Can Brazil win the 2022 World Cup?

In less than six days, the football feast "2022 World Cup" will be staged in Qatar.

Dear fans, you must have the intention of staying up late to watch the ball, and even predicted this year's champion in advance.

With the approach of the 2022 World Cup, predictions from all sides have also poured in. Manchester City coach Guardiola expressed confidence in Argentina's ability to win the championship; European professional forecasting institutions believe that France has more champions; Ball king Bailey also predicted that this year's champion should belong to England ... Of course, as in previous years, five-star Brazil is still the biggest favourite to win the World Cup.

So, maybe one last dance Messi can lead Argentines to Qatar in the World Cup? Can the French team led by Benzema and Mbappé break the championship spell and successfully defend their title? Can "five-star Brazil" avoid the heat and die, and take the World Cup back to Brazil 20 years later? Can other teams "hit the face" with various predictions and win the championship all the way?

Now, an international joint team uses artificial intelligence (AI) to give a new prediction that Brazil's "five stars" will become "six stars" and win the World Cup again after 20 years.

A joint research team from Dortmund University of Technology, technical university of munich and Luxemburg University. Combined with several statistical models representing team strength, team structure (market value or number of champions league players) and socio-economic factors (population or GDP) in the country of origin, a prediction based on conditional reasoning random forest learner is made.

Among them, the training data of random forest model comes from five World Cup matches from 2002 to 20 18, so as to predict the final result of the 2022 World Cup.

Figure | The probability of each national team participating in this World Cup winning the championship.

The forecast results show that Brazil has a 23.5% probability of entering the finals, and the probability of winning the championship is 15%, ranking first; Argentina is close behind, with a winning probability of11.2%; France and England are not in the top three teams, and the Netherlands ranks third with a probability of 9.7%; Germany and France ranked fourth and fifth respectively.

Following all the rules of the lottery and FIFA, the research team conducted 65,438+00,000 complete simulations using AI, and predicted the possibility of all teams entering different rounds and eventually winning the championship.

Figure | For each possible combination of opposing teams, the probability that one team will beat the other team in the knockout (green and purple indicate that the probability is higher and lower than 50% respectively).

Of course, the outcome of the game is far from predetermined-because some top teams have a relatively low probability of winning.

In this regard, Andreas Groll, a professor of statistics at Dortmund University of Technology, explained, "Prediction may also be wrong, which is the essence of prediction, otherwise the football match will become very boring. We only provide probability, not certainty. The probability of winning 15% also means the probability of losing 85%. "

However, the previous similar predictions were quite successful: Achim Zeileis, a professor at the School of Economics and Statistics of Innsbruck University, correctly predicted the 2008 European Cup final, the 20 10 World Cup champion and the 20 12 European Cup champion. This time, the model of Zeileis team will be used as part of a more comprehensive combination model proposed by the joint team.

Figure | Probability of each team entering 16,14 finals, semi-finals, finals and winning the championship.

"Due to the high temperature in Qatar in summer, the World Cup had to be postponed until winter. In the winter months, the major football leagues in Europe and South America have to interrupt their previous arrangements to adapt to the game, which makes each national team have less time to prepare for the game and players have less time to recover before and after the World Cup. Coupled with extreme weather conditions, the risk of injury to players has increased, "Zeileis said.

In Graul's view, it may no longer be an advantage to have more players playing in international leagues (such as the Champions League, Europa League and UEFA). "All these factors make it more difficult to predict the result of the match, because variables that have proved to be very meaningful in previous World Cups may not play a good role or play a different role."

The research team said that the model trained in this way can also be used for other future predictions, and better football match prediction may eventually provide more accurate weather prediction. Regardless of the final result, let's look forward to the upcoming World Cup.