Bayes Rule for Soccer Predictions Ahead of the World Cup
https://medium.com/codex/football-analytics-using-bayesian-network-for-analyzing-xg-expected-goals-705e63e597c2
Every four years, the entire world comes together to watch 32 countries with the best national football teams fight for a chance at the glorious World Cup. Simply by it’s name, The World Cup, it’s prestige and importance are highlighted for what it is, the most important sporting event in the world. But as fans count down the days until their favorite team kicks off, they ask themselves, will we win? Well, one way of approaching this question is to use basic mathematics, determining how many goals each team will score in a given game and see if it is greater or less than the opposing team. More specifically, the article above has already explored how we can use Bayesian Networks (an application of Bayes Rule), utilizing conditional probability, to predict the expected goals of a player in any given game.
In the article above, the author highlights the most basic idea of how to calculate the predicted goals of a player using Bayes Rule. They model a player scoring using P(A) = user scores a goal and P(B) = shot from player leads to goal, where P ( A | B ) is the odds of a player scoring a goal on any given shot. They then multiply this by the number of shots in a given game to calculate the expected goals of a player. The issue is though, that the P(B) as mentioned above, is affected by a multitude of factors, such as determining from where the shot is taken (inside or outside the box) or the coverage of the player on the shot (is it a free kick, penalty, with defenders in the straight line path, etc). As a way to address this issue, the author uses Bayesian Networks to create a conditional probability model that can, at it’s fundamental core, still apply Bayes Rule, but factor in these many variables while doing so. In the photo below, you can see the final result of the Bayesian Network from the article, such that the end probability of “is_goal” = P(B) = is a taken shot resulting in a goal, is determined by it’s incoming probabilities and the many variables that can lead to a shot either leading to a goal or being blocked/missing the goal.
As we look forward to the World Cup and cheer on our teams, think about how you could use statistics to better understand and predict the beautiful game of football. Or if your team is losing and you stare at the TV in frustration, using the Bayesian statistics in the article above might just give you the hope you need to need to not lose faith (and not smash your TV in).