Sports Analytics Weekly by kubeia.io - 2/2026

On the Cannon Stats discord (it’s a pretty fun place to hang out and talk Arsenal/Soccer and free to join) a subscriber shared this video titled “The 4 Flaws Of Expected Goals”.This seems specifically designed to needle me (and probably Adam too), but it’s a good video despite the bait-y title/thumbnail, and it really got me thinking.The first thing is that this is not really a video about the flaws or xG but rather how more casual fans use xG and some of the more common misunderstandings of what the intentions are.

Most attempts to describe players start and end with outcomes: goals, assists, chances created. But football is mostly something else — long stretches of circulation, structure, and coordination far from the box.This 2015 paper asked a deceptively simple question: can a player’s passing style be captured as a stable, quantifiable pattern?

A course by Andrej Karpathy on building neural networks, from scratch, in code.We start with the basics of backpropagation and build up to modern deep neural networks, like GPT. In my opinion language models are an excellent place to learn deep learning, even if your intention is to eventually go to other areas like computer vision because most of what you learn will be immediately transferable. This is why we dive into and focus on languade models.Prerequisites: solid programming (Python), intro-level math (e.g. derivative, gaussian)

Instead of just learning how to use a deep learning library, we are going to learn how to create one.We start with a blank file and NumPy, and we don’t stop until we have a functional autograd engine and a collection of layer modules. By the end, we will use it to train MNIST, simple CNN and simple ResNet.

Adapting transformer architectures to learn foundational player features from match sequences.

The Thinking Game takes you on a journey into the heart of DeepMind, capturing a team striving to unravel the mysteries of intelligence and life itself.

Filmed over five years by the award winning team behind AlphaGo, the documentary examines how Demis Hassabis’s extraordinary beginnings shaped his lifelong pursuit of artificial general intelligence. It chronicles the rigorous process of scientific discovery, documenting how the team moved from mastering complex strategy games to the ups and downs of solving a 50-year-old "protein folding problem" with AlphaFold.

Following its world premiere at the Tribeca Festival and a successful international tour, the film is now available here for all to watch for free.

People will behave the same before you make a policy and after you make a policy. Most people who work on causal inference know none of this is true, of course. And any seasoned machine learning engineer knows this as well when maintaining systems to continually retrain their stable of prediction models.

Dawid is of course not the first person to identify this problem. Fifty years ago, economist Robert Lucas pointed out that you can’t use historical data to predict the impact of economic policy because of feedback effects