Applied Machine Learning
Machine learning can be a powerful tool for gaining insight, especially from tracking data.
From data to on-field performance
Machine learned models are uniquely effective at finding patterns in tracking data. When linked to video this is a potent recipe for streamlining video analysis. Some of the patterns I have trained models to identify include off-ball runs, pressing, overloads, and phases of play.
At Second Spectrum I was the product owner of the reports in their soccer performance platform. As the platform’s starting point for analysis, the reports were loaded with intuitive visualizations that described shape, style, and strengths.
See how the above images contributed to post-match analysis: left, right!
Effective analysis of a fluid sport such as soccer requires creativity. At Second Spectrum I developed novel approaches to quantifying player acceleration and agility, player usage, and the on-ball value added by players.