April 2026 Forecast Accuracy Report

Standing benchmark memo for trading desks, operators, and procurement teams tracking whether Gramm still justifies a switch from the feed they use today.

This issue in numbers

Regions in public snapshot
7

Publicly benchmarked market regions in the current issue.

Active maximum horizon
15-day

Active forecast coverage spans day-ahead through 15-day runs.

Flagship benchmark move
39.6%

Relative California benchmark reduction from 4.8% to 2.9% MAPE.

California remains the cleanest public signal for a side-by-side switch decision.

ERCOT still deserves more attention on the bad days than on the average day.

Day-ahead through 15-day coverage now sits on one delivery surface.

Territory benchmark snapshot

California
CAISO · PG&E, SCE, SDG&E
Baseline
4.8%
Gramm
2.9%
Reduction
39.6%
Texas
ERCOT · Oncor, CenterPoint, AEP Texas
Baseline
5.1%
Gramm
3.4%
Reduction
33.3%
Mid-Atlantic & Midwest
PJM · PECO, ComEd, Dominion
Baseline
3.9%
Gramm
2.6%
Reduction
33.3%
Central U.S.
MISO · Entergy, Xcel, DTE
Baseline
4.3%
Gramm
2.9%
Reduction
32.6%
New York
NYISO · Con Edison, National Grid
Baseline
4.1%
Gramm
2.8%
Reduction
31.7%
New England
ISO-NE · Eversource, National Grid
Baseline
3.7%
Gramm
2.5%
Reduction
32.4%
Great Plains
SPP · AEP, OGE, Evergy
Baseline
4.6%
Gramm
3.1%
Reduction
32.6%

California is still the easiest place to start

The California day-ahead benchmark remains the simplest public read on whether Gramm is materially better than the incumbent baseline.

Read benchmark memo

In Texas, the bad days matter most

ERCOT buyers care less about average days than about the days when weather and price move together. Large misses are where load forecasting gets expensive.

Request market backtest

One API now covers tomorrow and the rest of the planning window

7-day and 15-day runs now sit on the same API surface as the short-horizon forecast, which matters for teams trying to avoid vendor handoffs.

Review API docs

Why teams keep this report open

Trading desks use it to decide whether the public result still justifies a side-by-side test.
Operators use it to see where lower load error is likely to change day-ahead decisions.
Procurement teams use it as a standing reference alongside security, SLA, and status pages.