Identifying playing styles in men's tennis
Our purpose was to identify playing styles in men's tennis to help underpin a ‘What Takes to Win’ model.
The problem was that there are multiple game styles to serve against each other and as we humans are visual creatures, we needed a simple visualisation to present to what the dataset is saying.
Using dimension reduction techniques and clustering this allowed us to identify the key performance indicators for players and associated game styles such as ‘All Court Player’, ‘Counter Puncher’, Aggressive Baseliner’ or ‘Big Server’.
Dimension reduction is a machine learning technique that reduces the number of attributes in a dataset whilst keeping as much of the relevant variation as possible. Essentially, removing/reducing features that aren’t as equally important and finding a low dimensional representation that captures the essence of the data.
The Data Science Methods
The Data Science methods helped to identify distinctions and similarities in game styles, as well as the KPIs associated with each game style. This work provides a foundation to answer further questions around Head-to-Heads as well as player development.
Questions to consider
How does this impact Player Development?
There is value in understanding how a player's game styles change over time, for reasons including player development, opponent analysis and game understanding.
- Should players be capable of playing in multiple different ways? How important is it to do this?
- How quickly can/should players change their game style?
- Are some game styles better against others? A Rock, Paper, Scissors scenario?
- How can game styles be implemented in a player's development?
- Are there specific weapons exclusive to game styles?
We would love to have an informal chat and share our knowledge and expertise with you.