Nota de investigación

Calibrating predictions differently for friendlies vs tournaments

Estado: Shipped (Variant 4 — per-tier Platt temperature scaling). Production calibrator uses a hybrid strategy: Platt for the tournament tier (where isotonic collapses to identity at n~70), isotonic for friendlies/qualifiers (where it's more expressive at n~400+). Gate passed. See results belowResumen + nota completa · 3,433 palabras

Resumen

The shipping ensemble calibrator (scripts/fit_ensemble_calibrator.py) fits per-class isotonic regression curves on the uniform-averaged three-component output (Elo bracket MC + Dixon-Coles + Hierarchical Poisson MAP). The first cut lifted holdout ECE on the 365-day common-subset training pool from 4.62pp uncalibrated → 2.70pp under the pooled-across-tiers fit (5-fold CV, n_train = 939, current artefact at data/wc2026/ensemble_calibrator.json).

A subsequent tier-aware refit (three sets…

Nota completa

Standard Pass

Lee la nota de investigación completa

Calibrating predictions differently for friendlies vs tournaments tiene 3,433 palabras. El Standard Pass desbloquea todas las notas de investigación completas, más el pronóstico completo y las valoraciones por selección y por jugador, válido durante todo el torneo.

Consigue el Pass — 15 $

24h self-service refund·No subscription, no auto-renewal·Access through 31 Dec 2026. See refund policy.