A computer model gives Brazil more than a 25 percent chance of winning the World Cup. According to the model, Belgium has almost a 19 percent chance of winning, and the Netherlands less than 5 percent.
According to a prediction model of the Alan Turing Institute in London, Brazil is the most likely winner of the 2022 World Cup. The publicly accessible model gives Brazil a 1 in 4 chance of winning the final, based on 100,000 tournament simulations. Belgium is also rated highly, with a chance of winning of almost 1 in 5. The Netherlands, on the other hand, has little chance: less than 1 in 21.
Many people, from bookies to bankers, have already used computer models to predict the winner of the 2022 FIFA Men’s World Cup in Qatar. However, most of those models are kept secret.
READ ALSO
Food lessons from space
Software developer Nick Barlow from the Alan Turing Institute and his colleagues have now developed a model that people to run on their computer at home. On an average laptop, 1000 tournament simulations take about 15 minutes.
“It is very important to us that most of the things we do open-source are,” says Barlow. “We encourage people to get involved, use and contribute to our Code.”
Additional weight
When Barlow and his team ran their model through the tournament 100,000 times, Brazil won 25.1 percent of the time. Belgium followed with 18.9 percent and Argentina with 13.2 percent. The Netherlands was in seventh place with 4.7 percent.
The researchers took as a starting point for their model a method that is often used for competitions in national competitions. In that method, each team gets a score for defense and attack.
However, the researchers modified their model in a few ways. They eliminated home field advantage, which is absent for all teams in Qatar (except the home nation). They also took into account differences in strength that became apparent when teams played against each other in international friendlies.
The researchers also modified the model to give more weight to the results of important past competitions, such as semi-finals and finals. Results from recent competitions also received extra weight. Finally, the researchers applied the model to previous tournaments to see how well the predictions matched the real results. Based on that performance, they also improved the model.
Super computer
Barlow’s model roughly corresponds to a statistician’s model Achim Sailing from the University of Innsbruck in Austria and his colleagues. They ran an algorithm on a supercomputer and found that Brazil is the likely winner with a 15 percent chance of winning.
Other models predict other winners. Insurance company Lloyd’s predicted that based on the collective insurance value of the players of each team England will win by beating Brazil in the final. The same model previously correctly predicted that Germany would win the 2014 World Cup and France the 2018 World Cup.
A model designed by statistician Matthew Penn from the University of Oxford and colleagues predicted Belgium as the most likely winner. This model correctly predicted the winner of Euro 2020 last year, as well as six of the eight quarter-finalists.