Forecast accuracy
How close each forecast on Ask the Grid came to what actually happened over the last 24 hours, measured against each market's own settled data. Lower is better throughout.
Price
The average gap to the real-time price at each market's trading hubs, hour by hour. The day-ahead auction sets its prices the day before delivery.
Demand
Each grid operator's own demand forecast against metered demand, as it stood 24 hours ago.
Wind
Each grid operator's wind forecast against metered wind output, using the most recent forecast for each interval.
Solar
Each grid operator's solar forecast against metered solar output, using the most recent forecast for each interval.
Grid conditions
The demand forecast behind each operator's supply outlook against metered demand, using the most recent forecast for each interval.
| Market | Error% of actual | Avg missMW | BiasMW |
|---|---|---|---|
| MISO | 1.5% | 1,112 | 1,080 low |
Also published, not scored
How scoring works
- Error
- The total miss divided by the total actual, as a percentage. A large grid and a small one compare fairly, and intervals near zero, such as solar at night, cannot inflate it.
- Avg miss
- The average absolute gap between forecast and actual, in the grid's own unit.
- Bias
- The average signed gap. High means the forecast ran above what settled; low means below.
- Which forecast
- Demand forecasts are scored as they stood 24 hours ago. Wind, solar and grid-conditions forecasts use the most recent forecast for each interval, which has had later information. Prices compare hourly averages of the real-time price with the day-ahead auction, cleared the day before delivery, and with each model's earliest run for the hour.
- Intervals
- An interval counts once when a forecast and a settled value both exist for it, and forecasts in the same comparison are scored on the same intervals. Most are hours; price intervals are counted per trading hub.