Tactical Performance Data Analyses in Elite Football

Deb, B (2026) Tactical Performance Data Analyses in Elite Football. Doctoral thesis, Liverpool John Moores University.

Preview

Text 2026debphd.pdf - Published Version Available under License Creative Commons Attribution Non-commercial. Download (4MB) | Preview

Abstract

Background
The analysis of strategies and tactics in elite football has seen an increasingly growing disconnect between advances in coaching methodology, performance analysis, and football analytics, despite improvements in technology and data availability. Significant methodological gaps remain in defining, classifying, and evaluating key tactical events (as measured by performance analysts), such as penetrative passes and game context, such as phases of play. Existing methodologies often lack comprehensive validation, repeatability, and integration with the tactical frameworks used by coaches and analysts. This doctoral thesis addresses these research gaps by utilising novel analytical frameworks, robust validation approaches, and developing a possession framework to serve as a foundation for future research in the field.

Methods
A series of retrospective studies were conducted using event data, merged event-tracking data, and expert notational analysis data within elite football, comprising international football matches between 2018 and 2022 and Premier League data from the 2021-2022 season. Chapter 2 utilised event data to define progressive passes, applying the novel Separation-Concordance (SeCo) framework to cluster these passes into repeatable, interpretable groups, allowing for risk-reward profiling. Chapter 3 integrated merged event-tracking data to automatically detect opposition defensive lines, creating spatiotemporal features for classifying penetrative passes into specific tactical categories. This rules-based model was validated against a gold-standard manual notational analysis. Chapter 4 developed an augmented possession framework combining event data with phases of play derived from expert notational analysis, assessing the efficacy of progressive passing across various tactical scenarios through multivariate logistic regression. Finally, Chapter 5 synthesised these various methodologies, demonstrating their applications in multiple case studies and how this can be implemented in applied practice at scale using a cloud technology stack.

Results
In Chapter 2, the application of the SeCo framework identified stable clusters of progressive passes, notably highlighting types such as "Mid Central to Mid Half Space" during build-up phases and "Mid Half Space to Final Central" into the attacking third balancing risk (turnover) and reward (shot creation) facilitating opposition profiling during the 2022 World Cup.

Chapter 3 demonstrated the classification of penetrative passes using spatiotemporal clustering methods, achieving high validation accuracy against notational analysis data (accuracy rates: behind 1st defensive unit = 0.92; 2nd unit = 0.87; beyond defensive shape = 0.80). These results enhanced tactical analysis by quantifying opposition shape and the disruption of defensive lines through penetrative passes.

The augmented possession framework introduced in Chapter 4 enabled evaluation of tactical efficacy across match phases. Progressive passes significantly increased the odds of achieving positive phase outcomes, including successful phase progressions (aOR 2.57, 95% CI 2.19-3.02), dangerous possessions (aOR 1.79, 95% CI 1.56-2.04) and critical chances (aOR 1.91, 95% CI 1.27-2.86). This analysis further revealed strategic insights such as the value of non-progressive switches of play in 1st Phase when under opposition pressure, highlighting tactical nuances missed by traditional possession value models.

Conclusion
This thesis provides significant methodological advancements in football analytics by effectively integrating data science with tactical concepts familiar to coaches and analysts. The SeCo framework ensures robust, repeatable classification of progressive passes. Alongside automatic classification of penetrative passes using merged event-tracking data, validation with notational analysis offers validated football metrics and definitions to measure player and team performance. The augmented possession framework bridges the gap between event-based analytics and tactical outcomes, offering novel, contextually rich insights into team performance and tactical strategies.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	football analytics; phases of play; passing taxonomy
Subjects:	G Geography. Anthropology. Recreation > GV Recreation Leisure > GV561 Sports > GV711 Coaching G Geography. Anthropology. Recreation > GV Recreation Leisure > GV561 Sports
Divisions:	Sport and Exercise Sciences
Date of acceptance:	16 July 2026
Date of first compliant Open Access:	8 September 2026
Date Deposited:	08 Sep 2026 10:52
Last Modified:	08 Sep 2026 10:52
Supervisors:	McRobert, AP, Fernández Navarro, J and Jarman, I
URI:	https://researchonline.ljmu.ac.uk/id/eprint/29067

View Item