First to Score, First to Win? Comparing Match Outcomes and Developing a Predictive Model of Success Using Performance Metrics at the FIFA Club World Cup 2025

Abstract

In the present study, 96 teams’ performances across 48 matches in the group stage of the FIFA Club World Cup 2025 were analyzed. Teams scoring first won 62.5% of matches (p < 0.05), while goals were evenly distributed between halves (p > 0.05) and showed marginal variation across six 15 min intervals, peaking near the 30–45 and 75–90 min marks. Parametric analyses revealed a significant effect of match outcome on possession, with winning teams exhibiting higher average possession (53.3%) compared to losing and drawing teams. Non-parametric analyses identified significant differences between match outcomes for goals scored, attempts at goal, total and completed passes, pass completion rate, defensive line breaks, receptions in the final third, ball progressions, defensive pressures, and total distance covered. Winning teams scored more goals and registered more attempts on target than losing teams, although some metrics showed no significant difference between wins and draws. Logistic regression analysis identified attempts at goal on target, defensive pressures, total completed passes, total distance covered, and receptions in the final third as significant predictors of match success (AUC = 0.85), correctly classifying 80.2% of match outcomes. These results emphasized the crucial role of offensive accuracy and possession dominance in achieving success in elite football.

1. Introduction

Understanding the key performance indicators (KPIs) that influence match outcomes in professional soccer has been the focus of extensive research, as these insights provide valuable guidance for coaching strategies and performance optimization. For instance, Harrop and Nevill [1] investigated performance differences in matches won, drawn, or lost by a professional League One soccer team, revealing significant variations in offensive metrics, while defensive variables remained largely stable across outcomes. Notably, losing matches involved a higher number of passes, but passing accuracy was lower in draws compared to wins and losses. Additionally, more passes occurred in the opposition half during losses, thereby emphasizing that while losing teams attempt more passes, the quality and effectiveness of passing, especially in advanced areas, are critical for achieving positive results [1]. Similarly, Kite and Nevill [2] conducted a three-season analysis of a professional English League One soccer team, examining how technical and tactical performance indicators related to team success, finding that winning matches were associated with fewer passes, thus underscoring that passing quality and effectiveness are more important than volume. Additionally, shots on target emerged as a key predictor of success, highlighting the importance of offensive precision [2].

Research has consistently shown that scoring the first goal in football provides a considerable competitive edge across various tournaments and levels of play. This advantage is evident in both domestic leagues and international competitions. Teams that score first tend to have winning rates exceeding 60% [3,4,5,6,7,8,9]. For example, an analysis of 1116 matches of the Chinese Football Super League (2014–2018) showed that teams scoring the first goal won 66.31% of matches and remained unbeaten in 87.01% (including draws) [3]. Similarly, a study examining 504 matches across three top European women’s leagues, Primera Iberdrola (Spain), D1 Féminine (France), and Frauen-Bundesliga (Germany), found high win rates for teams scoring first: 77.3% in Spain, 94.0% in France, and 95.0% in Germany [4]. These success rates notably exceeded those commonly observed in men’s football [4].

Moreover, an analysis of the 2014–2015 season across five major European leagues, the English Premier League, French Ligue 1, Spanish La Liga, Italian Serie A, and German Bundesliga, highlighted a significant advantage for teams scoring the first goal. This advantage was more pronounced for home teams than for away teams. Results indicated that home teams scored first in 57.8% of matches and secured 84.85% of the points available in those games, whereas away teams that scored first obtained only 76.25% of subsequent points [10]. These findings were further corroborated by investigations in the Greek Soccer League, which demonstrate significant differences in offensive and defensive KPIs among teams with different final rankings. Specifically, winning teams exhibit superior metrics in goals scored, shots on target, ball possession, passing accuracy, and defensive recoveries [9,11]. These national-level results reinforced the broader trends observed internationally, confirming the consistent importance of offensive and defensive KPIs in determining success. Regression analyses further emphasized that shots on target, counterattacks, and set-piece execution were key predictors of match outcomes [11]. Additionally, studies on tactical elements such as crosses reveal that successful attempts predominantly originate from specific field zones. This highlighted the spatial dimension of offensive strategies [12].

Furthermore, Stafylidis et al. [8] examined the UEFA Euro 2024 tournament, revealing that scoring the first goal significantly increases the likelihood of a positive match outcome. No significant temporal differences were found in goal distribution between halves or 15 min intervals. However, offensive metrics such as attempts on target and passes into the attacking third were strong positive predictors of winning [8]. Conversely, attempts on target outside the penalty area and crosses attempted negatively predicted success, highlighting the importance of precise offensive execution [8]. The study of Martínez et al. [7] concluded that scoring the first goal holds substantial importance in both the UEFA Euro and FIFA World Cup tournaments. Historically, the 1930s, 1950s, and 1970s were characterized as periods of higher goal-scoring rates across examined variables [7]. Overall, the effect of scoring first on match outcomes in qualifiers and final stages of the FIFA World Cup and UEFA Euro was approximately 78.46%, with a success rate of 77.77% for the FIFA World Cup and 79.16% for the UEFA Euro. Additionally, research on the 2018 FIFA World Cup indicated that 63% of goals were scored in the second half. The majority of these occurred within the first 15 min of the second half and in the closing stages of matches [13]. A related study on the Portuguese Premier League (2009–2015) reported that 58% of second goals occurred in the second half. There was also a significant correlation between the timing of the first and second goals. Cox regression analysis revealed that if the first goal occurred in the second half, the probability of a second goal being scored was tripled [14]. Temporal match analyses also showed that high-ranking teams tend to score more goals in specific phases, particularly in the first half and during the 15–30 min interval, suggesting that strategic intensity early in the match may be crucial for success [15]. However, certain competitions have shown no statistically significant differences in win rates based on the timing of goals scored within matches [4].

