# Evidence-Gated Forecast Test: live evaluation brief

Generated at: 2026-09-18T12:21:04.227688+00:00
Protocol: evidence-gate-california-dry-run-v1
Mode: dry_run
Counts toward prospective claim: false
Pipeline status: attention
Open incidents: 0
Claim assessment: no_prospective_claim_permitted_in_current_dry_run

## Current descriptive result

Score revision: provisional
Scored days: 0
Scored events: 0
Mean IGPE: None
Relative factor exp(IGPE): None
Direction: not_available_no_scored_events
Warning: A positive live IGPE is descriptive until the frozen event gate and uncertainty criteria are met; few events are not proof that the gated forecast beats ETAS.

## Region results

- California RELM: events=0, IGPE=None, direction=not_available_no_scored_events, missed_days=0, current_missed_streak=0, longest_missed_streak=0, primary_eligible=True

## Metric

IG_i = ln(lambda_gated(i) / lambda_ETAS(i))
IGPE = (1/N) * sum_i IG_i
Positive IGPE favors the gated forecast; negative IGPE favors ETAS.

## Required evaluation discipline

- State whether the current protocol is a dry run or a claim-bearing prospective test.
- Report event count and scored-day coverage before interpreting IGPE.
- Separate provisional scores from catalog-settled final scores.
- Inspect each region; do not infer broad geographic generalization from only a pooled mean.
- Report missed days, invalidated regions, open incidents, and forecast freshness.
- Treat missed_region_days and longest_consecutive_missed_days as historical totals; current_consecutive_missed_days describes the active outage streak.
- Do not call the gated forecast superior without the pre-registered event gate and uncertainty analysis.
- Do not interpret these rate forecasts as exact earthquake time, location, or magnitude predictions.

## Sources

- Live JSON: https://etas2.bboga.com/api/evaluation.json
- Method: https://etas2.bboga.com/en/about.html
- Dashboard data: https://etas2.bboga.com/api/dashboard