If Flatpeak is temporarily unable to fetch tariff data—due to provider outages or delays from market operators—it will automatically retry until the data becomes available. During these times, gaps may appear in tariff rates. To maintain continuity, Flatpeak uses machine learning and AI to populate gaps and predict future rates. When this happens:
  • The API returns predicted values with a confidence below 1.
  • Once the original data source becomes available, Flatpeak back-fills the missing periods and updates the confidence to 1.
This ensures uninterrupted operation, even when upstream data is temporarily unavailable.

​
How tariff prediction works

Below is a simplified overview of how Flatpeak estimates tariff rates, depending on the tariff and the type of segment that is missing:

​
No tariff data available

  • FIXED and TIME_OF_DAY tariffs:
    Flatpeak estimates rates starting with 0.8 confidence, decreasing by 0.2 every 90 days, down to a minimum of 0.2.
  • DYNAMIC tariffs where market rates are available:
    Predictions are based on market rates (where available), weather forecasts, and past pricing. Values are returned with approximately 0.6 confidence.
  • MARKET or DYNAMIC tariffs where market rates are not available:
    A zero value is returned with 0 confidence. (Note: predictions for these cases are expected in Q4 2025.)

​
Gaps in tariff data

  • FIXED and TIME_OF_DAY tariffs:
    Missing rates are interpolated based on the existing tariff, using the lowest confidence of the surrounding time periods.
  • MARKET tariffs:
    Missing rates are interpolated using the lowest confidence of surrounding periods, and extended historical data for periods with similar weather conditions.
  • DYNAMIC tariffs:
    Market rates are returned with 0.6 confidence during the gap.