Recent analyses have consistently highlighted the critical role of offensive efficiency, ball possession, and defensive solidity in determining success on the field [8,11,15]. For instance, an analysis of 1125 UEFA Champions League matches from 2009/2010 to 2017/2018 revealed evolving trends in technical performance, with passing and attacking metrics increasing in frequency and accuracy, while defensive variables declined or remained stable, indicating a strategic shift toward controlling play through enhanced passing rather than wide crossing [16]. Similarly, research analyzing 1291 players from the “Big Five” European leagues in the Champions League found significant technical differences, with Bundesliga players taking more shots, and Serie A players exhibiting fewer ball touches, passes, and lower pass accuracy, but compensating with a higher frequency of long balls compared to players from La Liga, Premier League, and Ligue 1 [17]. Furthermore, a study utilizing data from Premier League matches found that teams’ average pressing effectiveness was strongly correlated with shots created per pressure (positively) and shots allowed (negatively), as well as league points [18]. Another study identified key defensive metrics, particularly clearances and goalkeeper saves, as significant differentiators of match outcomes. Increased last-resort defensive actions were strongly associated with losing teams, while higher tackling success positively correlated with winning, thereby underscoring the critical role of defensive performance in determining football match success [19]. During the FIFA World Cup Qatar 2022, 33 out of 137 significant technical-tactical indicators differentiating qualified from non-qualified teams were identified, with offensive efficiency in set pieces, height of defensive line during offensive phase, and ability to reduce available playing space emerging as main indicators for team classification [20].

Regarding passing, progressive passing in football is crucial for creating positive possession outcomes. Analysis of international football found that “Mid Central to Mid Half Space” and “Mid Half Space to Final Central” progressive passes provide the best balance between risk and reward [21]. Additionally, research comparing top-tier and bottom-tier Premier League teams across 266 matches identified significant differences in technical performance. Top-tier teams recorded more shots, shots on target, successful take-ons, and higher pass completion rates. In contrast, bottom-tier teams produced more clearances, blocked shots, and tackles in the defensive third. Notably, for top-tier teams, each additional aerial duel won increased the probability of winning by 35% [22]. Moreover, a historical analysis of Serie A from the 2000/01 to 2009/10 using Data Envelopment Analysis, revealed a tactical paradigm shift. Since the 2005/06 season, improvements in offensive efficiency have been shown to yield relatively greater benefits than traditional defense-focused strategies [23]. Similarly, a longitudinal study of Spanish La Liga over eight seasons (2011–12 to 2018–19) identified a significant tactical evolution. The number of passes and team width remained stable, while there was a downward trend in shots, crosses, corners, total distance covered, team length, and goalkeeper–defense distance. These findings suggest a growing prioritization of defensive strategies over offensive emphasis in La Liga [24].

Another comprehensive study analyzing First Division Championship matches across 11 European countries during the 2021–2022 season developed a framework integrating tactical situations as key performance indicators (KPIs). The findings revealed that possession-based play, counterattacks during offensive transitions, and a balanced yet aggressive defensive strategy significantly enhance winning probabilities. Successful teams emphasized central attacks, minimized crosses, and executed strategic plays that generated goal attempts even with limited possession overall [25]. Furthermore, a recent study investigated shot effectiveness and goalkeeper interventions by analyzing 15,266 on-target shots from five major European leagues during the 2022/2023 season. The study identified significant associations between shot success and variables such as match context, target zones within the goal, field locations, and shooting foot. These factors were shown to substantially influence scoring outcomes [26]. An analysis of playing styles across European competitions identified marked differences between teams using wide versus narrow possession styles. Defensive ball pressure emerged as a key differentiator between these approaches [27]. Moreover, research employing machine learning techniques on the FIFA dataset achieved a position prediction accuracy of 99.84%. Using Recursive Feature Elimination, the study identified the top five features influencing player positional suitability [28]. Advanced neural network–based pattern analysis, integrated with success-oriented statistical frequency methods, enhances tactical performance evaluation. This approach offered promising avenues for understanding game dynamics beyond conventional static indicators [29]. An analysis of the 2014 FIFA World Cup in Brazil showed that winning teams scored significantly more goals overall and from set pieces. They also had higher shot accuracy and more shots on goal while receiving fewer yellow cards compared to losing teams. Binary logistic regression identified shot accuracy as the strongest predictor of match success [30]. Conversely, physical performance parameters showed no significant differences between winning and losing teams. This highlighted the dominant role of technical efficiency, particularly in scoring, in determining success at the World Cup level [30]. In addition, an analysis of the 2023–2024 UEFA Youth League revealed that attacking effectiveness was higher when teams were leading, faced less initial pressure, and initiated attacks from defensive or pre-defensive zones. Early penetrative actions and opponent pressure shortened attack duration. Counter-attacks were more frequently observed when teams were ahead [31].

