Base Rate

A base rate is the underlying frequency at which a defined outcome occurs across a specified population, the foundation any calibrated probability estimate is built from. A base rate estimated from too small or too narrow a sample is noise dressed up as a finding, not a validated base rate.

A base rate is the underlying frequency at which a defined outcome occurs across a specified, historical population, and it is the raw material every calibrated probability estimate in this platform's models is ultimately built from. A base rate is only meaningful at the level of granularity it was actually estimated at: a jurisdiction-level base rate computed from a large, deep resolved-matter sample supports a real estimate, while the same jurisdiction's county-level or judge-level base rate, if estimated from only a handful of resolved matters, is a small sample masquerading as a validated finding rather than genuine, granular evidence. Estimating a base rate at too coarse a level, averaging genuinely different sub-populations into a single reported figure, understates real variance that matters to an institution's actual exposure; estimating it at too fine a level without sufficient sample size trades that error for a different one, over-fitting to noise. A properly disclosed base rate states both the level of granularity it was computed at and the sample size supporting it, so a counterparty can judge whether the base rate represents genuine, validated evidence or a plausible-sounding number with too little data behind it to actually trust, a distinction that matters most in exactly the granular segments a portfolio's real exposure concentrates in.