The Relationship Between Playing Formations, Team Ranking, and Physical Performance in the Serie A Soccer League

1. Introduction

The execution of soccer-related bouts requires large a huge physiological load on players during competition. Research indicates that activity profile is role-positioning dependent [1,2,3,4,5,6] and contextual variables such as the playing formation [7,8] as well as the ranking of the opponents [9], can significantly affect the locomotor activity of professional soccer players. Although some research has explored the influence of these variables on locomotor activity, as Plakias & Michailidis [9] pointed out, the findings remain contradictory. Certain studies reported that team quality does not significantly impact running performance [10], whereas others reported the opposite [11]. Similar contradictions emerge when considering the opponents’ level. In fact, Modric et al. [10] found no significant effect of opponents’ level on locomotor activity, while Gonçalves et al. [12,13] observed that facing strong opponents increases the total distance covered by a team.

The playing formation and the playing style (e.g., defensive, direct, possession-based) used in the same playing formation affect the physical performance [7]. However, Bradley et al. [14] reported that high- and very high-intensity running distances (e.g., running over 20km/h) were similar in 4-4-2, 4-3-3 and 4-5-1 formations when ball possession was not considered.

When analyzing the locomotor activity of soccer players, it is crucial to recognize the association between high-intensity activities and the most decisive soccer game events [15]. Consequently, high-intensity actions warrant careful consideration in such analysis.

One of the high-intensity activities is represented by high-intensity running, which is a crucial element of soccer performance. Moreover, it serves as a valuable indicator of physical performance in soccer [6], differentiating various levels of play [6,16], the tactical role of players [17,18], and fluctuations throughout the competitive season [6]. It is even sensitive to physiological changes associated with the end of a training program [19].

The traditional speed-category approach, neglecting acceleration and deceleration, provides only a partial understanding of the actual game’s physiological and external load experienced during a match [20,21]. By considering the energy expenditure estimated from acceleration, deceleration and speed, following the method proposed by Osgnach et al. [20], a more comprehensive description of match demands has been possible. Osgnach’s method [20] quantifies players’ activity as the distance covered within arbitrarily chosen energy-expenditure categories, referred to as metabolic power (MP).

Greig and Siegler [22] highlight the importance of sprinting and acceleration in contributing to muscular fatigue due to their high neuromuscular demand. However, using absolute acceleration thresholds can lead to misclassification of high-intensity acceleration events, underestimating those with high initial running speed and overestimating those with low initial speed [23].

With the method proposed by Sonderegger et al. [24] it has been possible to consider the initial running speed and the population-specific maximal acceleration values at various initial speeds, thus improving the accuracy of detecting high-intensity acceleration actions.

Video match analysis is a valuable tool for evaluating soccer players’ performance. This technique, initially introduced and used to monitor the work-rate profiles of elite players [17,25], has become indispensable for assessing the physical and tactical behavior in training and competition. It enables complex analytical evaluations on a large sample size. In fact, a multiple-camera video system is pivotal in the analysis of high-intensity bouts, where detailed information can be collected [3].

Given the contradictory findings in previous research, as highlighted by Plakias & Michailidis [9] in their analysis of the Turkish first division soccer data, this exploratory study aims to investigate how ranking and playing formations influence the physical exertion of professional soccer players in the Italian First Division (Serie A). The secondary aim of this study was to compile data attained by professional soccer players considering different roles and playing formations to provide benchmarks and to facilitate the interpretation of the locomotor activity level of players.

2. Materials and Methods

2.1. Sample

We analyzed through semi-automatic tracking 212 Professional soccer players from 20 Italian Serie A teams. A total of 375 players match data were analyzed in this study, comprising 88 attackers, 74 box-to-box midfielders, 97 central defenders, 30 central midfielders, 32 wide defenders and 54 wide midfielders. Goalkeepers were not included in this investigation. Data were collected from all official home matches played by a single team, along with the corresponding matches of their opponents, using video match analysis. Only players who participated in the entire match (85-95 minutes) were included in the analysis. Data from players whose playing time fell outside this range was excluded (e.g., red card incidents).

