Abstract
erformance analysis in sports is a rapidly evolving field, where academics and applied performance analysts work together to improve coaches’ decision making through the use of perfor- mance indicators (PIs). This study aimed to provide a comprehensive analysis of factors affecting running performance (RP) in soccer teams, focusing on low (LI), medium (MI), and high-speed distances (HI) and the number of high-speed runs (NHI). Data were collected from 185 matches in the Turkish first division’s 2021–2022 season using InStat Fitness’s optical tracking technology. Four linear mixed-model analyses were conducted on the RP metrics with fixed factors, including loca- tion, team quality, opponent quality, ball possession, high-press, counterattacks, number of central defenders, and number of central forwards. The findings indicate that high-press and opponent team quality affect MI (d = 0.311, d = 0.214) and HI (d = 0.303, d = 0.207); team quality influences MI (d = 0.632); location and counterattacks impact HI (d = 0.228, d = 0.450); high-press and the number of central defenders affects NHI (d = 0.404, d = 0.319); and ball possession affects LI (d = 0.287). The number of central forwards did not influence any RP metrics. This study pro- vides valuable insights into the factors influencing RP in soccer, highlighting the complex interactions between formations
HI (d = 0.228, d = 0.450); high-press and the number of central defenders affects NHI (d = 0.404, d = 0.319); and ball possession affects LI (d = 0.287). The number of central forwards did not influence any RP metrics. This study pro- vides valuable insights into the factors influencing RP in soccer, highlighting the complex interactions between formations and physical, technical–tactical, and contextual variables. Understanding these dynamics can help coaches and analysts optimize team performance and strategic decision making. Keywords:football; performance analysis; tactics; formations; running performance metrics; contextual variables 1. Introduction Performance analysis in sports is a great and rapidly evolving field [1]. On one hand, academics continually provide new research data, while on the other, applied performance analysts (PAs) join the coaching staff of teams, assisting in improving coaches’ decision making [2–4]. PAs use performance indicators (PIs) to draw conclusions [5]. PIs can be categorized based on whether their data originate from event or positional-tracking data [6], while to strengthen the meaningfulness of these, the importance of contextual variables has been highlighted [7]. Event data provide information on technical and tactical issues [8], tracking data are useful for analyzing running performance (RP) and tactics [9], while contextual variables enhance the quality of both [10]. Running performance (RP) is a crucial element of the overall performance of soccer players, contributing along with technical, tactical, and mental components [11]. However, on its own, it is not sufficient to interpret match outcomes [12,13], and it can be influenced by many factors [14]. Therefore, when interpreting RP in soccer, one should include contextual variables that may also explain part of the mental components [15,16], as well as technical–tactical variables. Such a holistic approach can provide genuinely useful information to team coaches. The role of contextual variables in the RP of teams has been documented by research for many years, as Lago-Peñas [17] states in his review article. Regarding the most recent Sports2024,12, 196.
The role of contextual variables in the RP of teams has been documented by research for many years, as Lago-Peñas [17] states in his review article. Regarding the most recent Sports2024,12, 196.
