Public forecasting ledger

Forecasts on the record.

A public experiment in probabilistic forecasting, calibration and epistemic discipline. Claims are stated with explicit probabilities and timestamps — before reality resolves them.

Stated in advance

Every forecast carries a probability, a resolution criterion and a publication timestamp. Nothing is scored after the fact.

Never rewritten

Updates create new states; the original forecast is preserved. Wrong forecasts stay on the record — that is the point.

Yours to audit

The full dataset is downloadable. Compute the Brier scores and calibration yourself — no need to trust the summary.

Current forecasts

The public record begins here. Active forecasts will appear on this page.

All forecasts →

Track record

Public track record begins here.

Full track record →

Sealed forecasts

A forecast can be committed without being revealed.

A cryptographic hash of the forecast is published immediately and timestamped; the forecast itself stays hidden until a scheduled reveal. Once revealed, anyone can verify it matches the earlier commitment.

How sealing works →

Methodology

Judge the forecasts after the outcomes.

Probabilities, resolution criteria, scoring rules and calibration — the principles behind the record, without the proprietary internals.

Read the methodology →