smarterscout: The Why in Analytics smarterscout
-
- Sports
-
Dan Altman, creator of smarterscout, discusses the big issues underlying football/soccer analytics and sports analytics in general. Dan asks why we use data, why we measure the things we do, and why we choose certain analytical methods out of the many tools available. Drawing on his experience working with clubs in the Premier League, Major League Soccer, and other competitions around the world – as well as USA Olympic teams – Dan gets to the core of the questions every analyst faces.
-
Q & A
Dan answers questions from listeners, followers on social media, and smarterscout.com users on a wide variety of topics: top-down versus bottom-up metrics, spotting potential in young players, hard-to-measure skills in defending, adjusting metrics for differences between leagues, how to get into football/soccer analytics, and more.
Support the show -
Measuring style and skill
Contributions to winning aren't the only things that football/soccer clubs might want to measure. Playing style has an important role in recruitment, and specific skills like finishing and winning aerial duels can be the crucial raw materials for tactical success – and for the careers of young players. Dan explains various approaches for gauging style and comparing players within and across positions, then moves on to the breakthroughs that pushed skill ratings forward.
Support the show -
Expected goals from ball progression and tactical applications
Each action during a match changes the probability of scoring, and these changes can become part of a model of ball progression. Dan explains how expected goals from ball progression can complement expected goal from shot creation, both in evaluating players and in predicting results. He also shows how both kinds of expected goals can form the basis for tactical tools, sometimes most effectively when they work in combination.
Support the show -
Rating players with expected goals from shot creation
Breaking down goal difference into its component parts shows how expected goals from shot creation – the chance of scoring for each shot, estimated from historical data – can be the foundation for a model of the game. Dan explains how this model can be used to rate both attacking and defending, then considers the judgment calls implicit in using this kind of model to evaluate players.
Intro: "Crackle & Pop" by Digital Primitives
Outro: "No Holiday" by Digital Primitives
Support the show -
The building blocks of a model of the game
There's an infinite number of mathematical models that might explain the game of football/soccer, but how can we strike a balance between usefulness and simplicity? Dan considers the advantages and drawbacks of models using goals, and shots, then begins to explain how expected goals – a measure of shots' chances of scoring – might capture information for rating teams and players.
Intro: "Crackle & Pop" by Digital Primitives
Outro: "No Holiday" by Digital Primitives
Support the show -
The basics (and basis) of metrics
The first step in constructing a metric is deciding – and clearly specifying – the question to be answered by the data. Dan talks about formulating this question and dealing with the complex issues generated by even the simplest metrics. As part of the discussion, Dan asks, "Who is the best tackler in the Premier League?" To answer this question... well, how long do you have?
Intro: "Crackle & Pop" by Digital Primitives
Outro: "No Holiday" by Digital Primitives
Support the show
Customer Reviews
Informative
Look, Dan is in the weeds of understanding soccer/football/fútbol through data. It’s very informative.