Aegean Intelligence Group
BRIER SYSTEMS BY AEIG / RECORD INTEGRITYAEIG-WEB-BRIER-INTEGRITY-v1.0 · A DIVISION OF AEGEAN INTELLIGENCE GROUP LLC · 2026
BRIER SYSTEMS · RECORD INTEGRITY

The score can be checked without taking the firm on trust.

This is the low-prominence audit surface behind the BRIER Scorecard: public, machine-readable and absent from the site homepage. It records what was locked when, what counts toward the headline, what did not qualify, and the exact bytes behind each ledger version.

01 · CURRENT FINGERPRINT

The public ledger, fixed by its contents.

Snapshot 2026-09-11-10b09ed59c29 is an exact copy of the public claim ledger. Its SHA-256 fingerprint is:

10b09ed59c29fd2461fc09081d9020a33403c8ae28cd78cbd1d044e15f5c0491

Open this ledger snapshot · Open the machine-readable integrity manifest

02 · RECORD CLASSES

The headline score begins with native forward registration.

Only probability claims locked no later than publication enter the headline Brier score. Earlier published judgments entered before their windows closed remain visible as legacy prospective claims and score separately. Entries made after their windows closed remain visible for audit and never enter the headline score.

Record classPublic claims
brier-native40
legacy-prospective33
legacy-retrospective4

The forward benchmark may score publicly, but it is a holdout: it never trains the Learning Loop or changes a prior.

03 · SELECTION ACCOUNT

What did not qualify is counted.

284 candidate statements were reviewed. 77 entered the public record and 207 did not qualify. Candidate text remains private because it includes extraction working material; the disposition totals are public so selection cannot disappear from the denominator.

ReasonCount
duplicate or overlapping3
no probability21
no unambiguous window67
not publicly resolvable8
scenario construction not probability108
04 · DEPENDENCE AND QUALITY

Raw claims are not represented as independent trials.

The public record currently contains 42 conservative claim families. A family defaults to the source report until a narrower event-level family is supplied. Future scorecards can therefore report both raw n and family-adjusted evidence without pretending that several claims driven by one event are several independent tests.

CheckCountClaims
missing settlement specification0none
missing claim family0none
locked after resolution window4CLM-2026-083, CLM-2026-157, CLM-2026-187, CLM-2026-213

These checks are generated from the same ledger as the Scorecard. A non-zero count stays visible until corrected; it does not silently disappear.