This mathematical model predicts the possible outcomes of the World Cup

Posted by dom on January 26th, 2019

A Nigerian calculated which country will win the World Cup based on a mathematical model of his authorship. So far there is no undisputed favorite as both Brazil (17.9%), Germany (15.6%), Spain (15.3%) and Argentina (11%) have similar chances of winning the World Cup.

The Nigerian engineer and tipster from sure banker bets has used mathematics and programming to elaborate a prediction model for the World Cup Russia 2018 and has made some predictions somewhat different from the last ones published, both by Austrian mathematicians and by the simulator of the video game FIFA 2018 and there is no clear favourite to win the World Cup. The results have been published by the Spanish newspaper El País.

For his prognosis, engineer Kike Llaneras has used the Elo scoring system, similar to the one used in chess and other sports, which measures the relative capacity between teams, which win and lose points when they face each other. Llaneras has used this scoring system to generate three factors: the players, the expected results and their historical results. 

The three factors 

The first factor taken into account by the mathematician is the so-called "Elo of the players", which has a weight of 20% and uses big data from the soccer players of each team giving them a value based on their value in the transfer market and the strength of the club to which they belong. For example, according to these data, the world's most valuable player plays in Argentina: Lionel Messi who belongs to the strongest club (FC Barcelona, which has an Elo of 2025) and is one of the most expensive players of all (valued at 180 million euros).

The second factor is the "Elo esperado". For this, mathematicians use the expected goals. This statistic says how many goals a team should have scored (on average) with the shots it made in a match, taking into account many details of each shot, such as the distance, angle, type of shot or the previous play.

The third and final factor is the "Elo of the team" itself. This factor measures the strength of each team according to its results. This score is interchangeable as the winner of a match gets the loser's points. If the victory is by surprise the teams exchange more points. The exchange is also greater if the victory is by several goals and when the game confronts teams that go up in the score.

Modeling

Using all three factors, Llaneras uses a match model to train the mathematical model. In total, several tens of thousands of historical and data from more than 150 national teams are used, including those that did not qualify for this World Cup.

Then the model is checked for accuracy and parameters are adjusted to fit. More than victory/defeat/beat the results are measured on the basis of goals for and against as it allows the software to easily adapt to the format of the tournament, which is slightly different to the national leagues or the Champios League, for example.

Finally the last step is to simulate the tournament 10 thousand times, playing virtually every match of the group stage, round of 16, and so on until the final. A virtual ranking is updated during the virtual tournament. The model takes into account the rules of the group phase and the World Cup table to create successive matches and also considers the possibility of ties, extensions and penalties.

So far there is no undisputed favorite as both Brazil (17.9%), Germany (15.6%), Spain (15.3%) and Argentina (11%) have similar chances of winning the World Cup. Still, it serves to get an idea about what to expect over the next month as the model also makes predictions about which teams could move to the next rounds. Colombia, Uruguay, Peru and Mexico seem to have a chance to go a little further than expected.