Browse all

Machine learning for World Cup's winner

Football > AI

Machine learning for World Cup's winner Football > AI

If it is true that there is a sort of 'curse' that affects the World Cup's champions (in the last five editions of the World Cup, four times have not get through the group stage), there is a study carried out before the beginning of Russia 2018 which makes the elimination of Germany even more unpredictable. Andreas Groll and some colleagues from the University of Dortmund had in fact produced an in-depth analysis of 28 pages which made it possible to establish, among other things, the winning percentages of each team. All this thanks to a system of artificial intelligence and more than 100,000 simulation attempts, taking as basic parameters statistics and numbers of various kinds such as the country's population, the national GDP, the FIFA Ranking, the bookmakers' odds, the age average of each team, the number of players playing together in the clubs and how many Champions League player have win.


According to this calculation the most likey winner team is Spain, with a 17.8% chance to raise the Cup to the Moscow sky next July 15th, closely followed by Germany (17.1%) and Brazil (12.3%). But what amazes most is the observation of the simulation of the entire tournament, starting from the knockout stages. Despite the initial recommendations of the same researchers to take the predictions with the right seriousness, in fact, in case of making the quarter-finals the team favored for the title would be the Germany of Joachim Löw.