Your Team's Football Is Not a Feeling — It Is a Framework

What Is a Game Model — Beyond Formations

Ask ten coaches what their game model is. You will get ten answers — and most of them will describe a formation.

That confusion is the starting point. A Game Model (GM) is not a 4-3-3 or a 3-5-2. It is not a pressing trigger or a build-up pattern. It is the foundational framework that sits beneath all of those choices, shaping why a team plays the way it does — not just how (Plakias, 2023).

The framework has four components. First, the coach’s ideas: principles of play, preferred style, system, and strategies adjusted across the key moments of the game — attack, defensive transition, defence, attacking transition, and set pieces. Second, player capabilities — technical, tactical, psychological, physiological, and sociological. Third, the club’s culture, structure, and goals. Fourth, national culture, the often invisible force that makes Brazilian football feel fundamentally different from Italian football even before a ball is kicked (Plakias, 2023).

The game model is therefore a framework for team organisation, not a tactical choice. It is the foundation for training design, individual player development, and, crucially, the core of Tactical Periodisation. A formation is a product of the game model. Treating it as the model itself is like mistaking a floor plan for an entire building.

Game Model ≠ Playing Style — A Conceptual Distinction

Here is where language creates real problems. Coaches, analysts, and researchers frequently use “game model” and “playing style” interchangeably. They should not.

Playing style refers to the characteristic patterns a team exhibits during match play — recurring behaviours that are observable, measurable, and relatively stable across specific contexts. Think of it as the team’s behavioural signature on the pitch. A grounded theory analysis across 22 studies extracted 84 distinct codes describing how teams play, clustering them into six categories: whole-team styles independent of match phase, styles tied to specific phases, sub-phases of attack, styles based on the dominant phase, and physical-performance-related styles (Plakias et al., 2024).

The game model is the superordinate concept. Playing style lives inside it. “Styles of play,” “playing style,” and “game style” are the most widely used terms in the literature, and researchers have proposed them as the common scientific language going forward (Plakias et al., 2024). Conflating them with “game model” is not just imprecise — it collapses a hierarchy that matters for both coaching practice and research design.

Why does this distinction matter on the training ground? Because a game model tells you what to build towards. A playing style tells you what the team is actually doing. The gap between the two is where coaching decisions live.

Methodologies for Identifying Team Playing Styles

If playing style is an observable pattern, the next question is straightforward: how do you measure it?

A scoping review of 40 studies mapped the landscape. Twenty-nine used classical inductive statistics — predominantly Factor Analysis through Principal Component Analysis (PCA) — while ten used AI techniques such as machine learning and deep learning. One study combined classical statistics, AI, and observational methods (Plakias et al., 2023a).

PCA emerged as the most suitable inductive method for style identification. It reduces a large set of performance indicators into a smaller number of underlying factors — each representing a dimension of playing style. When applied to event data (passes, shots, tackles, crosses) or tracking data (distances, speeds, spatial positions), PCA can reveal the latent structure behind how a team plays without requiring the analyst to pre-define what “styles” should look like.

AI is the newer frontier, but it remains underdeveloped. Only ten of the 40 reviewed studies applied machine learning or deep learning, and the field is still working through fundamental questions about feature selection, model interpretability, and validation against expert observation (Plakias et al., 2023a).

The data sources themselves carry limitations. Most studies relied on event data from the major European leagues and FIFA World Cups. Tracking data — richer in spatial and temporal detail — appeared in only eight studies. And critically, research on the effectiveness of different playing styles (does a given style actually win more matches?) remains scarce, with only seven studies addressing this question directly (Plakias et al., 2023a).

One thing is clear: the tools exist to move beyond gut feeling. The question is whether the field will use them rigorously enough.

The Uncharted Territory of Individual Player Playing Styles

Team styles get most of the attention. Individual player styles are a different story — one that has barely been written.

