Case Study 1: Manchester City's Tactical Evolution Under Guardiola: A Data-Driven Analysis

Overview

Pep Guardiola arrived at Manchester City in the summer of 2016, inheriting a squad built by Manuel Pellegrini with considerable talent but a style that leaned toward direct, transition-based attacking play. Over the subsequent seasons, Guardiola reshaped City into one of the most dominant possession-based teams in Premier League history, winning multiple league titles and achieving a historic treble in 2022-23.

This case study uses the team style identification framework from Section 16.1 and the squad balance tools from Section 16.3 to quantify Guardiola's tactical transformation---tracking how City's style fingerprint evolved season by season, how squad composition changes supported tactical shifts, and how the team's chemistry metrics reflected the cohesion of Guardiola's system.

Data and Methods

Data Sources

We use match-level event data from the 2015-16 season (Pellegrini's final season) through 2022-23 (the treble season), comprising approximately 300 Premier League matches. For each match, we compute:

  • Possession percentage
  • PPDA (passes per defensive action in the attacking 40%)
  • Average pass sequence length
  • Direct speed of attacks (meters per second toward goal)
  • High press recoveries (turnovers won in the opposition third)
  • Defensive action height (average y-coordinate of defensive actions)
  • Cross frequency per 90
  • Set-piece xG share

Style Fingerprint Construction

Following Section 16.1.2, we standardize all metrics to z-scores using league-wide means and standard deviations for each season, producing a 6-dimensional style vector for each season. We also compute rolling 10-match style vectors to capture within-season evolution.

Results

Season-by-Season Style Evolution

2015-16 (Pellegrini): City's baseline style was moderate across all dimensions---slightly above average possession (54%), moderate pressing (PPDA: 12.2), average directness, and high cross frequency. The style fingerprint shows a team with no extreme tendencies---competent but tactically generic.

2016-17 (Guardiola Year 1): The most dramatic single-season shift in our dataset. Possession surged to 60.2% (+1.8 SD). PPDA dropped to 9.8 (-1.3 SD), indicating much more intense pressing. However, directness initially increased as Guardiola's pressing created turnovers in advanced positions that led to fast attacks. The team struggled defensively (defensive line height moved very high, but the center-back quality could not support it), finishing 3rd.

Style drift from 2015-16 to 2016-17: $D = 3.21$ (well above the league average of 0.8 for year-on-year change).

2017-18 (Guardiola Year 2): After major squad investment (Kyle Walker, Benjamin Mendy, Ederson, Bernardo Silva), the style crystallized. Possession reached 63.8% (+2.4 SD). Pressing remained intense but became more coordinated (high press recoveries per match rose from 9.2 to 11.1). Directness decreased as the team shifted toward patient, positional play. City won the league with 100 points.

This season showed the highest Team Chemistry Score in our dataset: 0.82 (league average: 0.61), driven by the De Bruyne-Silva-Sterling-Sane attacking combinations and the Walker-Stones-Otamendi-Mendy/Delph defensive unit that had months to gel.

2018-19 (Guardiola Year 3): Stylistic refinement rather than revolution. The key change was increased width---Guardiola deployed full-backs in more advanced, inverted positions, stretching the opposition and creating interior passing angles. The width dimension increased from +0.4 SD to +1.6 SD.

Style drift from Year 2 to Year 3: $D = 0.95$, indicating tactical stability with minor adjustments.

2019-20 (Guardiola Year 4): The loss of Vincent Kompany and Aymeric Laporte's long-term injury disrupted the defensive structure. Defensive line height remained high but team compactness decreased, indicating disorganization without the leader (Kompany) and the ball-playing center-back (Laporte). City's SDI at center-back dropped to 0.58, the lowest in the top 6. They finished second, 18 points behind Liverpool.

2020-21 (Guardiola Year 5): The signing of Ruben Dias transformed the defense. The team became more conservative in their pressing (PPDA rose to 11.4) but more compact and organized defensively. This was Guardiola's most "balanced" City team---no extreme style dimensions but elite execution across the board. SDI at center-back recovered to 0.81 with Dias, Stones, Laporte, and Ake.