This study introduced a novel contribution to the existing body of performance analysis literature by offering the first in-depth examination of match outcomes and key performance indicators within the specific context of the FIFA Club World Cup 2025. Unlike previous investigations, which primarily focused on national leagues, continental tournaments, or senior international competitions, the present research uniquely examined an intercontinental club tournament featuring elite teams from all football confederations. Crucially, this was the first analytical study of a FIFA tournament conducted under the newly adopted format that included an extended structure with both group stage matches and knockout rounds, marking a significant structural departure from earlier editions of the competition. The study initially compared winning, drawing, and losing teams by integrating temporal goal-scoring patterns with technical, tactical, and physical performance variables, thereby providing a multifaceted overview of performance profiles across different match outcomes. Building on this comparative approach, a second level of analysis involved the development of a predictive model of match success using selected indicators, including attempts at goal on target, total completed passes, receptions in the final third, defensive pressures, and total distance covered. Notably, Receiver Operating Characteristic (ROC) curve analysis, seldom applied in previous football performance research, was employed to evaluate the discriminative ability of these indicators. This methodological advancement extended prior technical and tactical analyses by introducing a more advanced and interpretable means of identifying the most influential performance variables associated with match outcomes. As such, the study provided original and context-specific insights into the determinants of success in elite-level international club football. To summarize, the aim of the present study was to investigate the impact of scoring the first goal and to conduct a temporal analysis of goal scoring during matches, alongside examining key performance indicators associated with different match outcomes during the group stage of the FIFA Club World Cup 2025. Specifically, the study aimed to analyze technical, tactical, and physical metrics to determine their relationship with winning, drawing, or losing. Additionally, it sought to develop a predictive model to identify the most significant factors influencing match success, thereby offering practical implications for performance optimization in elite football.

2. Materials and Methods

2.1. Sample

This investigation focused on the performance data of 96 teams participating in 48 group stage matches of the FIFA Club World Cup 2025. Matches from the knockout rounds were deliberately excluded to avoid potential inconsistencies arising from the occurrence of extra time in certain games, which could compromise the comparability of the dataset. The dataset comprised performance metrics from these 48 matches involving 96 teams, with a total of 96 recorded match outcomes owing to additional contextual considerations. Match outcomes were initially categorized into three groups: lose, draw, and win. The frequencies of these outcomes were 35 losses (36.5%), 26 draws (27.1%), and 35 wins (36.5%). For the binary logistic regression analysis, match outcomes were dichotomized into non-win (loss or draw, coded 0) versus win (coded 1). Within this binary classification, 61 instances (63.5%) represented non-win outcomes, and 35 instances (36.5%) represented wins.

2.2. Data Collection and Analysis Procedures

This study’s dataset was compiled from the official statistical records maintained by the FIFA Training Centre (https://www.fifatrainingcentre.com, accessed on 15 June to 1 July 2025). The data collection, organization, and subsequent analysis were conducted by one of the authors, who holds a UEFA coaching license and specializes in football performance analysis, during the period from 15 June to 1 July 2025. Performance indicators related to offensive and defensive play that were available for all 48 group-stage matches and consistently reported across all participating teams were included in the analysis. Metrics related to goalkeeper-specific actions were excluded, as they were not uniformly recorded and were beyond the primary scope of team-level performance evaluation. Furthermore, matches from the knockout phase were deliberately excluded, as the occurrence of extra time in certain fixtures could introduce structural inconsistencies and compromise the reliability of comparisons across match outcomes. Prior to analysis, all variables were inspected for completeness, consistency, and the presence of outliers. The final dataset was complete, with no missing values detected. Previous investigations utilizing official FIFA statistics include studies on the 2010 and 2018 World Cups [32,33], as well as recent research on similar tournaments [34]. Corresponding studies have also employed alternative platforms such as Whoscored and Kicker to assess team performance in major competitions, as demonstrated by analyses of the 2018 World Cup [35] and UEFA Euro 2012 [36].

The selected variables encompass both technical and tactical metrics. Key terms such as build up unopposed, defined as team progression from the defensive third without direct opposition, and line breaks, referring to successful passes or carries that bypass an opposition defensive line, are also used as defined within the FIFA Training Centre taxonomy.

2.3. Statistical Analysis

Data were analyzed and visualized with IBM SPSS Statistics (version 29 for Windows) [37], Jamovi (version 2.6.23.0 for Windows) [38], and JASP (version 0.19.3.0 for Windows) [39]. Descriptive statistics were computed for all variables, including means, medians, standard deviations, interquartile ranges, and 95% confidence intervals (CI). To examine the impact of scoring the first goal on match outcomes and the temporal distribution of goals, Chi-square goodness-of-fit tests were conducted to compare observed frequencies against expected equal proportions. Normality of distribution was assessed using the Shapiro–Wilk test. Variables meeting normality assumptions were analyzed with parametric tests, while non-normally distributed variables were analyzed using non-parametric methods. Effect sizes (ES) were calculated in accordance with Cohen’s [40] guidelines, utilizing η² for parametric data and ε² for non-parametric data. The thresholds for interpreting effect sizes were categorized as small (0.01–0.06), moderate (0.06–0.14), and large (>0.14). A one-way Analysis of Variance (ANOVA) was conducted to examine differences in possession percentage and other normally distributed performance metrics across match outcomes categorized as loss, draw, and win. Where significant effects were identified, post hoc Tukey tests were applied for pairwise comparisons. For variables violating normality assumptions, Kruskal–Wallis tests were employed to assess differences across match outcomes. Significant Kruskal–Wallis results were further explored with Dwass–Steel–Critchlow–Fligner pairwise comparisons to identify specific group differences.

To determine the variables that significantly influence match results, a generalized linear modeling approach was applied, specifically logistic regression, with the binary outcome variable coded as win versus draw or loss, consistent with methodologies employed in earlier research [1,2,8,11]. Binomial logistic regression analyses were performed to model the likelihood of match success (win vs. non-win) based on key match performance variables. Nine successive models (M₀ to M₉) were constructed, progressively incorporating predictors to evaluate their individual and combined effects on winning probability. Model fit was assessed by changes in deviance, Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and pseudo R² indices (McFadden, Nagelkerke, Tjur, Cox and Snell). Multicollinearity was evaluated using tolerance and variance inflation factors (VIF), with VIF values below 5 considered acceptable. The final logistic regression model’s discriminative ability was examined through Receiver Operating Characteristic (ROC) curve analysis, with the area under the curve (AUC) indicating predictive accuracy. Classification metrics including accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were computed at the optimal cutoff point determined by Youden’s Index. Additionally, supplementary ROC curve analyses for individual predictors were performed using the DiagROC module in Jamovi [38,41,42,43,44,45,46], to further assess their diagnostic accuracy and establish practical threshold values for match outcome discrimination. Statistical significance was set at p ≤ 0.05 for all tests.

