now (€/kWh)the forecast price of the hour in progress
Average / Lowest / Highest next 24 hover the 24 hours from the current hour
Average / Lowest / Highest next 7 daysover the 168 hours from the current hour
Start of the cheapest windowweekday and time (Helsinki), e.g. Tue 03:00
Cheapest window average (€/kWh)the average forecast price of that window
Forecast updated / Forecast untilwhen the model last published, and the end of its last hour

Times are in the Helsinki time zone, the clock of the Finnish price hours.

Dashboard widget​

The forecast curve (one step per hour) of the next 2, 4 or 7 days, with its lowest and highest hours marked; three tiles (now, average of the next 24 hours and of the next 7 days); the cheapest window, the update time and the end of the forecast. The tiles and the status lines can be hidden in the widget's settings.

Scene triggers​

Cheapest forecast hours begin​

Fires when the cheapest run of N consecutive hours (1 to 8) of the next 12, 24 or 48 hours begins: nothing cheaper is forecast before that horizon. Then the same choice stays silent until the horizon is over: with prices falling for a whole day, the cheapest window would otherwise start "now" every hour. Variables: average price, start and end (HH:MM).

The forecast price drops below a threshold​

Fires at the start of a run of hours under the threshold (0 to 0.20 €/kWh, VAT included), when the hour before was above it. Variables: the price of the hour, the number of hours under the threshold, start and end (HH:MM).

Scene action​

Find the cheapest forecast hours: the cheapest run of N consecutive hours (1 to 8) from now, within the next 6 hours to 7 days. Outputs: found, starts now, start date (YYYY-MM-DD), start and end (HH:MM), average price, minutes until the start. With no forecast known (the source unreachable since the start), the action fails in red.

Scene examples​

  • Water heater: trigger "Cheapest forecast hours begin" (3 hours, nothing cheaper in the next 24 hours) → turn the water heater's plug on; a second scene turns it off three hours later.
  • Car charging: every evening at 20:00, action "Find the cheapest forecast hours" (4 hours within 48 hours) → send a message "charge on {start_date} at {start_time}".
  • Negative prices: trigger "The forecast price drops below a threshold" (0 €/kWh) → notify.

Refresh and source usage​

The file is read every 30 minutes with If-None-Match: when it has not changed, GitHub answers 304 without content. When it changed, one call to the GitHub API (60 calls an hour per address without token) gives the time of its last commit; if GitHub refuses, the time the integration saw the change is shown instead. After a failure, the next tries come after 5, 10, 20, 40 then 60 minutes; the last forecast read stays in use meanwhile. The model is rerun a few times a day.

Troubleshooting​

  • The integration shows an error status: the forecast cannot be read (GitHub unreachable), or it covers less than a day ahead (the model stopped publishing). The last forecast read stays in use.
  • No state on the device: add the device from the Discovery tab, then use the Read the forecast now action.
  • Full detail in the integration logs, with LOG_LEVEL=debug.

Limitations​

  • Finland only for now: no open forecast without key was found for Sweden or Germany (November 2026). Other zones can be added as devices when one exists.
  • A forecast is not a price: expect errors of several cents, more on windy or very cold days. The model's site shows its past accuracy.
  • Hourly forecast, while the Nord Pool market trades in 15-minute periods since October 2025.
  • The forecast depends on a hobby project hosted on GitHub: it may stop.

Configuration settings​

These are the settings Spot Price Forecast asks for in its configuration screen in Gladys.

SettingTypeRequiredDescription
Spot price forecastsectionNoThe open model nordpool-predict-fi (vividfog, MIT) forecasts the Nord Pool Finland day-ahead prices hour by hour, about a week ahead, VAT included, without account or key. A forecast, not the published prices: for those, use a Nord Pool or ENTSO-E integration. Add the device from the Discovery tab and the widget from the dashboard. Not affiliated with Nord Pool.
Length of the cheapest windowselectNoThe device shows the start of the cheapest run of this many consecutive hours.
Search the cheapest window inselectNoFrom the current hour.
Language of the feature namesselectNoUsed when the device is added to Gladys: the names do not change afterwards.

How to install Spot Price Forecast in Gladys​

  1. In Gladys, open Integrations: Spot Price Forecast appears in the catalog, next to the native integrations, with a community badge.
  2. Click Install. Gladys pulls the Docker image (ghcr.io/guim31/gladys-spot-forecast:1.0.0), starts it in a sandbox isolated from the core, and generates the integration's interface (devices, discovery and configuration).
  3. Open the Configuration screen of the integration, fill in the settings, and save.
  4. You can also install it directly from its repository URL: https://github.com/guim31/gladys-spot-forecast.

Spot Price Forecast requires Gladys >=5.3.0. The catalog inside Gladys refreshes every hour, so a new version becomes available at most one hour after its release.

Not running Gladys yet? It is free and open source: follow the installation guide to get started.

About external integrations​

Spot Price Forecast is an external integration: a community integration packaged as a Docker container and published on GitHub, that Gladys installs in one click and runs in a sandbox isolated from its core. It is published and maintained by guim31, not by the Gladys core team.

Subscribe to the Gladys Assistant newsletter

A few emails per month about new releases and project news. Sent by Pierre-Gilles Leymarie, founder of the project. Unsubscribe anytime 🙂