Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Neighbor-based Approach to Pitch Ownership Models in Soccer

Tiago Mendes-Neves, Luís Meireles, João Mendes-Moreira

TL;DR

The paper addresses pitch ownership modeling in soccer using tracking data and introduces a KNN-based framework for pitch control. It leverages a small set of hyperparameters, namely the $lags$, $\xi$, and $\tau$, to emulate multiple literature approaches while accommodating uncertainty. The method uses KNN to approximate Voronoi regions, incorporates distance-based uncertainty, and applies smoothing to reflect contested spaces, enabling fast inference. While visually demonstrating the approach and its flexibility, the authors acknowledge limitations such as body blocking and the need for quantitative evaluation, and provide open-source code for reproducibility.

Abstract

Pitch ownership models allow many types of analysis in soccer and provide valuable assistance to tactical analysts in understanding the game's dynamics. The novelty they provide over event-based analysis is that tracking data incorporates context that event-based data does not possess, like player positioning. This paper proposes a novel approach to building pitch ownership models in soccer games using the K-Nearest Neighbors (KNN) algorithm. Our approach provides a fast inference mechanism that can model different approaches to pitch control using the same algorithm. Despite its flexibility, it uses only three hyperparameters to tune the model, facilitating the tuning process for different player skill levels. The flexibility of the approach allows for the emulation of different methods available in the literature by adjusting a small number of parameters, including adjusting for different levels of uncertainty. In summary, the proposed model provides a new and more flexible strategy for building pitch ownership models, extending beyond just replicating existing algorithms, and can provide valuable insights for tactical analysts and open up new avenues for future research. We thoroughly visualize several examples demonstrating the presented models' strengths and weaknesses. The code is available at github.com/nvsclub/KNNPitchControl.

A Neighbor-based Approach to Pitch Ownership Models in Soccer

TL;DR

The paper addresses pitch ownership modeling in soccer using tracking data and introduces a KNN-based framework for pitch control. It leverages a small set of hyperparameters, namely the , , and , to emulate multiple literature approaches while accommodating uncertainty. The method uses KNN to approximate Voronoi regions, incorporates distance-based uncertainty, and applies smoothing to reflect contested spaces, enabling fast inference. While visually demonstrating the approach and its flexibility, the authors acknowledge limitations such as body blocking and the need for quantitative evaluation, and provide open-source code for reproducibility.

Abstract

Pitch ownership models allow many types of analysis in soccer and provide valuable assistance to tactical analysts in understanding the game's dynamics. The novelty they provide over event-based analysis is that tracking data incorporates context that event-based data does not possess, like player positioning. This paper proposes a novel approach to building pitch ownership models in soccer games using the K-Nearest Neighbors (KNN) algorithm. Our approach provides a fast inference mechanism that can model different approaches to pitch control using the same algorithm. Despite its flexibility, it uses only three hyperparameters to tune the model, facilitating the tuning process for different player skill levels. The flexibility of the approach allows for the emulation of different methods available in the literature by adjusting a small number of parameters, including adjusting for different levels of uncertainty. In summary, the proposed model provides a new and more flexible strategy for building pitch ownership models, extending beyond just replicating existing algorithms, and can provide valuable insights for tactical analysts and open up new avenues for future research. We thoroughly visualize several examples demonstrating the presented models' strengths and weaknesses. The code is available at github.com/nvsclub/KNNPitchControl.
Paper Structure (12 sections, 4 equations, 5 figures)

This paper contains 12 sections, 4 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Pitch Ownership Model
  3. KNN-based Pitch Control
  4. Concepts
  5. Approximating Voronoi diagrams with KNN
  6. Introducing uncertainty with KNN's distance
  7. Introducing uncertainty with smoothing
  8. Combining the concepts
  9. Methodology
  10. Results
  11. Discussion
  12. Conclusion

Figures (5)

  • Figure 1: A visualization of the pitch control surface as described by Spearman - implementation by Laurie Shaw. The blue team dominates blue areas. The red team dominates red areas. Light-colored areas have higher uncertainty over who controls the area. White areas indicate that there is complete uncertainty over who controls this area.
  • Figure 2: An approximation of the Voronoi diagram using the KNN algorithm, for a resolution of $1m^2$.
  • Figure 3: The effect of modeling distance on the Voronoi diagram.
  • Figure 4: The areas of uncertainty obtained when smoothing are visible in the lighter spaces.
  • Figure 5: The results obtained with KNN-based Pitch Control compared with the original approaches, according to the parameters set in Section \ref{['sec:methodology']}. On the left, we present our approaches; on the right, we present the similar approaches available in the literature. The results we present are from the play Barcelona 1 - [2] Real Madrid on the 21st of April 2012, frame 132.