Nota di ricerca

Calibrating predictions differently for friendlies vs tournaments

Stato: 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 belowTopline + nota completa · 3,433 parole

Topline

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

Leggi la nota di ricerca completa

Calibrating predictions differently for friendlies vs tournaments contiene 3,433 parole. Lo Standard Pass sblocca ogni nota di ricerca nella versione integrale, oltre alla previsione completa e alle valutazioni per squadra e per giocatore, valido per tutto il torneo.

Acquista il Pass: $15

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