US electricity net generation · United States · 3 months ahead

2026-08: the model assigns an 80 percent probability to 416,085 to 444,099 million kilowatthours

Point estimate 420,242 million kilowatthours from Seasonal naive, against 354,691 million kilowatthours observed in 2026-05. seasonal_naive beats last_value on out-of-sample MAE by 79.2 percent over 45 cutoffs (minimum 24, margin 5 percent).

ModeledMetric page

Distribution over the history

Fan chart

Solid: observed (million kilowatthours). Dashed and shaded: model point estimates and 80 percent intervals from empirical residual quantiles. Filled dot: the record on this page. Green or red dot: the realised value, inside or outside its interval.

Quantiles: p5 410,151 · p10 416,085 · p25 421,627 · p50 431,353 · p75 436,756 · p90 444,099 · p95 449,915
Source: empirical quantiles of 36 in-window 3-step residuals, in log space; no normality assumed

The record

Immutable fields

Key: us-electricity-net-generation:USA:3
Id: e5913a6b-4c73-440a-bf74-3f6386c47e6b
Created: 2026-09-07 22:16 UTC
Data cutoff: 2026-09-07 22:16 UTC
Origin: 2026-05 = 354,691 (obs 245396)
Target period: 2026-08 (92 days)
Point: 420,242 million kilowatthours
Interval: 416,085 to 444,099 at 80 percent
Model: seasonal_naive models_v1.targets_v1
Baseline: last_value
Snapshot: 84da1a759387356a6376a2360ccde7d4dd66807b11b68b5cff11c41749247fd0
Data mode: ingested_at
Status: active

Drivers

Statistical association, not causal

Seasonal naive uses only the target's own history; there are no driver contributions.

Outcome

Not yet realised

The observation for 2026-08 has not been ingested. The weekly forecast.evaluate job attaches it when it arrives; the record above does not change.

Scenario assumptions

What the forecast takes for granted

No structural break in us-electricity-net-generation between the data cutoff and the target period.
The source keeps publishing the series on the same definition; a rebasing or redefinition invalidates the comparison.
The seasonal pattern of the last year repeats.

Falsifiers

What would show this forecast wrong

If the first published value of us-electricity-net-generation (USA) for 2026-08 is below 416,085 million kilowatthours or above 444,099 million kilowatthours, the 80 percent interval is falsified.
The model assigns a 10 percent probability to a value below 416,085 million kilowatthours and a 10 percent probability to a value above 444,099 million kilowatthours; interval misses should occur about one time in five over many forecasts, and a run of misses well above that rate falsifies the calibration.
The point forecast is above the last observed value (354,691 million kilowatthours); a realised value at or below that level falsifies the direction call.

Competing models

Scorecard for this target and horizon

Model	Status	Cutoffs	MAE	RMSE	Direction	80% coverage	Skill vs last value
Seasonal naive models_v1.targets_v1	champion	45	10,931	13,549	84%	89%	79.2%
Drift models_v1.targets_v1	challenger	45	52,650	65,260	53%	82%	-0.3%
Holt (exponential smoothing with additive trend) models_v1.targets_v1	challenger	45	56,069	69,669	20%	80%	-6.9%
Last value models_v1.targets_v1	baseline	45	52,472	64,853	0%	82%	0.0%
Linear trend models_v1.targets_v1	challenger	45	35,310	42,592	87%	78%	32.7%
Ridge regression on lags and drivers models_v1.targets_v1	challenger	45	45,020	58,589	60%	89%	14.2%

Model card

Seasonal naive

Purpose. Point and 80 percent interval forecasts of us-electricity-net-generation (USA) 1, 3, 6, 12 months ahead. Owner: Forecast Lab (HV 3.0).

Method

The forecast is the value observed in the same month one season (12 periods) earlier. The series is modelled in natural logarithms and transformed back, so intervals are asymmetric in level terms.

Uncertainty

Intervals are empirical: the model is refitted at earlier origins inside the training window, the realised errors at the same horizon are collected, and the interval is the 10th to 90th percentile of those errors around the point forecast. No normality is assumed.

Training and evaluation

Training window: 641 periods, 1973-01-01 to 2026-05-01. Evaluation: rolling-origin backtest over 45 cutoffs using pseudo real-time data (the earliest ingested value per period; no vintage archive exists for this series, so revisions between the cutoff and today are not reproduced).

Known limitations

Structural breaks after the cutoff are not modelled.
Driver contributions are statistical associations, not causal effects.
Promotion to champion requires beating the last-value baseline on out-of-sample MAE by at least 5 percent over the minimum number of cutoffs; otherwise the baseline stays champion.
Retraining: every scheduled forecast.run refits on the data then available; backtests re-run from /admin/forecasts.

Version chain

Supersedes

This is the first record for its key.