11  Analytical Methods in Sport Data Analytics

11.1 Introduction

As we’ve discussed in class, analytical methods in sport data analytics have evolved significantly over the past three decades, both in professional sporting context and within academic research .

Currently, key concepts (e.g., expected value of game states, win probability, and team strength) are applied across multiple sports using the statistical methodologies, computational tools, and diverse data sources we have and will encounter within the MSc SDA to generate actionable insights.

In football, for example, traditional notational analysis, which focuses on video-based event summaries, is being enhanced by technological advancements that allow dynamic analysis of complex variables (Robberechts et al, 2021). Machine learning, a newer tool in data analytics (and which we’ll cover in Semester 2), shows potential in identifying key performance indicators through the analysis of positional data, although its application has largely focused on predictive models rather than tactical improvements.

The increasing role of machine learning and advanced statistical tools suggests a shift toward more sophisticated, context-driven analysis that can enhance both theoretical understanding and practical performance in sport.

References

• Robberechts, P., Van Haaren, J., & Davis, J. (2021). A Bayesian approach to in-game win probability in soccer. Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. https://www.semanticscholar.org/paper/A-Bayesian-Approach-to-In-Game-Win-Probability-in-Robberechts-Haaren/842f17ef1871c70879bc9f452514e983ce49229d

11.2 Guided Reading

Both papers for this week are available for reading and download via the module reading list, which can be accessed via myplace.

• Baumer, B.S., Matthews, G.J. & Nguyen, Q. (2023) Big ideas in sports analytics and statistical tools for their investigation. Wiley interdisciplinary reviews. Computational statistics. doi:10.1002/wics.1612.

• Herold, M., Goes, F., Nopp, S., Bauer, P., Thompson, C. & Meyer, T. (2019) Machine learning in men’s professional football: Current applications and future directions for improving attacking play. International Journal of Sports Science & Coaching. 14 (6), 798–817. doi:10.1177/1747954119879350.

Key Observations

A number of key themes can be identified in our reading this week:

• The expansion of sport analytics

  • Sport analytics has grown significantly over the past 30 years, impacting both the professional sports industry and academia.

  • This growth is seen across multiple sports, highlighting common analytical concepts such as expected value of a game state, win probability, team strength, and sports betting market data.

• Analytical approaches and tools

  • Different sports share similar foundational concepts but differ in their analytical approaches.

  • Contemporary analysis involves statistical methods, computational tools, and diverse data sources, enabling the generation of actionable insights that can improve team and game performance.

• Focus on football

  • Football analysis has traditionally relied on notational analysis to study the sport, focusing on video footage to assess attacking play and match outcomes.

  • Recent technological advancements allow for the analysis of complex, dynamic variables, incorporating contextual parameters for a more in-depth evaluation of team performance.

• The role of machine learning in football

  • Machine learning, though relatively new in football, shows potential in identifying performance metrics through algorithmic analysis of positional data.

  • Despite its promise, machine learning research in football has primarily centered around predictive modeling rather than improving tactical understanding and performance.

  • Future directions for machine learning in football focus on overcoming current challenges to offer deeper tactical insights.

• Interdisciplinary relevance

  • Both papers emphasise the intersection of data science and sport, with a focus on using advanced statistical and computational methods to derive insights from complex datasets.

  • The growing role of machine learning highlights a shift toward more sophisticated, context-driven analysis that could enhance both theoretical understanding and practical application in various sports.

11.3 Questions for Reflection

• How could the integration of machine learning in sport data analytics move beyond predictive modeling to provide deeper insights into tactical strategies and player performance across different sports?

• What, for you, are the potential limitations or challenges in applying the expected value of game states, win probability, and team strength across different sporting disciplines, given the idiosyncrasies of different sports?

• In what ways could advanced analytical methods, such as dynamic contextual analysis in football, reshape traditional approaches like notational analysis to provide a more comprehensive understanding of game performance?