Criterica Group — The institutional data science platform for regulated outcomes. A Splitifi company.
Glossary

Governance Artifact

A non-predictive component of a model registry, a reference table, a retired experiment, a quality-control check, distinct from a production predictive model. Citing governance artifacts and production models under a single undifferentiated total overstates the size of the genuinely predictive fleet.

← All Terms

A governance artifact is a component of a model registry that supports the institution's modeling operation without itself generating a prediction used to price capital: a reference table encoding a jurisdiction's procedural taxonomy, a retired experiment kept for institutional memory rather than active use, or a quality-control check that validates data quality without producing an outcome probability of its own. Governance artifacts are a legitimate and often necessary part of a well-run modeling operation, but they are a different category of thing than a production predictive model, and an institution that cites its governance artifacts and its production predictive models under a single undifferentiated models total is presenting a number that overstates the size of its genuinely predictive fleet, sometimes by a substantial multiple. This platform discloses its production predictive model count and its governance artifact count as separate, distinct figures specifically to avoid this overstatement, since citing the combined total as though it represented predictive capability alone would materially mislead a diligence-minded reader, even though every component of the combined total may be, on its own terms, an entirely legitimate and necessary part of the registry that no institution needs to hide or apologize for maintaining alongside its production fleet.

Where This Appears
What an Institutional Data Room Should Contain
Related Terms
Model RegistryProduction ModelSignature UniquenessLeakage

Working through a diligence process?

Institutional partners evaluating a position against this platform's outcome and duration models are welcome to reach out directly.

I am a
Name
Organization
Email
Message