How scientists are attempting to predict the World Cup 2018 winner

First off was a model based on a range of bookmakers’ odds which are compiled by professional statisticians, and which would lose the bookies a great deal of money if they were wrong.

With this model, unsurprisingly, Brazil stormed ahead as the clear favourites with a 16.6% of chance of winning the whole tournament, followed by Germany at 12.8% and Spain close behind at 12.5%.

But all is not necessarily lost. England fans can take heart that these kinds of conventional statistical approaches are hardly cutting-edge science.

The researchers believe that more complicated analysis is needed to really identify who is likely to bring the World Cup home from Moscow.

And new machine learning techniques, including a method called the “random-forest approach”, has identified a different favourite.

Dr Andreas Groll at the Technical University of Dortmund in Germany and several colleagues compared three different modelling approaches all based on performance in the four previous tournaments from 2002 until 2014.

The best of those has led the scientists to believe they have improved the predictive power of their algorithm substantially.

Their work however doesn’t simply select the best team at the beginning of the tournament, but predicts the likelihood of their progression through the various stages to select a favourite.

After repeatedly simulating the upcoming tournament, the scientists believe that Spain and not Brazil are in fact the favourites to lift the trophy – but only until the quarter-finals.

The way that their algorithm works means that the probability of a chain of events happening changes once previous events in the chain have been completed.

As a crude example, it is extremely unlikely to throw a coin so it lands heads nine times in a row, but after eight heads the chance is increased to 50%.

As soon as the algorithm is run again among the last eight, Germany becomes the most likely winner – something which the scientists say is accounted for by Germany having a more difficult group and first knockout stage.

The scientists have also provided survival probabilities for all of the teams at all of the stages of the tournament, and England fans will quickly notice that this sees their team getting knocked out by Germany in the quarter-finals.

England fans should definitely note that the data the researchers used to measure the teams is heavily based on the performances in past tournaments, however.