Live scores active — 10 states
MI · CA · NY · KY · GA · TX · WA · FL · LA · PA

Predict which ZIP codes are at risk before the storm hits.

Give your operations team ZIP-level advance notice to pre-position before the outage hits.

High-voltage transmission towers at dusk
Storm detection rate
94%
Aug 2021 derecho
Storm detection rate94%
Years of training data10 yrs
States covered10
Counties scored daily1,000+
The predictive pipeline

Advanced warning, backed by 10 years of federal outage data and machine learning.

01

EAGLE-I federal grid data

10 years of utility outage records from the DOE's authoritative power outage database — comprehensive ZIP-level coverage going back to 2014.

02

Live NWS weather alerts

Tornado warnings, severe thunderstorm alerts, high-wind advisories, NOAA Local Storm Reports, and drought indices — the same signals the model was trained on.

03

Risk scores refreshed hourly

The model re-scores 1,000+ counties across 10 states every hour and delivers results via REST API. Trained on 10 years of EAGLE-I federal outage data using 28 weather, grid, and infrastructure signals.

Built for grid stakeholders

Intelligence for everyone who depends on the grid.

Emergency managers, utility operators, and insurance analysts all make better decisions with advance outage lead time. FORESCORE provides the same prediction layer to all three.

EM

Emergency Management

NWS tells you a storm is approaching. FORESCORE tells you which ZIP codes will actually lose power — enabling advance pre-positioning of generators, shelters, and first responders.

UTL

Regional Utilities

Every crew pre-positioned in the wrong area multiplies response costs. FORESCORE predicts which counties need pre-staging before damage occurs — from distribution feeders to transmission lines.

INS

P&C Insurance

Historical risk scores for any ZIP code and date, correlated against EAGLE-I outage records. Built for claims validation, catastrophe underwriting, and grid-failure peril modeling.

County-level outage risk heatmap across the US

Predictive heatmap coming Fall/Winter 2026

Accuracy

On Michigan's worst storm of 2021 — a fast-moving derecho across 72 counties — FORESCORE detected 94% of affected areas before the event peaked.

Backtested against DOE EAGLE-I federal outage records. AUC 0.705–0.847 across six utility territories. No customer data, no SCADA feeds used.
94%
Storm detection rate
0.847
Top territory AUC
0.705
Avg territory AUC
Pricing

Start predicting outages in your region today.

Start with a Proof of Value Audit. We run our model against your territory and past events and deliver an accuracy report.

Enterprise
Starter
1 state
  • Hourly risk scores via REST API
  • Live outage status per county
  • 30-day history
  • 500,000 county queries/month

MI · CA · NY · KY · GA · TX · WA · FL · LA · PA

Regional
Up to 5 states
  • Everything in Starter
  • Alert webhooks (beta)
  • 1-year risk history · 90-day outage history
  • 2M county queries/month
National
All 10 states
  • Everything in Regional
  • Unlimited county queries
  • Bulk data export (CSV/Parquet)
  • All 52 states — coming soon

Request FORESCORE API access. Start predicting.

We'll reach out to provision your API key.

We'll be in touch soon