With less than a week to go before the ball starts rolling at Al Bayt Stadium, a new edition of the FIFA World Cup in Qatar is about to begin. As expected, many people have turned to technology to try to predict the early results and ultimately a champion. One of those people is Achim Zeileis from the Department of Statistics at the University of Innsbruck in Austria, who, together with an international team of researchers, simulated the tournament 100,000 times. A small spoiler? The Cup would not be traveling to Europe this year…
On November 20 at seven in the evening (local time), Qatar and Ecuador will face off in the first match of the 2022 World Cup, a privilege that previously belonged to the defending champion. Unfortunately, this tournament has much more than football at its core, with serious organizational problems, multiple corruption allegations, and more than 6,500 workers dead since the Asian country was confirmed as host.
However, that deserves a separate article. What brings us here today is the probability of winning the tournament, and what machine learning has to say. Researcher Achim Zeileis and an international team from the Department of Statistics at the University of Innsbruck created a special model capturing past, present, and future of the participating teams. Who will win the 2022 World Cup?
Brazil Will Be the Champion of Qatar 2022, According to an AI
Zeileis and his team place Brazil in first place with a 15% chance of victory. Immediately after comes Argentina with 11.2%, and the Netherlands closes the podium with 9.7%. Germany and France almost tie for fourth and fifth place (9.2 vs. 9.1), and Spain appears only in sixth place with a probability of 7.5%.
The system combines historical information from the last eight years with a bivariate Poisson distribution, future expectations with a consensus model using data from 28 bookmakers, and the current state of the teams through a “random forest” method trained with information from the last five World Cups. Secondary values were also incorporated, ranging from the FIFA ranking to market valuation, as well as the population and GDP of each country. From there, the system managed to predict a number of goals for each team, which were then used in 100,000 simulations.
The article ends with two additional charts: a heat map that explores the head-to-head matchups between the teams, and the probabilities of each one advancing to the next phases. What does all this leave us? The tournament promises to be especially competitive, at least from a statistical point of view.
Official site: Click here
