The Brier ledger is Aequara's public accountability surface. Every published prediction registers as a Brier prediction with a stated horizon. At horizon resolution, the prediction is scored, the error is published, and the running cohort calibration is updated. We earn trust by making mistakes legible. The alternative is the rating-agency model that pretends predictions are descriptions.
A sample of the last twelve resolved predictions across NPX-100 and sub-indices. Trustee Tier members access the full queryable ledger via Cipher.
The Brier score (Glenn Brier, 1950) is the mean squared error between forecast probability and realized outcome. 0.0 is perfect; 0.25 is what you'd score by always predicting 50/50. Aequara's 0.143 is materially better than naive — but the score alone is incomplete. The discipline is in the calibration: are confident predictions actually right at the rate we claimed?
Look at the ledger above. The 0.99-confidence prediction on issuer-pay revenue resolved true. The 0.46-confidence on sector decoupling resolved miss — and that's correct behavior: when we're uncertain we say so, and uncertain predictions miss frequently. A calibrated forecaster's miss rate matches their stated confidence; an overconfident one's doesn't. Read the ledger that way.
Every cohort. Every cycle. The math is open. If our 0.85-confidence predictions don't resolve true at ≥85% rate, our calibration claim is wrong — and you should be able to verify that without taking our word for it.
Each prediction is signed by a named steward. Trustee Tier sees per-steward Brier histograms. Stewards who calibrate poorly over multiple cycles rotate out of decision authority.
Use Aequara's published predictions as priors for your own analysis. Or use them as a calibration baseline against your own forecasting discipline. The data is public. The methodology is documented.
The 12 rows above are a sample. The full ledger contains every published prediction Aequara has ever made — by index, by steward, by horizon, by confidence band — with cohort statistics computed live. Public read, Trustee write.