3. Results

3.1. Impact of the First Goal on Match Outcome and Temporal Analysis of Goal Scoring

The distribution of first goal effects significantly deviated from the expected equal distribution (χ²(2) = 20.4, p < 0.001), with teams scoring first winning 62.5% of the matches, drawing 27.1%, and losing only 10.4%, highlighting the strong association between scoring the first goal and increased likelihood of winning (Figure 1). In practical terms, teams that scored the first goal were over six times more likely to win than to lose, demonstrating its decisive impact on match outcome.

Additionally, the timing of goals within the match was analyzed (Figure 2). A goodness-of-fit test comparing the proportion of goals scored in the first (49.3%) versus second (50.7%) half revealed no statistically significant difference (χ²(1) = 0.028, p = 0.868), indicating an approximately equal distribution of goals across halves.

Further, goal scoring across six 15 min intervals was examined to determine temporal trends. No significant difference was found in the distribution of the observed goal frequencies from expected equal proportions (χ²(5) = 11.0, p = 0.051). Goals were most frequent in the 30–45 min (22.9%) and 75–90 min (22.9%) intervals, suggesting potential periods of increased offensive effectiveness near the end of each half.

3.2. Performance Indicators Comparison Between Match Outcomes

A one-way ANOVA analysis was conducted to examine differences in possession percentage across match outcomes (loss, draw, win). The analysis revealed a significant effect of match result on possession (Figure 3), F(2, 93) = 12.20, p < 0.001, η² = 0.208. Specifically, possession was significantly higher in winning matches (M = 53.3% ± 12.8, 95% CI [48.9, 57.7]) compared to losses (M = 39.1% ± 12.4, 95% CI [34.9, 43.4]; p < 0.001), indicating that winning teams controlled possession approximately 14 percentage points more than losing teams. Possession in wins was also significantly higher than in draws (M = 45.6% ± 10.0, 95% CI [41.5, 49.6]; p = 0.039). No significant difference was observed between draws and losses (p = 0.099).

Due to non-normal distributions of other performance variables, Kruskal–Wallis tests were conducted for comparisons across match outcomes (Figure 1, Figure 2 and Figure 3). Significant differences were found for goals (χ²(2) = 47.20, p < 0.001, ε² = 0.497), attempts at goal (χ²(2) = 10.97, p = 0.004, ε² = 0.116), attempts at goal on target (χ²(2) = 18.78, p < 0.001, ε² = 0.198), total passes (χ²(2) = 20.73, p < 0.001, ε² = 0.218), total completed passes (χ²(2) = 19.83, p < 0.001, ε² = 0.209), defensive line breaks (χ²(2) = 7.92, p = 0.019, ε² = 0.083), receptions in final third (χ²(2) = 15.77, p < 0.001, ε² = 0.166), ball progressions (χ²(2) = 8.93, p = 0.012, ε² = 0.094), defensive pressures (χ²(2) = 21.85, p < 0.001, ε² = 0.230), and total distance covered (χ²(2) = 8.86, p = 0.012, ε² = 0.093).

Post hoc Dwass–Steel–Critchlow–Fligner pairwise comparisons revealed that winning teams scored significantly more goals (Mdn = 2, IQR = 2; M = 2.91 ± 1.9, 95% CI [2.28, 3.55]) than losing teams (Mdn = 0, IQR = 1; M = 0.51 ± 0.7, 95% CI [0.26, 0.77]; W = 9.09, p < 0.001) and drawing teams (Mdn = 1, IQR = 1; M = 0.92 ± 1.2, 95% CI [0.45, 1.39]; W = 6.88, p < 0.001). Attempts at goal were higher in wins (Mdn = 14, IQR = 7; M = 15.29 ± 7.0, 95% CI [12.90, 17.67]) than losses (Mdn = 10, IQR = 7; M = 10.09 ± 5.5, 95% CI [8.18, 11.99]; W = 4.69, p = 0.003), with no significant difference between wins and draws (drawing teams: Mdn = 13, IQR = 10.5; M = 13.6 ± 7.8, 95% CI [10.45, 16.71]; p = 0.568), showing that teams that won produced around five more goal attempts per match than those that lost. Attempts at goal on target were also greater in wins (Mdn = 6, IQR = 3; M = 6.17 ± 3.4, 95% CI [4.99, 7.35]) than losses (Mdn = 3, IQR = 2; M = 3.17 ± 2.3, 95% CI [2.38, 3.96]; W = 6.16, p < 0.001), with no significant difference compared to the drawing teams (drawing teams: Mdn = 3, IQR = 4; M = 4.35 ± 2.8, 95% CI [3.2, 5.5]). On average, winning teams registered three more shots on target than losing teams, underlining the importance of shooting precision.

