Monte Carlo simulator and interactive analytics for the FIFA World Cup 2026 — ELO + Poisson engine, injury-adjusted squads, prediction-market demo, and bookmaker edge analysis.

• Monte Carlo engine — run 1K / 10K / 50K / 100K full tournaments in the browser via Web Worker (~15–25K sims/sec).
• Dual simulation pass — every run computes probabilities with and without squad absences (injuries/suspensions), so you can compare the impact on champion odds.
• Rich dashboard after each run:
  • Prediction delta — how absences shift top-team probabilities
  • Champion probabilities with Wilson 95% confidence intervals and with/without toggle
  • Stage matrix — reach R32 / R16 / QF / SF / Final / win %
  • Group standings — expected finish distribution per team
  • Bracket tree — most likely knockout path
  • Match calendar — 104 fixtures with modal W/D/L and score distributions (click a row for detail)
  • Tournament & goal stats, surprise teams, market edge vs Polymarket/Kalshi
• Team & match drawers — ELO, absences, stage probabilities, score heatmaps, scorers
• Sticky section nav, confetti on sim completion, hero image gallery
• Mobile responsive — hamburger nav, scroll-friendly tables, stacked demo layouts

Play-money sandbox that reuses the home-page simulation — no second run when you open the demo.

• $1,000 play-money wallet (persisted in localStorage)
• Settle markets against a random sampled tournament outcome from your sim
• Confetti on successful YES/NO trades
• Filters: all · winner · groups · matches

Historical validation on World Cups 2014, 2018, 2022 — calibration buckets, home-bonus sweep, recent-form blend, and penalty-shootout model evaluation.

Full write-up of ELO expectancy, Poisson goals, knockout penalties (Bayesian shrinkage on 103 historical shoot-outs), Monte Carlo aggregation, Wilson CIs, and known limitations.

• Hero: “Who will win the 2026 World Cup?” with live sim controls
• Header developer contact card (Dexoryn · Telegram · Discord · X) on every page
• Footer credits + Made by Dexorynlabs with social links

Performance: ~35K sims/sec in Node; ~15–25K in browser worker. 100K dual-pass run typically ~5–10 s depending on device.

Data: official FIFA 2026 draw (5 Dec 2025) · ELO from eloratings.net · optional live odds via /api/odds

npm install --legacy-peer-deps
npm run dev          # http://localhost:3000
npm run build        # production build
npm test             # vitest
npm run test:pipeline # test + build + sim smoke

Copy .env.example to .env.local for odds API keys and optional Redis (REDIS_URL or REDIS_MEMORY=true).

npm run scrape-elo       # refresh national-team ELO
npm run fetch-odds       # Polymarket / Kalshi winner odds
npm run fetch-absences   # squad absences (injuries, suspensions)
npm run fetch-backtest   # historical WC match data for backtest
npm run fetch-results    # live results (ESPN)
npm run fetch-cards      # card accumulation data
npm run redis:health     # ping Redis (or memory backend)

We = 1 / (10^(-dr/400) + 1)
dr = ELO_A − ELO_B + home_bonus   (+100 for host nations in group stage)

λ_team = clamp(1.30 + 0.18 · (ELO_team − ELO_opp + home_bonus) / 100,  0.15,  6.0)
goals ~ Poisson(λ_team)

Regulation draw → penalty shoot-out modeled with Bayesian shrinkage on historical PK rates (not a coin flip). Penalty goals are excluded from goal aggregates.

Key players out (injury/suspension) apply an ELO penalty per squad before each tournament draw. The worker runs a counterfactual pass with absences disabled for comparison.

• Champion probabilities sum to 100% (exact over N sims)
• Average goals per match ≈ 2.6 (in line with WC history 2.5–2.7)
• Top contenders align with ELO / bookmaker consensus (Spain, Argentina, France, Brazil, Portugal)

See /methodology in the app for the full spec.

Private / all rights reserved unless otherwise noted in repository settings.