Sports Analytics Weekly by kubeia.io - 5/2025

Set piece expert Stuart Reid has a guest post on how to design great corners.

In football, the objective is to score more goals than your opponent. Ergo, every decision taken or action carried out on a football pitch should be (and almost exclusively is) an attempt to increase your own team’s chances of scoring a goal and/or decreasing your opponent’s chances of scoring a goal. Usually there are a subset of micro decisions and executions that contribute to the success of any of these given actions, but the end goal remains constant.How a team/manager sets about achieving those objectives is the fun part, which brings me on to the focal point of this blog post: Route One/Direct/Long Ball football.

Time to bite the bullet and go in on this one again. I’ve written about overfitting extensively, both on this blog and in papers. A central line of my research for the last ten years has been motivated by the observation that overfitting doesn’t exist. We need to abandon the term. Let me try to motivate a different view through the lens of evaluation.

The prototypical statistical prediction problem is bit prediction: Having seen a long list of 0s and 1s, your task is to predict whether the next bit equals zero or one. For example, how might I predict whether Steph Curry will make his next free throw? I might look at all the times Steph attempted free throws this season. From this, I could compute his free-throw percentage (94.0%). I can then use this percentage to inform my prediction (I’m guessing he’s going to make it).

Now in its 28th edition, the Deloitte Football Money League profiles the highest revenue generating football clubs in world football. The 2023/24 season unlocked a new frontier, as Real Madrid became the first football club to record €1 billion in revenue.