2021-22 and 2022-23: The final evolution toward the "false 9" system, with the subsequent arrival of Erling Haaland dramatically increasing directness and xG per match while maintaining possession dominance. The style fingerprint showed an unprecedented combination: top-2 in possession, top-3 in pressing, and top-2 in xG---dimensions that are typically negatively correlated across the league.

Key Findings

  1. Non-linear transformation: Guardiola's style change was not gradual. Year 1 showed a massive jump, followed by refinement. The "big bang" approach to tactical change is visible in the style drift metric.

  2. Squad investment follows style: Transfer activity was systematically aligned with style evolution. The 2017 summer signings (Walker, Ederson, Mendy, Bernardo Silva) were precisely targeted at positions where the 2016-17 style had been limited by personnel.

  3. Chemistry lags style change: The Team Chemistry Score was lowest in Guardiola's first season (0.55) and peaked in his second (0.82), confirming that tactical changes precede chemistry development by approximately one season.

  4. Depth as insurance: The seasons where City underperformed (2016-17, 2019-20) were precisely the seasons where SDI was weakest at critical positions.

  5. The Haaland inflection: The 2022-23 season showed that Guardiola could successfully integrate a "non-Guardiola" player type (a traditional striker) by adapting the system---directness increased without sacrificing possession, demonstrating tactical flexibility within an established framework.

Discussion

The Coach Effect vs. the Player Effect

One limitation of this analysis is the difficulty of separating coaching influence from player quality. Did City become more possession-dominant because Guardiola demanded it, or because he signed players who are naturally possession-oriented? The answer is almost certainly both, and the interaction is the key insight: Guardiola's recruitment strategy was inseparable from his tactical vision.

The style clustering analysis reveals that City's signings (Bernardo Silva from Monaco, Rodri from Atletico Madrid, Grealish from Aston Villa) came from diverse tactical backgrounds, yet all integrated quickly into City's system---suggesting that Guardiola's coaching, not just player selection, drives the style convergence.

Replicability

Other managers have attempted to replicate Guardiola's style (Arteta at Arsenal, Maresca at Leicester/Chelsea). Using the same style fingerprint framework, we can measure how closely their teams approximate City's profile. Arteta's Arsenal by 2022-23 was the closest match in the Premier League: Euclidean distance of 1.2 from City's fingerprint, compared to 3.5+ for most other teams.

Methodology Notes

The analysis uses the code provided in code/case-study-code.py, which implements: - compute_season_style(): Aggregates match-level events to season-level style vectors - compute_rolling_style(): Computes rolling 10-match style vectors - plot_style_evolution(): Creates multi-panel radar charts showing season-by-season evolution - compute_style_drift(): Measures the Euclidean distance between consecutive season vectors - compute_team_chemistry(): Aggregates pairwise passing chemistry from event data

Exercises

  1. Using the framework developed in this case study, perform a similar analysis for another long-tenured manager (e.g., Klopp at Liverpool 2015-2024, Simeone at Atletico Madrid 2012-present, or Ancelotti at Real Madrid 2021-2024).

  2. Compare the style drift associated with a managerial change (e.g., Wenger to Emery to Arteta at Arsenal) with the within-manager style drift observed in this study. Is a managerial change always a larger disruption than within-manager evolution?

  3. Using the squad depth framework, identify the "weakest link" in City's squad for each season and evaluate whether it correlates with their competitive outcome (title won vs. not).

References

  • Dixon, M. J., & Coles, S. G. (1997). Modelling association football scores and inefficiencies in the football betting market. Journal of the Royal Statistical Society: Series C, 46(2), 265-280.
  • Guardiola, P. (various interviews). Post-match press conferences, Manchester City FC.
  • Fernandez, J., & Bornn, L. (2018). Wide open spaces: A statistical technique for measuring space creation in professional soccer. MIT Sloan Sports Analytics Conference.
  • Herold, M., et al. (2022). Machine learning approach to identify team performance indicators in professional football. International Journal of Sports Science & Coaching, 17(3), 531-544.