Module probability 

Empirical and isotonic probability calibration, out-of-sample metrics, and validation manifests.

Maps raw heuristic indicator/strategy scores to monotone calibrated probabilities fitted exclusively on training data and frozen for evaluation on unseen test datasets.

Structs

CalibratedProbability

A calibrated probability output distinguishing raw confidence from empirical likelihood.

CalibrationMetrics

Out-of-sample calibration performance metrics and baseline comparisons.

IsotonicCalibrator

A non-parametric monotonic probability calibrator using the Pool Adjacent Violators Algorithm (PAVA).

ValidationExperimentManifest

Verifiable experiment manifest documenting the full training and out-of-sample evaluation context.

Functions

block_bootstrap_brier

Circular block bootstrap estimating confidence intervals for Brier Score under time dependence.

compute_calibration_metrics

Computes comprehensive calibration metrics comparing predicted probabilities against actual outcomes.