Total passes were significantly higher in wins (Mdn = 600, IQR = 261; M = 579.97 ± 166.6, 95% CI [522.73, 637.22]) than losses (Mdn = 380, IQR = 162; M = 408.23 ± 130.8, 95% CI [363.30, 453.16]; W = 5.96, p < 0.001), and wins exceeded draws (drawing teams: Mdn = 444, IQR = 167.75; M = 448.00 ± 113.1, 95% CI [402.3, 493.7]); p = 0.004). These results indicate that winning teams completed on average over 170 more passes than losing ones. Similarly, total completed passes were greater in wins (Mdn = 528, IQR = 266; M = 519.69 ± 170.6, 95% CI [461.10, 578.28]) compared to losses (Mdn = 323, IQR = 174; M = 344.80 ± 132.1, 95% CI [299.41, 390.19]; W = 5.84, p < 0.001), with wins also exceeding draws (Mdn = 386.50, IQR = 157.25; M = 385.00 ± 114.5, 95% CI [338.74, 431.26]; p = 0.006).

Defensive line breaks differed between wins (Mdn = 12, IQR = 7; M = 13.17 ± 5.7, 95% CI [11.20, 15.14]) and losses (Mdn = 8, IQR = 6; M = 9.71 ± 4.9, 95% CI [8.03, 11.40]; W = 3.86, p = 0.017), but not between draws and wins (Mdn = 9, IQR = 6.8; M = 10.4 ± 5.3, 95% CI [8.24, 12.53]; p = 0.141). Completed line breaks varied also significantly by match result, F(2, 93) = 13.70, p < 0.001, η² = 0.227. Winning teams recorded more completed line breaks (M = 118 ± 28.6, 95% CI [105.0, 124.7]) than losing (M = 81.0 ± 29.1, 95% CI [73.1, 93.1]) and drawing teams (M = 92.0 ± 19.2, 95% CI [81.6, 97.1]; both p < 0.001).

Total distance covered (Figure 4) was significantly higher in wins (Mdn = 110.4 km, IQR = 10.1; M = 111.56 ± 5.5 km, 95% CI [109.68, 113.43]) compared to draws (W = 3.85, p = 0.018) and higher in losses (Mdn = 111.9 km, IQR = 9.35; M = 111.59 ± 5.9 km, 95% CI [109.56, 113.63] than draws (Mdn = 106.3 km, IQR = 5.8; M = 107.55 ± 4.9 km, 95% CI [105.56, 109.53]); W = −3.60, p = 0.029), showing that teams that won or lost covered approximately 4 km more than teams in drawn matches, suggesting that total distance may reflect match intensity but is not a reliable discriminator of match outcome. However, the effect of match outcome on zone 4 sprinting (20–25 km/h) did not reach the threshold of significance (F(2, 93) = 2.91, p = 0.060, η² = 0.059, (Win: Mean = 5.49, SD = 0.76, 95% CI = 5.75 to 6.02, Lose: Mean = 5.42, SD = 0.91, 95% CI = 5.73 to 6.04, Draw: Mean = 4.84, SD = 1.02, 95% CI = 5.25 to 5.66)).

Receptions in the final third were higher in wins (Mdn = 126, IQR = 117; M = 165.37 ± 102.7, 95% CI [130.09, 200.66]) than losses (Mdn = 82, IQR = 46; M = 91.86 ± 59.4, 95% CI [71.45, 112.27]; W = 5.57, p < 0.001), but no significant differences were observed when compared to the drawing teams (Mdn = 108.5, IQR = 75.0; M = 123.73 ± 81.9, 95% CI [90.64, 156.83]; p > 0.05).

No significant differences were observed for crosses (p = 0.237). Teams that lost matches had a median of 14.0 crosses per game (IQR = 10.5), with a mean of 14.54 (SD = 8.11) and a 95% confidence interval ranging from 11.76 to 17.33. Teams that drew recorded a median of 16.0 crosses (IQR = 13.75), a mean of 19.15 (SD = 13.90), and a 95% confidence interval between 13.54 and 24.77. Winning teams had a median of 17.0 crosses (IQR = 9.0), a mean of 18.91 (SD = 11.27), and a confidence interval ranging from 15.05 to 22.78.

Regarding second balls, losing teams had a median of 57.0 recoveries with an IQR of 23.5, a mean of 55.67 (SD = 15.44), and a 95% confidence interval from 60.97 to 66.28. Drawing teams had a median of 61.0 with an IQR of 15.0, a mean of 58.74 (SD = 11.60), and a confidence interval between 63.42 and 68.11. Winning teams demonstrated the highest median of 65.0 s ball recoveries with an IQR of 14.5, a mean of 62.01 (SD = 13.60), and a confidence interval ranging from 66.69 to 71.36 (Figure 5). Ball progressions were greater in wins (Mdn = 24, IQR = 8; M = 24.46 ± 8.1, 95% CI [21.69, 27.22]) compared to draws (Mdn = 17.5, IQR = 7; M = 19.35 ± 6.8, 95% CI [16.58, 22.11]; W = 3.95, p = 0.014), with marginal differences versus losses (Mdn = 20.0, IQR = 10.5; M = 19.78 ± 7.5, 95% CI [17.21, 22.34]; p = 0.055).

Defensive pressures were significantly higher in wins (Mdn = 198, IQR = 74; M = 209.86 ± 57.5, 95% CI [190.10, 229.61]) compared to losses (Mdn = 271, IQR = 87; M = 283.20 ± 66.9, 95% CI [260.23, 306.17]; W = −5.03, p = 0.001) and draws (Mdn = 214.0, IQR = 62.0; M = 220.92 ± 49.5, 95% CI [200.94, 240.91]; p < 0.001), with no difference between draws and wins (p = 0.592). A one-way ANOVA on direct defensive pressures showed a significant effect of match result, F(2, 93) = 4.43, p = 0.015, η² = 0.087. Post hoc comparisons indicated greater direct defensive pressures in wins (M = 43.00±, 95% CI [39.19, 46.76]) compared to losses (M = 48.00±, 95% CI [46.61, 53.22]; p = 0.019). Differences between draws and losses (draws: M = 43.70 ± 11.1, 95% CI [39.21, 48.19]; p = 0.064) or draws and wins (p = 0.962) were not significant.

