Forecast method and evaluation

Forecast publication criteria

Probabilities are published only when continuous coverage and verified effective times meet the requirements.

• At least 10 precisely timed global resets

  0 records currently qualify.

• Continuous, traceable historical coverage

  Verify each effective time and disclose gaps in coverage.

• Preserve each forecast and its outcome

  Published probabilities stay unchanged and are evaluated after 24 and 48 hours.

View historical intervals

Forecast outcomes over the last 30 days

No valid forecast has been published yet, so no Brier score or calibrated accuracy is available. Published forecasts will preserve their timestamp, model version and outcome window.

Evaluation is limited to periods with maintainer-declared continuous coverage and uses currently verified records. Hourly forecast windows overlap, so samples are not independent.