Sports2024,12, 196 2 of 12 research, several contradictory findings have emerged. Modric et al. [18] found that away matches were associated with increased total distance (TD) and low-intensity (<4 m/s) running. Jerkovic et al. [19] found a greater amount of distance covered in the running zone (4–5.48 m/s) for away matches, while Gonçalves et al. [20] observed higher values of high- intensity running (60–100% of the individual maximum running speed) in home matches compared to away matches. Additionally, Modric et al. [18] found that team quality was not associated with RP, whereas Aquino et al. [21] reported that the top-ranked team covered greater total distance with high acceleration compared to the bottom-ranked team. Finally, regarding opponent quality, Modric et al. [18] and Jerkovic et al. [19] concluded that the quality of the opponent team was not associated with RP. Conversely, Gonçalves et al. [22] found that starters playing against strong opponents exhibited higher values of distance in high-speed running (5.5–7 m/s), and Gonçalves et al. [20] found that matches against strong opponents resulted in greater total distance covered. However, analyzing situational variables in isolation seems to offer only a limited understanding of the complicated nature of team sports performance, since research has shown that ball possession, as well as other technical and tactical actions, is also affected by the location of the match and the opponents’ strength [17]. In examining the effects of tactical behavior, several researchers have studied the impact of ball posses- sion on RP. For example, Modric et al. [23] found that in the UEFA Champions League 2020–2021, teams with high ball possession covered more distance in high-speed running (5.5–7 m/s) and sprinting (>7 m/s) than teams with low ball possession. In contrast, da Mota et al. [24] found that in the 2014 FIFA World Cup, teams with high ball possession covered similar distances at medium (3–3.9 m/s) and high speeds (>3.9 m/s) but cov- ered greater distances in total and at low speed (≤3 m/s) compared to teams with low ball possession. Additionally, Modric et al. [23] found no differences in counterattacks and high-pressing regardless of
[24] found that in the 2014 FIFA World Cup, teams with high ball possession covered similar distances at medium (3–3.9 m/s) and high speeds (>3.9 m/s) but cov- ered greater distances in total and at low speed (≤3 m/s) compared to teams with low ball possession. Additionally, Modric et al. [23] found no differences in counterattacks and high-pressing regardless of whether teams covered greater or smaller distances both overall and at high intensities. Conversely, Forcher et al. [25] identified differences in RP between teams that adopt a counterattacking style compared to those that favor a ball possession style, and Low et al. [26] found differences in RP between teams employing deep-defending and high-press strategies. These studies highlight the complex and varied effects of tactical behaviors on RP in soccer, indicating that different styles of play can significantly influence physical demands on players, something that also emerged in the research of Plakias et al. [27], who made a direct comparison between two opposite styles in 19 tactical situations. Finally, regarding formations, which constitute another tactical aspect of soccer [28], Modric et al. [29] demonstrated that the values for almost all of the RP metrics are greater with three central defenders than with two central defenders. This conclusion was also reached by Tierney et al. [30], who found greater distances covered at high speeds (≥5.5 m/s) with three central defenders. Additionally, Borghi et al. [31] found that forma- tions with two central forwards cover greater distances at high intensities (>5.5 m/s) than those with one central forward (3-5-2 > 4-4-2 > 4-3-3). This finding aligns with the results of Bradley et al. [32] (4-4-2 > 4-3-3 > 4-5-1) and Arjol-Serrano et al. [33] (4-4-2 > 4-2-3-1), but not with those of Aquino et al. [34] and Vieira et al. [35], who found that mean speed and high-intensity activities are greater in the 4-3-3 formation compared to the 4-4-2. It is worth noting that all the aforementioned studies on formations had sample sizes ranging from 20 to 59 matches, limiting the generalizability of the conclusions regarding formations. For this reason, the authors focused
et al. [34] and Vieira et al. [35], who found that mean speed and high-intensity activities are greater in the 4-3-3 formation compared to the 4-4-2. It is worth noting that all the aforementioned studies on formations had sample sizes ranging from 20 to 59 matches, limiting the generalizability of the conclusions regarding formations. For this reason, the authors focused more on the differences between player positions. All the aforementioned conflicting findings indicate that RP may be influenced differ- ently in each competition, depending on the level and type of competition. Additionally, examining factors in isolation may yield some results, but it cannot explain the complex nature of soccer, where physical, technical–tactical, and contextual factors interact with each other. Building on the existing literature, we hypothesized that a combination of eight factors, including location, team quality, opponent team quality, ball possession, high-press, counterattacks, number of central defenders, and number of central forwards, significantly