2.2. Procedure

Teams were categorized into high (HR), medium (MR), and low (LR) ranking based on their final standing in the Italian championship: 1st-7th (HR), 8th-14th (MR), and 15th-20th (LR). The playing formations analyzed were 4-4-2, 3-4-3, 4-3-3, and 3-5-2. Comparisons among different team playing formations, both within and across the ranking categories, were conducted.

For the second aim of the study, the T-score method was employed to provide benchmarks and to facilitate the interpretation of the locomotor activity level of players [26]. The T-score offers a more intuitive alternative to the z-score [27], calculated as: (Z-score x 10) + 50, with a score of 50 rather than 0, equaling the mean. For enhanced interpretation, these T-score values were combined with qualitative descriptions ranging from “extremely poor” (<20) to “excellent” (>80).

The following kinematic variables were analyzed: average metabolic power (AMP, w·kg^-1), average speed (AS, m·min^-1), high metabolic power distance (HMPD, >20w·kg^-1), very high metabolic power distance (VHMPD, >35w·kg^-1), high-speed running distance (HSR, distance covered above 20km/h), and finally, very high-speed running distance (VHSR, distance covered at more than 25km/h). Accelerations events were defined based on Sonderegger’s equation [24] modified by Savoia et al. [28], where an event was considered an acceleration if it exceeded 50% of the a_max achievable by the player considering the initial speed. High acceleration data were defined as the percentage of the total acceleration time (H-acc). High decelerations were defined as a percentage of the total deceleration time through an absolute threshold (greater than 2m·s^-2, H-dec).

Missing data or data that did not meet the inclusion criteria were excluded. Subsequently, the players were categorized based on their playing formation and role, as shown in Table 1.

The experimental procedures were approved by the local Human Ethics Committee of Liverpool John Moores University (No. 12/SPS/003). The study complied with the Declaration of Helsinki.

2.3. Video Match Analysis

Match analysis was performed using the validated multi-camera video analysis system Stats Perform’s SportVU (Stats Perform, Chicago, US), tracking at up to 25-Hz rates. The Technical University of Munich (TUM) determined the measurement accuracy of this device with a typical error of 2.7% for total distance [29]. Raw data were provided via cartesian coordinates by K-Sport (K-Sport World SRL) and primary data have been smoothed at 5-Hz. The Stats SportVU tracking system transports the data of performance by extracting and processing coordinates of players (X, Y) and the ball (X, Y, Z) through HD cameras as well as sophisticated software and statistical algorithms [29]. Player movements were captured during matches through cameras located at the roof level. Data were analyzed using STATS Viewer and K-Sport Dynamix, and through K-Filter software package (K-Sport World SRL) processed to create a dataset on each player’s physical and technical performance.

2.4. Statistical Analysis

A Shapiro-Wilk test was used to test the normal distribution of the data. Not following a normal distribution, a non-parametric statistical analysis was applied to the data. Comparisons between groups were accomplished by the Krustal-Wallies [30] test, that is a valid non-parametric alternative to one-way ANOVA. It extends the two-samples Wilcoxon test when there are more than two groups to compare. When the p-value was < 0.05, the Dunn test [31], with Bonferroni adjustment, was applied to discriminate which group was different from the other. The Epsilon squared (η²) was reported as effect size (ES) according to Tomczak & Tomczak [32], η² ≤ 0.06 (small effect), 0.06 < η² < 0.14 (moderate effect); η² ≥ 0.14 (large effect). Significance was accepted at an alpha level of p ≤ 0.05. All statistical analyses were performed using R (version 4.1.1) [33] and the package rstatix [34].

3. Results

The first comparison was conducted to see if there were any differences between the teams according to their position in the rankings. Results are synthesized in Table 2, Table 3, Table 4 and Table 5.

Significant statistical differences (p <.001) were found among different rankings for the variables H-acc and AS. Dunn test with Bonferroni adjustment showed that teams in the MR reported a lower H-acc than HR and LR (ES=0.05), while there were no significant differences between HR and LR. Moreover, HR teams reported lower AS than MR and LR teams (ES=0.06), with no differences, in this case, between MR and LR.

