Forschungsnotiz

Calibrating predictions differently for friendlies vs tournaments

Status: 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 belowZusammenfassung + vollständige Notiz · 3,433 Wörter

Zusammenfassung

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…

Vollständige Notiz

Standard Pass

Die vollständige Forschungsnotiz lesen

Calibrating predictions differently for friendlies vs tournaments umfasst 3,433 Wörter. Der Standard Pass schaltet jede Forschungsnotiz in voller Länge frei, plus die vollständige Prognose und die Team- und Spielerbewertungen, gültig für das Turnier.

Pass kaufen: $15

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