Sports2024,12, 196 3 of 12 affect various aspects of running performance. Specifically, we anticipate that these factors will differentially impact low-speed distance (LI), medium-speed distance (MI), high-speed distance (HI), and the number of high-speed runs (NHI). Using a large sample of matches that included all teams from the first Turkish league in the 2021–2022 season, we aimed to address an additional gap in the existing literature. Specifically, most studies examining factors affecting team RP often rely on samples from a single team with GPS data access or a limited number of matches from international competitions like the UEFA Champions League or the FIFA World Cup. Therefore, the purpose of this study is to provide a comprehensive analysis of the factors (including formations, technical–tactical variables, and contextual factors) that affect the RP of soccer teams, focusing on LI, MI, HI, and NHI. This holistic approach aims to enhance the understanding of how different elements interact to affect RP in soccer. By doing so, it seeks to fill the existing gaps in the literature and offer practical insights for optimizing team performance at a competitive level. Understanding these dynamics can help coaches and analysts design better training and game strategies, ultimately improving team performance and success in matches. 2. Material and Methods 2.1. Sample This study analyzed data from the Turkish first division’s 2021–2022 season, encom- passing 20 teams over 38 matchdays, with each matchday featuring 10 games. Instatscout supplied data for the first 24 matchdays, covering 240 matches. Due to missing data for two matches and the exclusion of 53 matches because of red card incidents, the final sample included 185 matches, resulting in 370 (i.e., 2×185) observations, with each team in a match providing one observation. 2.2. Procedure Data collection of the running variables was performed using InStat Fitness’s optical tracking technology (https://football.instatscout.com/, accessed on 1 May 2024), a FIFA- certified system known for its high precision and reliability, as confirmed on the FIFA website [36] and referenced in previous studies [37,38]. For the 2021–2022 season, InStat’s system was designated as the official Electronic Performance and Tracking System (EPTS) for
of the running variables was performed using InStat Fitness’s optical tracking technology (https://football.instatscout.com/, accessed on 1 May 2024), a FIFA- certified system known for its high precision and reliability, as confirmed on the FIFA website [36] and referenced in previous studies [37,38]. For the 2021–2022 season, InStat’s system was designated as the official Electronic Performance and Tracking System (EPTS) for the league [27]. For the technical–tactical variables the data were obtained from the Instatscout platform. As reported in previous studies, the reliability of Instatscout data is very high (K values 0.90 to 0.98) [39–41]. 2.3. Ethics Ethics committee approval of the current study was gained from the University of Thessaly (No. 1973, 12 October 2022). Additionally, InStat Ltd. granted written consent on 8 November 2022, for the utilization of the data in this research, ensuring adherence to all ethical standards for research and publication. 2.4. Statistical Analyses Initially, cluster analysis was applied to create categorical variables concerning: (a) the quality of the teams and the opponents as strong or weak (based on the num- ber of points each team collected in the final championship standings), (b) ball possession as high or low (based on the team’s possession percentage in each observation), (c) the counterattacks of the match as many or few (based on the total number of counterattacks for both teams in the match for each observation), and (d) the high-pressing actions as many or few (based on the number of high-pressing actions of the team in each observation). Next, after checking the normality of the distribution of the four dependent variables shown in Table, a linear mixed-model analysis was performed four times, correspond- ingly. In all cases, the eight independent variables shown in Table factors, and the variable TEAM was used as a random factor. The definitions in Table high press, counterattacks, and ball possession are derived from the glossary of Instatscout,
Sports2024,12, 196 4 of 12 the platform from which the data for the respective variables were obtained. The vari- able TEAM represented the 20 different teams participating in the Turkish league in the 2021–2022 season. The thresholds for creating the running variables are those used in previous research [27,29,42]. However, in this study, the intensities of standing (<0.2 m/s), walking (0.21–2 m/s), and jogging (2.01–4 m/s) were combined as low-intensity, the inten- sity of running (4.01–5.5 m/s) was named medium-intensity, and the intensities of high speed (5.51–7 m/s) and sprint (>7 m/s) were combined as high-intensity. All statistical analyses were conducted using the IBM SPSS statistical package (version 29.00, IBM Corpo- ration, Armonk, NY, USA), with a significance level set atp< 0.05. Cohen’s d was utilized to measure the effect size. The effect sizes were defined as follows: trivial (d = 0.0 to 0.19), small (d = 0.2 to 0.49), medium (d = 0.5 to 0.79), and large (d≥0.8) [43]. Table 1.The variables used when performing the four linear mixed-model analyses. Type Abbreviation Definition Dependent variables LI Low-intensity (0–4 m/s) distance MI Medium-intensity (4.01–5.5 m/s) distance HI High-intensity (>5.51 m/s) distance NHI Number of high-intensity runs (>5.51 m/s) Independent variables (fixed factors) LC Location (home/away) TQ Team’s quality (strong/weak) OQ Opponent’s quality (strong/weak) CD Number of central defenders (two/three) CF Number of central forwards (one/two) HP High-press (many/few): Pressing until 30 m from the opponent’s goal. Pressing, in its turn, is counted for the opponents of a team that is building its attack when players are actively trying to get the ball back. BP Ball possession (high/low): Percentage share of one team’s ball possession in the total ball-in-play time. Ball possession is the sum of all time periods between the start of possession to the moment of transition or to the moment the ball went out. CA Counterattacks (many/few): Attack from the open play that starts with winning the ball from a defensive position and then quickly transitioning to offense while the prior attacking team is caught in an offensive formation; the length of possession during the attack cannot exceed
start of possession to the moment of transition or to the moment the ball went out. CA Counterattacks (many/few): Attack from the open play that starts with winning the ball from a defensive position and then quickly transitioning to offense while the prior attacking team is caught in an offensive formation; the length of possession during the attack cannot exceed 8 s before the possession transitions or ends; alternatively, the length of possession can last between 8 and 30 s, but the speed of attack cannot be less than 2.6 m/s. A counterattack cannot begin with a pass from a goalkeeper if he controlled the ball for more than 4 s before the action. Independent variable (random factor) TEAM The 20 teams that participated in the Turkish first division 2021–2022 season.
