Covariate shift detection bounds expand predictably under long conversion lag, requiring density ratio variance scaling to prevent false positive retraining alerts.
Direct density ratio estimation across sparse data streams requires temperature scaled score mapping and adaptive kernel widths to prevent unbounded variance.
Early venture conversion metrics require sparse baseline reference classes and extended observation windows to prevent mistaking low-sample noise for strategic signal.
Stochastic particle filtering estimates non-linear venture cash runway bounds by recursively updating particle weights against multi-channel transactional feeds.
Dynamic Bayesian recalibration continuously updates historical unit economic priors as acquisition and retention distributions shift over time.
Early capital depletion tracking relies on Bayesian prior distributions constructed from historical reference classes to isolate structural overhead from experimental burn.
Decay rate vectors measure output degradation across early execution reference classes by comparing live baseline readings against historical performance windows.
Outside view forecasting replaces optimistic internal projections with empirical distributions drawn from structurally similar historical ventures.
Reference class forecasting corrects founder optimism by anchoring internal projections to historical distributions of outcomes from similar ventures.
Production degradation velocity quantifies metric degradation rates by comparing short sliding observation windows against static invariant telemetry anchors.
The terms are written where they belong. Each field holds three desks, and every entry a desk writes raises the terms it uses, each term given a meaning, a mechanism and the places it appears. The record on these pages stays first person and hand made; the fields grow the nomenclature.
