Soccer Analytics with Machine Learning

Chapter 7. Deep Learning for Soccer Analytics: From Intuition to Implementation

A Note for Early Release Readers

With Early Release ebooks, you get books in their earliest form—the author’s raw and unedited content as they write—so you can take advantage of these technologies long before the official release of these titles.

This will be the 7th chapter of the final book. Please note that the GitHub repo will be made active later on.

If you’d like to be actively involved in reviewing and commenting on this draft, please reach out to the editor at [e-mail].

So far in this book, we have mostly worked with models that depend on carefully chosen features and relatively direct relationships between inputs and outputs. That is an excellent place to begin. It teaches discipline, interpretability, and the logic of building models step by step. But it also has limits.

Soccer is not always kind enough to present itself in a few clean columns. Its signal is often fragmented across many interacting variables: player movement, match state, tactical structure, momentum, fatigue, and the subtle feedback loops that emerge over the course of a game. What matters is not always one stat, or even one cluster of stats, but the way many pieces combine. And in many cases, the most useful representation of that structure is not obvious in advance.

This is where deep learning becomes compelling. Deep learning is built around neural networks: flexible models that can learn layered representations ...