Not Argentina, machine learning algorithm predicts this team will win the 2026 FIFA World Cup

The FIFA World Cup featured the famous eight-limbed fortune teller “Paul the Octopus.” In the age of AI and machine learning, predicting World Cup winners has become more sophisticated. Take Achim Zeilais, professor of statistics at the University of Innsbruck, for example. “Machine learning algorithms and subsequent simulations leverage data, expertise and statistical models” to predict who will win the 2026 FIFA World Cup, the biggest tournament to date.

What was the process used to predict the winner?

Achim Zeileis says his algorithm proceeds in two steps. “In the first step, sophisticated statistical models are combined with expert insights from bookmakers and the transfer market to determine the strength of every team and their players. In the second step, machine learning algorithms determine how to best combine strength estimates with other information about the team,” he wrote in The Independent.

“We ran 100,000 simulations to determine the most likely course for the tournament. The results show Spain are favorites to win with a 14.5% chance of winning, closely followed by England and France with 12.4% each and Germany with 11.2%.”

“Portugal and Argentina are also more likely to win the title, with 8.9% and 8.2% respectively.”

deep dive

Zeileis said his algorithm looked at four variables.

“Firstly, all domestic matches over the past eight years form the basis for a ‘retrospective’ estimation of team strength. “Secondly, the ‘future’ strength estimates are derived from the quoted odds of various international bookmakers and reflect the opinion of experts regarding the upcoming tournament,” he further wrote.

“Thirdly, individual player evaluations are created based on their contribution to goals at club and international level. And finally, the player’s current qualities and future potential are reflected in their expected market value. These are available on the Transfermarkt website, which uses a wisdom of the crowd approach to estimate the unknown actual market value.”