Ilgiz SatkynovSoftware engineerBay Area, CA

World Cup Predictor

Predicts World Cup matches from team ratings, then simulates whole tournaments.

Status
Complete analysis with a published results page.
Type
Software project
Built with
  • Python
  • XGBoost
  • Monte Carlo simulation
  • Jupyter
Open websiteView source
World Cup Predictor page with the headline “World Cup outcomes, modeled and simulated.”
The results page.

What it does

Each match is predicted from EA Sports FIFA ratings, Elo ratings and FIFA rankings, then the full tournament is simulated many times to estimate each team’s chances, including the 48-team format.

XGBoost was chosen from five models using leave-one-World-Cup-out validation.