“It is difficult to make predictions, especially if they concern the future” is a famous catchphrase that is attributed to Karl Valentin, Mark Twain or Niels Bohr. The truism also applies to this year’s European Football Championship for men. Programmers and statisticians were only partially correct in the tournament so far, despite their sophisticated methods. Because it is difficult to count against chance.
France and England in the European Championship final?
Researchers at the Karlsruhe Institute of Technology (KIT) programmed an algorithm for the outcome of the European Championship and saw the teams from France and England – the latter with a slight advantage – in the final. As is well known, France retired on Monday evening. England, which defeated the German team in the round of 16 on Tuesday, can still reach the final.
Coincidences are difficult to predict
But until then there are still many imponderables. “A stupid ricochet from the post, an unfortunate own goal – it is just the coincidences that can counteract any statistically calculated probability,” says Alexandros Stamatakis, Professor of High Performance Computing at KIT and research group leader at the Heidelberg Institute for Theoretical Studies (HITS). He developed software together with Ben Bettisworth from HITS.
The problem with this: “Football results during a European Championship are extremely difficult to predict because there are few goals overall and many of them happen by chance,” explains Stamatakis. “The strength of a team cannot necessarily be expressed in results.”
You saw that on Monday in the last 16 match between France and Switzerland. “France was stronger in all respects, but chance did not want it.” There was a penalty shoot-out: France lost 4-5. It is easier, for example, to simulate the Bundesliga: Because everyone plays against everyone for a season and coincidences average out over the course of the season, says Stamatakis.
EM was simulated 100,000 times
Stamatakis and Bettisworth built on the findings of researchers from the Technical Universities of Dortmund and Munich as well as statistics experts from the University of Innsbruck, among others. Before the start of the tournament, they had calculated the probability of winning in pairs and, among other things, combined the strengths of the teams and the market value of players.
The EM was then simulated 100,000 times game by game. The result is similar to that of the KIT researchers: France becomes European champion. The statisticians gave a probability of 10.1 percent for the German team to win a tournament.
On the basis of their colleagues’ data, Stamatakis and Bettisworth now determined the probability of winning the EM themselves – because no simulations were necessary, within microseconds. They were helped by an algorithm from bioinformatics that is used to create family trees – such as those of coronaviruses and the extent to which these viruses are related.
KickForm experts see Belgium as the European champion
Other experts are also concerned with predicting football games: For example, the startup KickForm, for which the physicist and author of the book “The Perfect Tip”, Andreas Heuer, developed the algorithm. Among other things, he calculates with the so-called Elo numbers, in which teams are rated according to their results.
In the final, he and KickForm see Belgium and England – and Belgium as future European champions. Deka-Bank economists, on the other hand, calculated the already eliminated France as European champions, but were at least correct with the round of 16 prediction for the game England against Germany.
Lady elephant Yashoda was wrong too
Who will be European champion in the end? Maybe animals oracles aren’t that bad after all. Paul the octopus, who has since passed away, for example, correctly guessed the seven games with German participation including the final at the 2010 World Cup in South Africa.
A dachshund, a snow leopard and a goat tried their luck at the 2018 World Cup. For the current EM, the elephant lady Yashoda from Hamburg is in great demand. For Tuesday evening, however, she had predicted a victory for the Germans.