Evaluating the “Players First” Game Model Using Large-Scale Tracking Data

Abstract

Introduction & Purpose

Game models – integrated collections of principles that determine a desired team identity and player behaviours – are paramount in modern elite football coaching. In practice, quantitative methods typically operate on untested ontologies derived from pre-existing game models (Corsie et al., 2024; Rocha & Lopes, 2025). This work pilots an alternative paradigm evaluating game models from an outside perspective. We evaluate Players First (PF), a game model of the Austrian Football Federation (ÖFB) that assigns players into four in-possession and four out-of-possession roles (“PF roles”) based on their positions relative to the ball and the penalty area. We examine whether PF roles distinguish teams, and whether the role-specific tasks recommended by PF are sound and consistent. For the latter analysis, we use Dangerous Accessible Space (DAS) (Bischofberger & Baca, 2026), relating positional geometry to tactical meaning.

Methods

We use positional and event data from 402 matches across major European competitions: the Men’s European Championship 2024, the Women’s European Championship 2025, the Nations League 2025, and the European Qualifiers for the Men’s World Cup 2026. PF roles and tasks are operationalized according to the Players First specification (Figure 1). A PERMANOVA shows whether the role configuration in individual tracking frames differs (a) between teams and (b) between formations. Furthermore, we evaluate PF’s role-level recommendations using geometric analyses and DAS-based metrics. DAS and role counts are corrected by longitudinal ball position using LOESS regression.

Results

PERMANOVA shows that team identity explains significant variance in offensive PF role distribution (R² = 0.29, F(53,740) = 6.05, p < 0.001) while the starting formation explains none (R² = 0.01, F = 1.48, p = 0.122). Role counts correlate with team strength: stronger teams deploy more Stretching players (r = 0.68) and fewer Preventing players (r = -0.59). Defensively, stronger teams tend to have fewer Defending (r = -0.56), more Controlling (r = 0.71), and fewer Counterattacking (r = -0.42) players (all p < 0.001). On the player level, consistency is mostly established: As prescribed, Preventing players have the highest Accessible Space (676m²), Defending players have the most DAS Covered (0.61m²; 0.40m²), and Defensive DAS of Counterattacking players increases when positioned closer to the opponent goal (F(3,1916329) = 313580, p < 0.001, η²_p = 0.146). Some tests show weaker or non-linear effects such as the influence of Stretchers on defensive width (η²_p = 0.012).

Discussion

The significant differences of PF role configuration across teams suggests formation-independent tactical features. This view is supported by substantial correlations with team quality. The tasks recommended by PF are shown to be plausible although some follow directly from role geometry.

Conclusion

We attempted the first published validation of an elite level game model in football, finding that Players First proposes a reasonable assignment of roles that shows associations with team style and success. Many of its recommendations on the level of individual roles appear plausible as measured by DAS. We hope this work inspires future research to support the development of coaches’ game models instead of taking them for granted.

References

Bischofberger, J., & Baca, A. (2026). Dangerous accessible space: A unified model of space and value in team sports. Journal of Big Data, 13(1), Article 76. https://doi.org/10.1186/s40537-026-01387-8

Corsie, M., Malone, M., & Swinton, P. A. (2024). Translating novel collective behaviour measures to concepts and principles of play as understood by football coaches. International Journal of Sports Science & Coaching, 19(4), 1443–1458. https://doi.org/10.1177/17479541241238991

Rocha, R. F., & Lopes, R. J. (2025). Assessing and visualizing principles of play in soccer. In Proceedings of the 13th International Conference on Sport Sciences Research and Technology Support (icSPORTS 2025) (pp. 90–99). SciTePress. https://doi.org/10.5220/0013716200003988

Figure caption

Figure 1: Geometric construction of the (A) in-possession roles and (B) out-of-possession roles prescribed by Players First.