Methodology · GAI
Grid Availability Index
Composite 0–100 score measuring near-term ability of a sub-region to absorb new large load without triggering network upgrades, queue restudy, or capacity scarcity pricing.
Inputs
Source feeds and weights.
| Feed | Class | Weight |
|---|---|---|
| ISO interconnection queue (active MW vs cluster cycle) | MODELED INTELLIGENCE | 30% |
| Substation ATC + planned network upgrades (FERC 715) | MODELED INTELLIGENCE | 25% |
| Capacity auction clearing trajectory | LIVE PUBLIC DATA | 20% |
| EIA-930 hourly balancing area load shape | LIVE PUBLIC DATA | 15% |
| Utility IRP signal & rate-case posture | MODELED INTELLIGENCE | 10% |
Model
Formula and plain English.
Formula
GAI = 100 − w₁·QueueDepth − w₂·HeadroomDeficit − w₃·CapacityPriceΔ − w₄·LoadShapeStress − w₅·UtilitySignal
In plain English
Start at 100, subtract weighted penalties for crowded queues, low substation headroom, rising capacity prices, peakier load shapes, and constrained utility posture.
Operations
Cadence and confidence.
Refresh cadence
Daily composite
Confidence interval
Reported with ±6 point P50/P90 band on backtested 18-month horizon.
Limitations
What this index does not do.
- L/01Backtested against post-2022 PJM, MISO, ERCOT prints; earlier regimes are out-of-distribution.
- L/02Sub-region resolution is balancing-area, not substation. Site-level reads require engagement.
- L/03Does not price political risk to specific projects.
Change log
Recent calibration history.
2026-Q2
Recalibrated queue-depth weighting to reflect ISO cluster reform timelines.
2026-Q1
Added EIA-930 load shape stress term.