Regarding the forced turnovers, the ANOVA did not reveal statistically significant differences between match outcome groups (F(2, 93) = 1.66, p > 0.05). Descriptive statistics revealed that teams that lost had a mean forced turnovers value of 41.66 (SD = 7.38, 95% CI = 39.12, 44.19). Teams that drew recorded a similar mean of 41.54 (SD = 7.79, CI = 38.39, 44.68). Winning teams exhibited a slightly higher mean forced turnovers of 44.40 (SD = 6.62, CI = 42.13, 46.67).

Regarding the percentages (%) of phases of play, a one-way ANOVA revealed a significant effect of match outcome on build up opposed, F(2, 93) = 5.58, p = 0.005, η² = 0.107, indicating differences across loss, draw, and win groups (Figure 6). Specifically, the mean values were 14.49 ± 3.3 (95% CI [13.37, 15.61]) for losses, 13.62 ± 3.5 (95% CI [12.21, 15.02]) for draws, and 16.46 ± 3.6 (95% CI [15.21, 17.71]) for wins. Non-parametric Kruskal–Wallis tests indicated significant differences by match outcome for: Build up unopposed (χ²(2) = 7.53, p = 0.023, ε² = 0.079), with means of 28.29 ± 9.4 (95% CI [25.07, 31.50], median = 26, IQR = 16.5) in losses, 33.54 ± 8.6 (95% CI [30.09, 36.99], median = 35.5, IQR = 11.25) in draws, and 34.40 ± 9.1 (95% CI [31.28, 37.52], median = 36, IQR = 12.5) in wins. Winning teams had significantly higher build up unopposed than losing teams (W = 3.68, p = 0.025), but did not differ from drawing teams (p = 0.883).

In contrast, progression did not differ significantly by match outcome, F(2, 93) = 1.02, p = 0.363, η² = 0.022, with mean progression scores of 18.34 ± 3.4 (95% CI [17.19, 19.50]) for losses, 17.58 ± 3.9 (95% CI [15.99, 19.16]) for draws, and 18.91 ± 3.6 (95% CI [17.68, 20.15]) for wins.

Significant differences by match outcome were also indicated in possession related variables such as final third (χ²(2) = 6.14, p = 0.046, ε² = 0.065). Median values increased from 12.00 in losses to 17.00 in wins. Mean values and 95% confidence intervals were as follows: 13.86 ± 6.00%, 95% CI [11.79, 15.93] for losses; 16.54 ± 10.90%, 95% CI [12.13, 20.95] for draws; and 19.34 ± 10.40%, 95% CI [15.82, 22.87] for wins. Post hoc Dwass–Steel–Critchlow–Fligner pairwise comparisons revealed that wins exceeded losses (W = 3.56, p = 0.032), with no difference between wins and draws (p = 0.343).

Long ball differed significantly among match outcomes, χ²(2) = 8.56, p = 0.014, with a moderate effect size (ε² = 0.090). Post hoc Dwass–Steel–Critchlow–Fligner pairwise comparisons revealed that winning teams recorded significantly fewer long balls than losses (W = −4.03, p = 0.012), with no difference between wins and draws (p = 0.334). Specifically, winning teams showed significantly fewer long ball occurrences (mean = 2.60, 95% CI [1.73, 3.48], median = 2.00, IQR = 3.00, SD = 2.55) compared to losing teams (mean = 4.43, 95% CI [3.30, 5.56], median = 4.00, IQR = 3.00, SD = 3.30). Draws presented intermediate values (mean = 3.42, 95% CI [2.27, 4.57], median = 2.50, IQR = 3.75, SD = 2.85).

Attacking transition also differed significantly, χ²(2) = 7.54, p = 0.023, ε² = 0.079. Attacking transition was significantly lower in wins versus losses (W = −3.78, p = 0.021), but did not differ between wins and draws (p = 0.193), with winning teams showing reduced occurrences (mean = 12.57, 95% CI [10.89, 14.25], median = 11.00, IQR = 6.50, SD = 4.88) relative to losses (mean = 15.74, 95% CI [13.97, 17.52], median = 15.00, IQR = 8.00, SD = 5.18). Draws had intermediate values (mean = 14.50, 95% CI [12.47, 16.54], median = 14.00, IQR = 6.75, SD = 5.04).

Counter attack and set piece did not differ significantly across match outcomes (p = 0.131 and p = 0.087, respectively). For counter attack, losing teams had a median of 2.00, a mean of 2.23 (95% CI [1.75, 2.71]), a standard deviation of 1.40, and an interquartile range (IQR) of 2.00. Drawing teams showed a median of 2.00, a mean of 1.81 (95% CI [1.37, 2.25]), a standard deviation of 1.10, and an IQR of 1.00. Winning teams had a median of 1.00, a mean of 1.63 (95% CI [1.21, 2.05]), a standard deviation of 1.22, and an IQR of 1.00. Regarding set piece, losing teams recorded a median of 6.00, a mean of 6.31 (95% CI [5.68, 6.95]), a standard deviation of 1.86, and an IQR of 2.50. Drawing teams had a median of 6.00, a mean of 6.58 (95% CI [5.74, 7.41]), a standard deviation of 2.06, and an IQR of 2.50. Winning teams showed a median of 6.00, a mean of 5.63 (95% CI [5.01, 6.25]), a standard deviation of 1.82, and an IQR of 2.00.

