DATA · POWER RATING ENGINE
World Cup Analytics Hub
Built on the directed graph of who beat whom: personalised PageRank computes the power ladder, while Monte Carlo simulation projects qualification, progression and title odds. All computed in-house, evolving with every real result.
104 of 72 group games played
2,500 Monte Carlo simulations
Jul 26 23:24 Last data sync (MYT)
Power Ladder
Title Odds
Group Qualification
Match Forecasts
Title & Progression Odds
2,500 simulations · computed Jul 26 21:24
↻ Refresh forecast
Team Title Final Semis Title odds
Group Qualification Odds Includes the best-eight-thirds qualification rule
Methodology
Power ratings (PageRank): Teams are nodes; every finished match draws a directed edge from loser to winner (draws count half each way), weighted by goal margin. On this graph we run damping-0.85 personalised PageRank , with a restart vector of pre-tournament priors (Elo scale) — solving the cold start while real results steadily take over the rating. The stationary distribution r maps back to Elo via E = 1700 + 400·log₁₀(r·N).
Forecasts & probabilities: The Elo gap sets expected goals; independent Poisson distributions yield win/draw/loss odds and the most likely score. We then run 2,500 Monte Carlo simulations of every remaining match to estimate group qualification (incl. best-eight thirds), round-by-round progression and title odds. Finished results stay fixed.
Note: until the official bracket is set, knockout pairings use an unbiased random draw of the 32 qualifiers; priors only matter pre-tournament and are quickly corrected by real data. For reference only.