A systematic review found only 12 studies addressing individual playing styles, and nine of them used AI techniques. The research is recent: aside from one paper published in 2004, everything else appeared in the last four years (Plakias et al., 2023b).

The problems are structural. No study offered a position-based theoretical framework — a model that explains what different styles look like for a centre-back versus a winger versus a central midfielder. Zero percent of the reviewed studies examined how contextual variables (match status, home/away, opponent quality) influence individual style. Only 25% reported the reliability and validity of their data sources. And barely 42% differentiated between playing positions at all (Plakias et al., 2023b).

There is also a gap between disciplines. Sports science studies approach player style with different goals than data science studies. Sports scientists want to understand performance in context; data scientists want to classify and predict. Neither is wrong, but the two communities are not yet speaking the same language (Plakias et al., 2023b).

This matters because the practical stakes are high. If you want to recruit a player who fits your game model, you need to know their playing style — and you need that profile to be position-specific and context-sensitive. La Liga data shows that strong-team defenders participate far more in attacking organisation than weak-team defenders, and strong-team wingers cross less but contribute more overall in attack (Liu et al., 2015). A winger’s style cannot be described by one set of numbers regardless of team quality and tactical role.

The evidence from the English Premier League reinforces this point. When analysts use general positional categories (e.g., “wide defender”), they systematically over- or under-estimate what a player actually does. A full-back and a wing-back both fall under “wide defender,” yet a wing-back covers 15% more high-intensity running distance. Their tactical actions — overlapping runs, recovery runs, pressing frequency — differ substantially. Using broad categories masks these differences, leading to flawed benchmarks and misguided training prescriptions (Ju et al., 2023).

Why Coaches’ Eyes Fail to Agree

Here is a number that should concern anyone who relies on video analysis: when qualified coaches watched the same 20 minutes of Bundesliga footage and answered the same five questions about what they saw, their agreement was essentially zero.

Across three studies involving 15 to 24 coaches each, Fleiss-Kappa coefficients ranged from -.036 to .236. For context, a kappa of 0 means agreement no better than chance. A kappa of .236 — the highest recorded across all conditions — qualifies as “fair” at best (Furley et al., 2024).

The questions were not obscure. Coaches were asked to identify the opponent’s formation, describe their use of space, characterise their build-up play, note group tactical actions, and name the key player. On formation alone, 15 coaches produced 30 different responses. On build-up play, 39 distinct themes emerged from the same clip. Even when given a tactical overhead view instead of a standard broadcast angle, the highest-qualified coaches only reached 38% agreement on the formation — and that was the best-case scenario (Furley et al., 2024).

The implication is not that coaches are incompetent. It is that the task itself — extracting consistent tactical information from video — is far harder than we assume. Each coach brings their own mental model, their own attentional biases, and their own vocabulary to the same footage. Without a systematic framework for observation and classification, “what I see” will always differ from “what you see.”

This is precisely where quantitative playing-style identification methods earn their value. Not as replacements for coaching expertise, but as a shared language that reduces the noise.

Practical Implications — From Game Model to Training Design

So what changes when a team takes the game model seriously — not as a buzzword, but as a structural framework?

Training design shifts first. If the game model is the foundation of Tactical Periodisation, then every training session should express an element of it. Small-Sided Games (SSG) become design tools, not fitness drills — manipulated to rehearse the principles of play embedded in the model (Plakias, 2023). The constraints you set (pitch size, player numbers, rules of engagement) should reflect the behavioural patterns you want to see on match day.

Match preparation becomes more structured. Rather than relying on individual coaches’ subjective readings of opponent footage — readings that may barely overlap with each other — teams can anchor their analysis in quantitative style profiles. PCA-derived style dimensions give analysts a common reference point: does this opponent play a possession-heavy, slow-build style, or a direct, transition-focused one? The answer is no longer dependent on who watched the video.