Sports2024,12, 196 5 of 12 3. Results 3.1. Cluster Analyses Based on the number of points they accumulated in the championship, the teams were classified into two categories (strong/weak). The strong category included teams that ranked 1st–13th with points ranging from 52 to 81 (cluster center 61.00), while the weak category included teams that ranked 14th–20th with points ranging from 20 to 47 (cluster center 36.57). Based on the number of high-pressing actions performed by the teams in each match, the observations were classified into two categories (many/few). The many category included observations where the team had 10 to 21 high-pressing actions (cluster center 12.50), while the few category included observations where the team had 0 to 9 high- pressing actions (cluster center 5.96). Based on ball possession percentage, the observations were classified into two cate- gories (high/low). The high category included observations where the team had 50.01% to 80% possession (cluster center 56.54), while the low category included observations where the team had 20% to 49.99% possession (cluster center 42.67). Finally, based on the number of counterattacks performed by both teams combined in each match, the observations were classified into two categories (many/few). The many category included observations where the match had 30 to 53 counterattacks (clus- ter center 34.76), while the few category included observations where the match had 13 to 29 counterattacks (cluster center 23.74). 3.2. Linear Mixed Models Table independent variables. Table the four models that were generated for the corresponding four dependent variables. Table 2.The mean of the dependent variables across the categories of the independent variables. Independent Variables Independent Variables’ Categories Dependent Variables LI MI HI NHI CD Two CDs 85,192.247 18,740.194 9780.509 632.877 Three CDs 85,224.288 19,181.388 10,078.614 652.853 CF One CF 85,132.495 18,919.433 9969.323 645.492 Two CFs 85,284.040 19,002.149 9889.800 640.238 TQ Strong 85,489.889 19,456.661 10,089.752 650.853 Weak 84,926.646 18,464.921 9769.370 634.877 OQ Strong 85,239.146 19,129.081 10,043.755 648.218 Weak 85,177.389 18,792.501 9815.367 637.512 HP Many 85,117.847 19,205.067 10,096.827 655.501 Few 85,298.688 18,716.515 9762.295 630.228 BP High 84,801.273 18,967.825 9991.655 644.705 Low 85,615.262 18,953.757 9867.467 641.024 CA Few 85,304.201 18,946.646 9680.661 640.152 Many
18,919.433 9969.323 645.492 Two CFs 85,284.040 19,002.149 9889.800 640.238 TQ Strong 85,489.889 19,456.661 10,089.752 650.853 Weak 84,926.646 18,464.921 9769.370 634.877 OQ Strong 85,239.146 19,129.081 10,043.755 648.218 Weak 85,177.389 18,792.501 9815.367 637.512 HP Many 85,117.847 19,205.067 10,096.827 655.501 Few 85,298.688 18,716.515 9762.295 630.228 BP High 84,801.273 18,967.825 9991.655 644.705 Low 85,615.262 18,953.757 9867.467 641.024 CA Few 85,304.201 18,946.646 9680.661 640.152 Many 85,112.334 18,974.936 10,178.46 645.578 LC Home 85,320.289 19,026.764 10,055.527 647.732 Away 85,096.246 18,894.818 9803.596 637.998
Description
Analyzes factors influencing running performance in Turkish soccer teams.