Know tomorrow's solar and wind output today.
Predigrid trains a model on each asset's production history and blends it with numerical weather predictions — delivering probabilistic forecasts from 15 minutes to 10 days ahead.
History teaches the model your asset. Weather tells it what comes next.
Connect production history
Upload meter or SCADA exports, or connect your data platform. Twelve months of 15-minute data is enough to start; more improves seasonal accuracy.
Blend weather models
Irradiance, cloud cover, wind speed at hub height and temperature from several numerical weather prediction models are weighted per site and updated every run.
Receive probabilistic forecasts
P10, P50 and P90 output per interval — in the dashboard, via REST API or as scheduled CSV delivery to your trading and dispatch systems.
One forecasting layer for the whole portfolio
{{ f.title }}
{{ f.body }}
From the next quarter-hour to next week
Each horizon runs its own model tuned to the decision it supports — balancing, intraday trading, day-ahead bidding or maintenance planning. Accuracy is reported against your own measured output, not a benchmark site.
Typical normalised mean absolute error across the reference portfolio, 2025.
Everyone who has to commit to a number
Asset owners & IPPs
Reduce imbalance costs, meet grid-code forecast obligations and plan maintenance in low-yield windows.
Energy traders
Bid day-ahead and intraday with quantiles, not a single line — and size positions to the uncertainty.
Grid & system operators
Aggregate forecasts by node or region to anticipate ramps, congestion and reserve needs.
Send 12 months of production data. Get a backtest within five working days.
Compared against your current forecast provider, interval by interval. No commitment.