Player recruitment gains precision. If playing style is measurable and position-specific, then scouting can move beyond “this player looks like they would fit” towards “this player’s behavioural profile aligns with our game model’s demands at this position.” The data from women’s international football illustrates how different contexts produce genuinely different styles — more direct play, less pressing, fewer off-the-ball movements — not as deficiencies but as context-specific adaptations shaped by physiological and tactical realities (Ju et al., 2025). Style is not universal. It is shaped by the system it lives within.

And self-evaluation improves. A team can compare what its game model prescribes against what its playing style data shows. The gap between intention and behaviour — visible in the data, not just in a coach’s post-match instinct — becomes the target for development.

A few things to hold in mind:

• The game model is a four-component framework (coach ideas, player capabilities, club culture, national culture) — not a formation or a single tactic.
• Playing style is observable, measurable, and sits inside the game model — not alongside it.
• PCA is the most established method for identifying team styles; AI is promising but still early.
• Individual player style research is severely underdeveloped — missing position-based frameworks, contextual variables, and cross-disciplinary integration.
• Coaches watching the same footage reach near-zero agreement on basic tactical questions — systematic methods are not optional.
• None of this replaces coaching judgment. It sharpens it.

The question is not whether your team has a playing style. Every team does. The question is whether you can describe it with enough precision to act on it — or whether “our football” remains a feeling that means something different to everyone in the room.

References

1. Furley, P., Mehta, S., Raabe, D., & Memmert, D. (2024). Objectivity of match analysis in football: Testing the level of agreement between coaches’ interpretations of video data. International Journal of Sports Science & Coaching, 20(1), 45-55. https://doi.org/10.1177/17479541241278603
2. Ju, W., Doran, D., Hawkins, R., Evans, M., Laws, A., & Bradley, P. (2023). Contextualised high-intensity running profiles of elite football players with reference to general and specialised tactical roles. Biology of Sport, 40(1), 291-301. https://doi.org/10.5114/biolsport.2023.116003
3. Ju, W., Kim, B., Lee, J., & Choi, S. (2025). Gender disparities in match performance at FIFA World Cups: More direct play, less off-ball movement, and less pressing in women’s matches. International Journal of Performance Analysis in Sport. https://doi.org/10.1080/24748668.2025.2523700
4. Liu, H., Gómez, M., Gonçalves, B., & Sampaio, J. (2015). Technical performance and match-to-match variation in elite football teams. Journal of Sports Sciences, 34(6), 509-518. https://doi.org/10.1080/02640414.2015.1117121
5. Morgulev, E. & Lebed, F. (2024). Beyond key performance indicators. German Journal of Exercise and Sport Research, 54(3), 335-340. https://doi.org/10.1007/s12662-024-00944-8
6. Plakias, S. (2023). An integrative review of the game model in soccer: Definition, misconceptions, and practical significance. Trends in Sport Sciences, 30(3), 55-66. https://doi.org/10.23829/TSS.2023.30.3-1
7. Plakias, S., Moustakidis, S., Kokkotis, C., Tsatalas, T., Papalexi, M., Plakias, D., Giakas, G., & Tsaopoulos, D. (2023a). Identifying soccer teams’ styles of play: A scoping and critical review. Journal of Functional Morphology and Kinesiology, 8(2), 39. https://doi.org/10.3390/jfmk8020039
8. Plakias, S., Moustakidis, S., Kokkotis, C., Papalexi, M., Tsatalas, T., Giakas, G., & Tsaopoulos, D. (2023b). Identifying soccer players’ playing styles: A systematic review. Journal of Functional Morphology and Kinesiology, 8(3), 104. https://doi.org/10.3390/jfmk8030104
9. Plakias, S., Kasioura, C., Pamboris, G. M., Kokkotis, C., Tsatalas, T., Moustakidis, S., Papalexi, M., Giakas, G., & Tsaopoulos, D. (2024). A grounded theory for professional soccer teams’ playing styles: Towards a consensus. International Journal of Sports Science & Coaching, 20(1), 159-172. https://doi.org/10.1177/17479541241300605