Public forecast documentation
Forecast calibration evidence
Calibration asks whether stated probabilities and intervals agree with observed frequencies. It is not, by itself, proof of useful predictive skill.
Historical audit snapshot created 2026-09-10 — not a live/current forecast
Calibration is reliability, not directional correctness
A probabilistic forecast is calibrated when events assigned a stated probability occur at roughly that frequency over a suitable collection of comparable forecasts. Calibration asks whether probabilities or intervals are reliable; it does not ask only whether a point direction happened to be right.
Directional correctness, Brier score, price error, and interval coverage answer different questions. A forecast can have acceptable coverage because its intervals are broad, or have directional hits without well-calibrated probabilities.
What the dated evidence supports
The September 10, 2026 audit reports a direction Brier score of 0.250491 against the constant 50% reference of 0.25, and 95% observed coverage for nominal 80% bands. These are aggregate observations, not a complete calibration analysis.
For score meaning and the 0.25 reference, use the direction Brier-score interpretation. For nominal versus observed interval coverage, use the confidence-band coverage interpretation.
The forecast methodology page provides the shared cadence, overlap, probability, resolution, and evidence-boundary definitions.
Evidence not yet available
No probability bins, observed frequencies by bin, reliability curve, interval-width distribution, or horizon-level coverage were supplied, and the observed target span is not a prespecified evaluation window. It would be inaccurate to call the current material a complete calibration study or to treat either aggregate as proof of predictive skill.
Data provenance and export status
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.
Citation-ready public export status: NOT READY. The retained historical evidence contains 56 total SOL forecast rows and 40 resolved rows, with resolved horizon counts of 6h n=13, 12h n=12, 24h n=9, 48h n=6 and observed target span 2026-09-07T07:14:30Z–2026-09-10T10:59:20Z. These are point-in-time observations, not an immutable issued-and-resolved archive, complete history, or prespecified evaluation window. A current public feed may expose rows, but it is not an immutable versioned research archive. Metrics calculated from a changing feed are not automatically comparable to this dated snapshot. A research export would need versioned records with forecast ID, asset and venue, UTC issue and target timestamps, horizon, point prediction, interval bounds, direction probability, model version, outcome source, observed price, resolved-at time, scoring-rule version, and content hash. No research export is available from these pages.
The existing Dataset JSON-LD preserves the audit date rather than implying freshness.