HSM Review

Machine Learning Predictions of Winners of World Cup Teams

Oxford mathematical model predicts route to the men's @FIFAWorldCup

The model is based on correlating differences in a number of goals with the differences in their ELO ratings.

There is a strong positive relationship between the number of goals scored and the ELO ratings. So for each simulated match, we can look at who the teams are and what their ratings are according to eloratings.com, and use that to predict what the number of goals difference will be.

Repeat this simulation 100,000 times for every match in the tournament. This gives us a model of how often we expect each team to win the trophy.

The model - - forecasts:

  • England to lose in the quarter-final
  • Argentina vs Brazil in the semi-final
  • Brazil to beat Belgium in the final

KSA has defeated Agentina, and Japan has defeated Germany. Lets see how good this algorithm will be at the rest of the predictions.