Commit Graph

3 Commits

Author SHA1 Message Date
NilsBriggen d3e8df96ef Recency-weighted v4 model: fast-Elo blend, Team-DC majority, nightly refit
Validated on the strict 2018-2022 window and confirmed on the untouched
2022-2026 test set (RPS 0.1703 vs 0.1721 over 4,448 matches):
- the Elo member now blends 30% of a 3x-faster Elo walk, so recent
  results move ratings much harder
- ensemble weight shifts from 75/25 toward Elo to 45/55 toward the
  time-decayed Team-DC member — the recent-form model now leads
- Team-DC refits nightly (and at boot) on the 15-year window plus every
  finished 2026 match, via a new committed-at-build dcTrain.json
  (server-only, excluded from the PWA precache)
- a recent-form goal multiplier was also tested and did NOT validate;
  it ships disabled, and backtest.json records the whole decision

buildBacktest.ts grew the experiment harness: wider xi/k grids, the
fast-Elo and form variants, a pre-registered ship bar (>=0.0005
validation RPS), and a 7-day refit-cadence check. Older ratings.json
files still work — all new model fields are optional with golden
regression coverage.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 23:34:18 +02:00
NilsBriggen f3b2a69b31 v3 Phase B core: Team-DC + backtested ensemble, squad values
- src/lib/model/teamDc.ts: per-team attack/defence Dixon-Coles via weighted MLE
  (Maher iterations), exponential time-decay, shrinkage for thin histories,
  learned home multiplier; log-pool of score matrices. 7 vitest cases.
- buildBacktest.ts rewritten as a strict three-way-split bake-off: params
  pre-2018, tune (xi, k, w) on 2018-2022 validation, final scores on untouched
  2022-2026 test (4,448 matches). Result: ensemble RPS 0.1721 beats Elo-DC
  0.1726 and Team-DC 0.1801; ECE stays 0.01. Tuned: xi=0.5 k=5 w=0.75.
- Runtime now IS the backtested model: ratings.json carries teamDc+ensembleW;
  matchMatrix() pools both members for predictions, Monte Carlo and advance
  probs; ModelEngine re-rating preserves ensemble params; lambdas reported as
  matrix expectations (drives in-play).
- scripts/buildSquadValues.ts: Transfermarkt FIWC participants page -> all 48
  squad market values (France 1.52bn ... Qatar 20m). Display/availability layer
  only — June-2026 values are NOT in the backtested model (would leak).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-11 17:23:51 +02:00
NilsBriggen d604bb14ff Phase 2: predictive model — Elo + Dixon-Coles + Monte Carlo
- buildRatings.ts: walks 49k historical results → World-Football-Elo per team,
  data-calibrated goals model (goals-per-Elo slope, mean goals) + MLE-fit
  Dixon-Coles rho. Top: Spain/Argentina/France (the real 2026 favourites)
- src/lib/model: elo, poisson/dixon-coles, predict, host-advantage, monteCarlo
  (full 48-team sim — group sampling, best-third bipartite allocation, knockout
  advance probs). 20 vitest cases incl. exact per-round count invariants
- Server ModelEngine: live Elo re-rating after each result, per-match W/D/L,
  20k-sim odds, odds-over-time history; broadcast on finished-result changes
- Client: championship board, heat-shaded odds table, lazy-loaded title-race
  chart (Recharts split to its own chunk), match-prediction bars, bracket
  advance overlay
- Verified: odds render, chart populates as injected results re-rate teams live

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-11 13:25:53 +02:00