National Average

Model Parameter <1d
n=672
1d
n=672
2d
n=672
3d
n=631
4d
n=615
5d
n=615
6d
n=615
7d
n=615
8d
n=615
9d
n=615
10d
n=615
11d
n=615
12d
n=580
13d
n=236
AgilePredict MAE 0.08 2.16 4.27 4.67 4.48 4.44 4.50 4.68 4.63 4.80 4.65 4.33 4.22 5.14
AgilePredict RMSE 0.45 3.64 7.04 8.29 8.37 8.22 8.44 8.58 8.61 8.63 8.77 8.12 6.55 7.43
AgilePredict Bias +0.08 +1.19 +0.16 -0.26 -0.32 -0.08 -0.29 -0.85 -0.57 -0.44 -0.54 -0.32 -0.39 -1.03
Model Offset n MAE RMSE Bias
AgilePredict <1d 672 0.08 0.45 +0.08
AgilePredict 1d 672 2.16 3.64 +1.19
AgilePredict 2d 672 4.27 7.04 +0.16
AgilePredict 3d 631 4.67 8.29 -0.26
AgilePredict 4d 615 4.48 8.37 -0.32
AgilePredict 5d 615 4.44 8.22 -0.08
AgilePredict 6d 615 4.50 8.44 -0.29
AgilePredict 7d 615 4.68 8.58 -0.85
AgilePredict 8d 615 4.63 8.61 -0.57
AgilePredict 9d 615 4.80 8.63 -0.44
AgilePredict 10d 615 4.65 8.77 -0.54
AgilePredict 11d 615 4.33 8.12 -0.32
AgilePredict 12d 580 4.22 6.55 -0.39
AgilePredict 13d 236 5.14 7.43 -1.03
Showing 672 actual prices and 672 predictions for 1d ahead.

This chart compares the actual agile price with predictions made for the selected lead time.

The prediction offset is measured from when the forecast was created to the price slot being predicted.

MAE is the average absolute error, in p/kWh.

RMSE is similar but gives more weight to larger errors.

Bias is the average signed error, so positive means predictions were too high and negative means they were too low.