High press (%) differed significantly among match outcomes (Figure 7), χ²(2) = 9.85, p = 0.007, with a moderate effect size (ε² = 0.104). Post hoc comparisons revealed that winning teams exhibited significantly higher high press values than draws (p = 0.040) and losses (p = 0.012). Descriptively, means and 95% confidence intervals (CI) were 5.14 [4.16, 6.13] for losses (median = 5.00, IQR = 4.00, SD = 2.87), 4.65 [3.56, 5.74] for draws (median = 4.50, IQR = 4.50, SD = 2.70), and 7.40 [6.04, 8.76] for wins (median = 7.00, IQR = 4.50, SD = 3.96).

Mid press (%) did not reach s statistical significance, χ²(2) = 4.87, p = 0.081, ε² = 0.053. Means were 4.31 [3.73, 4.90] for losses (median = 5.00, IQR = 2.00, SD = 1.69), 4.03 [3.24, 4.83] for draws (median = 4.00, IQR = 2.00, SD = 1.97), and 5.00 [4.40, 5.60] for wins (median = 5.00, IQR = 2.50, SD = 1.75).

Low press (%) showed a significant effect, χ²(2) = 6.25, p = 0.044, ε² = 0.066, with winning teams presenting statistically significant lower values than draws (p ≤ 0.05). Means and 95% CI were 0.54 [0.37, 0.72] for losses (median = 1.00, IQR = 1.00, SD = 0.51), 0.62 [0.42, 0.82] for draws (median = 1.00, IQR = 1.00, SD = 0.50), and 0.31 [0.15, 0.48] for wins (median = 0.00, IQR = 1.00, SD = 0.47).

High block (%) was non-significant, χ²(2) = 5.91, p = 0.068, ε² = 0.057. Means were 4.60 [3.62, 5.58] for losses (median = 4.00, IQR = 2.50, SD = 2.85), 4.92 [3.84, 6.01] for draws (median = 4.50, IQR = 3.00, SD = 2.68), and 6.23 [5.12, 7.34] for wins (median = 7.00, IQR = 5.00, SD = 3.24).

Mid block (%) was tested with one-way ANOVA and was not significant, F(2, 93) = 2.24, p = 0.113, η² = 0.046. The means and 95% CI were 20.60 [17.98, 23.22] for losses (median = 21.00, IQR = 9.50, SD = 7.62), 20.15 [17.50, 22.81] for draws (median = 21.00, IQR = 6.75, SD = 6.56), and 17.29 [14.99, 19.58] for wins (median = 17.00, IQR = 8.50, SD = 6.69).

Low block (%) varied significantly, χ²(2) = 10.65, p = 0.005, ε² = 0.112. Winning teams had lower low block values compared to losses (p = 0.004). The means with 95% CI were 26.71 [21.17, 32.26] for losses (median = 22.00, IQR = 22.50, SD = 16.14), 22.19 [16.12, 28.26] for draws (median = 19.00, IQR = 20.50, SD = 15.03), and 16.00 [12.22, 19.78] for wins (median = 12.00, IQR = 12.00, SD = 11.01).

Non-parametric Kruskal–Wallis tests indicated that recovery (%) did not reach statistical significance among match outcomes (p = 0.133). Descriptive statistics (Figure 7) showed that losing teams had a median recovery of 4.0 (IQR = 3.5), mean = 4.83, 95% CI [4.15, 5.51], SD = 1.98; drawing teams had a median of 4.0 (IQR = 3.0), mean = 4.31, 95% CI [3.58, 5.04], SD = 1.81; and winning teams had a median of 5.0 (IQR = 4.0), mean = 5.54, 95% CI [4.69, 6.39], SD = 2.48.

Defensive transition (%) showed significant differences (χ²(2) = 7.64, p = 0.022, ε² = 0.080), being higher in wins compared to losses (W = 3.79, p = 0.020), but not differing significantly from draws (p = 0.701). The median defensive transition was 11.0 (IQR = 6.5) for losses, mean = 12.57, 95% CI [10.89, 14.25], SD = 4.88; 14.0 (IQR = 7.0) for draws, mean = 14.54, 95% CI [12.49, 16.59], SD = 5.07; and 15.0 (IQR = 8.0) for wins, mean = 15.77, 95% CI [13.98, 17.56], SD = 5.20.

Counter-press (%) was significantly χ²(2) = 9.19, p = 0.010, ε² = 0.097) higher in wins compared to losses (W = 4.14, p = 0.010), with no significant difference compared to draws (p = 0.464). The median counter-press was 8.0 (IQR = 4.0) for losses, mean = 9.00, 95% CI [7.94, 10.06], SD = 3.08; 10.0 (IQR = 4.25) for draws, mean = 10.23, 95% CI [8.97, 11.49], SD = 3.12; and 11.0 (IQR = 4.0) for wins, mean = 11.29, 95% CI [10.10, 12.47], SD = 3.45.

Regarding set-piece related activities (Figure 8), total set plays showed a marginal effect of match outcome, F(2, 93) = 3.09, p ≤ 0.050, η² = 0.062, but pairwise comparisons revealed no significant differences between losing (M = 30.49, 95% CI [28.20, 32.77], SD = 6.65), drawing (M = 34.31, 95% CI [31.39, 37.23], SD = 7.23), and winning teams (M = 33.97, 95% CI [31.59, 36.35], SD = 6.93).

Free kicks did not differ significantly across match outcomes, with means of 12.60 (95% CI [11.14, 14.06], SD = 4.26), 12.62 (95% CI [10.75, 14.49], SD = 4.63), and 13.83 (95% CI [12.32, 15.33], SD = 4.38) for losing, drawing, and winning teams, respectively.

