Brazil is the 2022 World Cup champion? This time it was AI that said it

In less than 5 days, the football feast "2022 World Cup" will be staged in Qatar.

Dear fans, I believe you have already made plans to stay up late to watch the game, and even predicted this year's champion in your hearts.

As the 2022 World Cup approaches, predictions are coming in. Manchester City coach Guardiola said he is confident in Argentina's ability to win the championship; European professional prediction agencies believe that France is more likely to win the championship; Pele also predicted that this year's championship should belong to England... Of course, as in previous years, five-star Brazil is still the biggest favorite to win the World Cup.

So, can Messi, who may be in his last World Cup dance, lead Argentina to realize its dream of reaching Qatar? Can the French team, led by Benzema and Mbappe, break the championship curse and successfully defend the title? Can the "Five-Star Brazil" avoid being the favorite and bring the World Cup back to Brazil after 20 years? Can other teams "slap" various predictions and win the championship as dark horses?

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

A joint research team from universities including the Technical University of Dortmund, the Technical University of Munich, and the University of Luxembourg combined multiple statistical models representing team strength with information such as team structure (market value or number of Champions League players) and socioeconomic factors of the country of origin (population or GDP), and made predictions based on a conditional inference random forest learner.

Among them, the training data of the random forest model comes from five World Cup matches from 2002 to 2018, in order to predict the final results of the 2022 World Cup .

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

The prediction results show that Brazil has a 23.5% chance of entering the final and a 15% chance of winning the championship , ranking first; Argentina follows closely behind with a 11.2% chance of winning the championship; France and England are not among the top three favorites to win the championship, and the Netherlands ranks third with a probability of 9.7%; Germany and France rank fourth and fifth respectively.

Following the tournament draw and all FIFA rules, the research team used AI to conduct 10,000 full simulations , predicting the probability of all teams entering different rounds of the tournament and ultimately winning the championship.

Figure | The probability of one team defeating another in the knockout stage for each possible combination of opposing teams (green and purple indicate probabilities above and below 50%, respectively).

Of course, the outcome of the match is far from predetermined – as some of the top teams have relatively low odds of winning.

In this regard, Andreas Groll, professor of statistics at the Technical University of Dortmund, explained, "Predictions can also be wrong. This is the nature of predictions, otherwise football matches will become very boring. We only provide probabilities, not certainty. A 15% chance of winning also means an 85% chance of failure ."

However, similar predictions have been quite successful before: Achim Zeileis, a professor at the Faculty of Economics and Statistics at the University of Innsbruck, and his team correctly predicted the 2008 European Championship final, as well as the 2010 World Cup winner and the 2012 European Championship winner. This time, Zeileis's model will be part of a more comprehensive combined model proposed by the joint team.

Figure | The probability of each team entering the round of 16, quarterfinals, semifinals, finals and winning the championship.

"Due to the high temperatures in Qatar during the summer, the World Cup had to be postponed to the winter. During the winter months, all major football leagues in Europe and South America had to interrupt their previous schedules to adapt to the competitions, which left national teams with less time to prepare and players with less time to recover before and after the World Cup. This, coupled with extreme climatic conditions, increased the risk of players getting injured ," said Zeileis.

In Groll's view, having more players playing in international leagues (such as the Champions League, Europa League, and UEFA Cup) may no longer be an advantage. "All these factors make it more difficult to predict the results of the game, because variables that have proven to be very meaningful in previous World Cups may not work well or work differently."

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

Please leave your predictions in the comment section.

Reference Links:

https://www.eurekalert.org/news-releases/971175

https://www.zeileis.org/news/fifa2022/