Public forecast documentation
Forecast accuracy: evidence and limits
The audit supplies two aggregate statistics, but not enough row-level outcomes to independently reproduce accuracy by horizon or market regime.
Historical audit snapshot created 2026-09-10 — not a live/current forecast
| Metric | Observed | Reference | Careful interpretation |
|---|---|---|---|
| Direction Brier score | 0.250491 | 0.25 for a constant 50% probability | Slightly worse than the no-skill 50% baseline; no demonstrated directional skill. |
| Nominal 80% band coverage | 95% | 80% nominal coverage | Higher coverage may indicate bands were overly wide; this is not 95% predictive accuracy. |
What cannot be concluded
These aggregates do not establish profitability, predictive superiority, statistical significance, or performance on non-overlapping holdout data. Row-level observations and sample counts by horizon are needed for independent reproduction.
Data provenance and recommended public export
The statistics shown here come only from the forecast audit snapshot created 2026-09-10. Current values remain on the external live product. The local scheduled-worker forecast table is a distinct implementation and is not represented as that external production feed.
Recommended source: a versioned, downloadable JSON and CSV dataset at the live product, with immutable issued and resolved rows, UTC timestamps, source venue, sample count n, scoring formulas, model version, and a content hash. The Dataset JSON-LD on this page preserves the audit date rather than implying freshness.