In Table 2, where differences among playing formations within the same ranking group were assessed, statistical differences were also detected. In the HR group, differences were found for H-acc and H-dec. Specifically, the H-dec 3-4-3 formation yielded lower results compared to the 3-5-2 and 4-3-3 formations (ES=0.05). Whereas for H-acc 4-3-3 has value higher than 3-4-3, 3-5-2 and 4-4-2. No statistical differences were established among variables in the MR group. Finally, in the LR group, the only difference found was for H-acc, where 4-3-3 < 4-4-2 playing formation (ES=0.16). No other statistical differences were detected.

The t-score values combined with qualitative description for each formation and role are reported in Table 6, Table 7, Table 8 and Table 9.

4. Discussion

The aim of this study was to investigate the influence of team ranking and playing formation on the locomotor activity of professional soccer players in the Italian First Division. Additionally, the study also aimed to establish benchmarks combined with qualitative descriptors to provide insight into role-specific locomotor activity of players and to help defining performance levels as above or below average.

4.1. Differences Among Rankings

Only three statistical differences were detected when different ranked teams were analyzed. HR teams reported more H-acc than the MR teams, partially in agreement with Aquino et al. [11], who noted that high-ranked teams performed more acceleration compared to the bottom-ranked ones. However, in this investigation, accelerations were comparable between low- and high-ranked teams, emphasizing that the technical and tactical aspects that come into play when trying to avoid relegation play a crucial role in lower-ranked teams, significantly impacting their physical effort.

HR teams showed significantly lower average speed during the match compared to MR and LR teams. This contrasts with the findings of [11], who reported that the top-ranked team covered more distance (and thus had higher average speed) than lower-ranked teams. Our results suggest that average speed may be less critical for match outcomes, and that high-intensity activities are more important to consider [15].

4.2. Differences among Playing Formations within the Same Ranking Level

In the HR group the 3-4-3 playing formation reported lower H-dec than the 3-5-2 and 4-3-3 formation. This result is partially supported by Tierney et al. [35] which identified this decreasing order in terms of differences between playing systems: 3-5-2 > 3-4-3 > 4-3-3 > 4-4-2.

Borghi et al. [36], and Tierney et al. [35] reported that the 3-5-2 formation exerted the greatest amount of accelerations. However, our findings showed that the 4-3-3 formation had the highest H-acc values, with greater acceleration compared to 3-4-3, 3-5-2, and 4-4-2 formations. These results are consistent with the findings of Morgans et al. [7] who reported that 4-3-3 formation resulted in more acceleration than the 3-5-2 formation when comparing teams primarily focused on defending collectively in a deep position (with very low ball possession/low-block). Nevertheless, our findings were not consistent across all ranking groups, highlighting that the playing formation may influence locomotor activities differently among teams of varying ranks. These differences could be attributed to the way a “flat” midfield defends, with an extra man in the center, given that this role requires expending a lot of energy both in possession and out of possession.

4.3. Benchmark of Locomotor Activity

The second purpose of this study was to compile normative data and create benchmarks for AMP, AS, and different high-intensity variables for each role attained by professional soccer players. This approach enables the analysis of players’ kinematic variables, allowing us to understand if their performance is above or below average, as supported by Laterza & Manzi [26]. Moreover, the data collected could be used to assess players’ fatigue and detect symptoms of overreaching or overtraining. If a player consistently exhibits poor performance over a prolonged period, this could be an early sign of overtraining [37]. In addition, these benchmarks may represent a useful tool to assess the performance of junior professionals competing for the first year at a professional level. They can help determine if their level is comparable with more experienced professionals, provide insights into their training needs, and facilitate the monitoring of their performance parameters over time [26].

