Rolling evidence
Model Scoreboard
Rolling forecast quality before model promotion.
Score rows appear only after a forecast was committed before the delivery date and official OREE rows later became available for scoring.
Forecast scoring
MAE/RMSE are price metrics; dispatch regret and value capture connect forecasts to BESS economics.
| Model | Window | MAE | RMSE | Dispatch regret | Value capture | Boundary |
|---|---|---|---|---|---|---|
| No scored forecast pairs yet. | ||||||
Promotion ladder
Deterministic index, public forecasts, rolling scores, schedule selector, then research challengers.
- Stage 0Realized deterministic index
- Stage 1NBEATSx/TFT forecast challenge
- Stage 230+ scored delivery days
- Stage 3Schedule-selection backtest
- Stage 4V2+ optimization candidate
- Stage 5DT/HF DT gated challenger
Research boundary
DT/HF DT evidence can be public only as a challenger to V2+ after enough source-backed score history.
- Market executionfalse
- Proposed bidsnot emitted
- External EMS integrationnot claimed
- Default modelnot DT/HF DT