US Power Demand Forecasts

Hourly US electricity demand actuals and day-ahead demand forecasts for 12 balancing authorities, benchmarked against EIA's own day-ahead forecast.

Hourly US electricity demand (actuals) and day-ahead demand forecasts for 12 balancing authorities, benchmarked against EIA’s own day-ahead forecast. Available as a data product on the Snowflake Marketplace.

12 / 12BAs beat EIA day-ahead
+37.3ppbest margin (PSEI)
+9.1ppaverage margin
Dailyrefresh, 90-day rolling window

Coverage

PSEI PSCO FPC SWPW BHBA GVL SWPP PACE CISO NEVP AZPS PJM

Selected because Aclastic’s forecast has beaten EIA’s day-ahead forecast (lower WMAPE) on a 90-day trailing basis for every BA in this set.

Tables

US_POWER_DEMAND_ACTUALS

Hourly actual demand per BA, rolling 90-day window, refreshed daily.

ColumnTypeDescription
balancing_authorityVARCHAREIA BA code
ba_nameVARCHARFull BA name
regionVARCHARNERC region
utc_timeTIMESTAMP_NTZHour start, UTC
local_timeTIMESTAMP_NTZHour start, BA local time
demand_mwhFLOATActual hourly demand (MWh)

US_POWER_DEMAND_FORECASTS

Day-ahead forecast per BA, rolling 90-day window, refreshed daily.

ColumnTypeDescription
balancing_authorityVARCHAREIA BA code
ba_nameVARCHARFull BA name
regionVARCHARNERC region
target_utc_timeTIMESTAMP_NTZHour being forecast, UTC
target_local_timeTIMESTAMP_NTZHour being forecast, BA local time
prediction_run_atTIMESTAMP_NTZWhen the forecast was issued
horizon_hoursFLOATLead time from issuance to target hour
aclastic_forecast_mwhFLOATAclastic’s forecast
eia_forecast_mwhFLOATEIA’s published day-ahead forecast, for benchmarking

Accuracy (90-day trailing, day-level WMAPE)

BAAclasticEIAMargin
PSEI3.08%40.41%+37.34pp
PSCO6.60%25.14%+18.54pp
FPC5.36%23.78%+18.43pp
SWPW3.74%13.98%+10.24pp
BHBA2.67%8.85%+6.18pp
GVL6.53%11.24%+4.71pp
NEVP5.29%8.41%+3.12pp
AZPS6.34%9.12%+2.78pp
SWPP3.84%6.52%+2.68pp
PACE5.55%8.20%+2.65pp
CISO6.25%8.35%+2.10pp
PJM3.16%3.53%+0.36pp

Example queries

Latest actual demand for a BA

SELECT ba_name, utc_time, local_time, demand_mwh
FROM us_power_demand_actuals
WHERE balancing_authority = 'CISO'
ORDER BY utc_time DESC
LIMIT 24;

Aclastic’s next-24h forecast vs EIA’s

SELECT ba_name, target_local_time, aclastic_forecast_mwh, eia_forecast_mwh
FROM us_power_demand_forecasts
WHERE balancing_authority = 'PJM'
  AND target_utc_time > CURRENT_TIMESTAMP()
ORDER BY target_utc_time
LIMIT 24;

Accuracy comparison by BA, last 30 days

SELECT
    f.balancing_authority,
    AVG(ABS(f.aclastic_forecast_mwh - a.demand_mwh) / NULLIF(a.demand_mwh, 0)) * 100 AS aclastic_mape,
    AVG(ABS(f.eia_forecast_mwh      - a.demand_mwh) / NULLIF(a.demand_mwh, 0)) * 100 AS eia_mape
FROM us_power_demand_forecasts f
JOIN us_power_demand_actuals a
  ON f.balancing_authority = a.balancing_authority
 AND f.target_utc_time     = a.utc_time
WHERE f.target_utc_time >= DATEADD(day, -30, CURRENT_TIMESTAMP())
  AND a.demand_mwh <> 0
GROUP BY f.balancing_authority
ORDER BY aclastic_mape;

Daily demand by region

SELECT region, DATE_TRUNC('day', utc_time) AS day, SUM(demand_mwh) AS total_mwh
FROM us_power_demand_actuals
GROUP BY region, day
ORDER BY day DESC, region;

Update cadence

Both tables refresh daily — new actuals and forecast rows appended, rolling 90-day window maintained.

Questions

Contact sales@aclastic.ai.