Analyzing various playing formations and role positions is crucial in soccer, as each distinct role demands a unique activity profile [17,18]. For instance, the average distance covered above 25km/h by attackers differs among playing formations and roles. A distance that might be considered average in one formation could be subpar in another. To illustrate, an attacker in a 3-5-2 formation might cover 260 meters at high speed, which could be significantly less than what’s expected for the same role in a 4-3-3 formation (see Table 6, Table 7, Table 8 and Table 9). This analysis provides invaluable insights for coaches, allowing them to tailor training programs to the specific demands of different roles and formations. Furthermore, benchmarks offer additional benefits. By examining the range of performance levels for each role, we can identify positions where performance is more consistent (i.e., a smaller range). This suggests a more clearly defined activity profile for that role. For example, in a 3-4-3 formation, the box-to-box midfielder’s performance might be more consistent than that of a wide midfielder. This research has successfully provided readily available data for professional soccer coaches, enabling them to quickly assess their athletes’ performance levels. Additionally, the data can help identify players with greater work capacity, potentially allowing coaches to assign them specialized tactical roles that leverage their superior abilities without compromising their performance.

4.4. Limitations

While this study provides valuable insights, it is important to recognize its inherent limitations, which may influence the interpretation and generalizability of the findings.

This study is based on a sample of matches that is not uniform in terms of home and away games. The game location factor must be considered in the analysis of the players’ physical data, as it represents a critical piece of information. It is directly correlated with the style of play and consequently influences the intensity of the performance, as demonstrated by Hands et al. [38] and Beato et al. [39]. Moreover, other contextual variables, such as ball possession, match results, and playing strategies (e.g., high-press, counterattacks, deep-defending) both from an individual and collective tactical perspective, were not considered, which could also impact the outcomes (Bradley and Ade [40], Ju et al. [41], Plakias et al. [42]).

Researchers and practitioners should also be mindful of some aspects of this study before using the presented normative data. The data collected are referred to the Premier Division Championship (Serie A) players, meaning that professional soccer players competing in other championships (e.g., the Spanish LaLiga and the English FA Premier League) might have different activity profiles, as supported by Dellal et al. [43].

Lastly, to the best of our knowledge, this methodological approach, developed by Sonderegger et al. [24], utilizes a spatial reference (distance covered in meters), whereas in this study, the quantitative variable was temporal (the sum of short time intervals as a percentage of the total time spent accelerating during the match). In practice, however, this approach does not consider the total number of accelerations (n^o. of events), which can make comparisons with other studies difficult. Following the previous concept, still in terms of time spent, a fixed threshold of 2m·s^-2 was used for decelerations. Therefore, readers should be mindful when interpreting our speed variations data (H-acc and H-dec), as comparison with other studies may require careful consideration.

5. Conclusions

This study provides insights into the influence of team ranking and playing formation on the locomotor activities of professional soccer players in the Italian First Division. The results revealed that HR teams exhibit a higher percentage of high-intensity accelerations compared to MR teams, emphasizing the importance of high-intensity efforts over average speed in determining match outcomes. However, these differences varied across rankings, highlighting the variability in physical demands based on team strategy and opposition.

The study also demonstrated that playing formations significantly impact locomotor activities, with the 4-3-3 formation showing greater acceleration demands than others, such as 3-5-2 and 4-4-2. These differences were most pronounced in HR teams, underscoring the strategic role of formation in optimizing player performance. However, the lack of consistent trends across all ranking groups suggests that the effectiveness of a formation may vary depending on the team’s ranking.

Furthermore, the benchmarks and normative data provided for various roles and formations offer valuable tools for coaches to assess player performance and detect signs of overtraining. By understanding the role-specific demands within different formations, coaches can better tailor training programs to enhance player readiness and performance.

In the authors’ opinion, due to the fact that accelerations represent one of the most predictive variables associated with the outcome of the match [44], it was essential to improve the reliability of the accelerations data using the method proposed by Sonderegger et al. [24].

Future research should address the current investigation’s limitations and explore the evolving dynamics of locomotor activities in current soccer. Nevertheless, the data generated in this study contribute to a better understanding of the physical demands in soccer and provide a foundation for further investigations.