Similarly, penalties showed no significant differences, with means near zero for all groups (Lose: M = 0.09, 95% CI [−0.01, 0.18], SD = 0.28; Draw: M = 0.19, 95% CI [0.03, 0.36], SD = 0.40; Win: M = 0.20, 95% CI [0.04, 0.36], SD = 0.47).

Throw ins did not show significant differences, with means of 14.11 (95% CI [12.84, 15.39], SD = 3.72), 15.73 (95% CI [13.93, 17.53], SD = 4.45), and 15.00 (95% CI [13.25, 16.75], SD = 5.08) for losing, drawing, and winning teams, respectively.

Corners differed significantly, χ²(2) = 6.97, p = 0.031, ε² = 0.073, where draws recorded more corners (Mdn = 4.5, M = 5.77, 95% CI [4.13, 7.41], SD = 4.06) compared to losses (Mdn = 3.0, M = 3.69, 95% CI [2.82, 4.55], SD = 2.52), with a significant pairwise Wilcoxon test (W = 3.39, p = 0.043). Winning teams had a median of five corners (M = 4.94, 95% CI [3.98, 5.90], SD = 2.80), not differing significantly from draws or losses.

3.3. Logistic Regression Analysis

A logistic regression analysis (

Table 1. Significant predictors from logistic regression analysis for match outcome (win vs. draw/loss).

Predictor	B	SE	Odds Ratio	95% CI Lower	95% CI Upper	Wald	p
Attempts at Goal on Target	0.333	0.118	1.395	1.102	1.754	8.001	0.005 *
Defensive Pressures	−0.019	0.008	0.982	0.966	0.997	5.443	0.020 *
Total Passes (Complete)	0.009	0.003	1.009	1.003	1.015	8.624	0.003 *
Total Distance Covered (km)	0.140	0.067	1.150	1.008	1.268	4.333	0.037 *
Receptions in Final Third	−0.015	0.006	0.985	0.973	0.998	5.376	0.020 *

Note: * p < 0.05.

) was conducted to examine the association between multiple match performance metrics and match outcome (win vs. non-win, with win coded as 1). Nine successive models (M₀ to M₉) were estimated, progressively incorporating additional predictors to evaluate their influence on the likelihood of winning.

Across models, the decrease in deviance and the improvements in Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were minimal and statistically non-significant (all ΔΧ² p-values > 0.05), indicating limited enhancement in model fit with the inclusion of additional predictors. The pseudo R² statistics (McFadden, Nagelkerke, Tjur, and Cox and Snell) were consistently very low, suggesting that the models explained only a modest proportion of the variance in match outcomes.

Despite the overall limited explanatory power, certain performance variables consistently emerged as significant predictors across models. Specifically, attempts at goal on target demonstrated a significant positive association with winning, as each additional shot on target increased a team’s odds of winning by approximately 40%, reflecting the critical value of offensive efficiency. For instance, in the fully adjusted model (M₉), each additional attempt on target increased the odds of winning by approximately 40% (OR = 1.40, 95% CI [1.10, 1.75], p = 0.005). This underscores the critical importance of offensive accuracy in determining match success. Defensive pressures were inversely associated with winning (OR ≈ 0.98 across models, p < 0.05), suggesting that a higher frequency of defensive pressures corresponded to a decreased likelihood of victory. In practical terms, a team performing 10 additional defensive pressures would experience an approximate 17% reduction in their odds of winning, indicating that frequent defensive activity may be symptomatic of teams lacking sustained control. This finding may reflect the possibility that teams facing greater defensive pressure are more likely to concede possession or control. Total passes completed showed a modest but significant positive effect in later models, for example, in M₉ (OR = 1.009, 95% CI [1.003, 1.015], p = 0.003), indicating that efficient ball circulation contributes to positive match outcomes. Specifically, an increase of 100 completed passes was associated with a 9% greater chance of winning, supporting the value of accurate ball movement. Total distance covered exhibited a positive association with winning (OR = 1.15, 95% CI [1.008, 1.27], p = 0.037 in M₉), which may reflect the physical effort and work rate associated with success. Receptions in final third were negatively related to winning in the final model (OR = 0.985, 95% CI [0.973, 0.998], p = 0.020), suggesting that, quantitatively, each additional reception in this zone was linked to a 1.5% decrease in the odds of winning. In practical terms, an increase of 10 receptions corresponded to a 14% lower likelihood of victory. This suggests that frequent entries into advanced areas, without productive final actions, may reflect ineffective possession in the final third. Other variables, such as defensive line breaks, forced turnovers, crosses, and zone 4 sprinting, did not achieve statistical significance throughout the modeling process.

Tolerance and variance inflation factors (VIFs) indicated acceptable multicollinearity among included predictors, with VIF values ranging from 1.55 to 3.93, suggesting no severe collinearity issues that would compromise model estimates.

The final model exhibited satisfactory classification metrics, with an overall accuracy of 80.2%, an area under the ROC curve (AUC) of 0.85, sensitivity of 65.7%, specificity of 88.5%, and an F-measure of 0.71. These indices suggest the model has good discriminative capability to correctly classify match outcomes based on the included performance variables.

The ROC curve analysis (Figure 9) demonstrated the strong discriminative ability of the final logistic regression model, with an area under the curve (AUC) of 0.85, indicating excellent predictive accuracy. At the optimal classification threshold (cutoff = 0.5), the model achieved an overall accuracy of 80.2%, correctly classifying a large majority of match outcomes. The sensitivity, or true positive rate, reflecting the model’s ability to correctly predict winning matches, was 65.7%, while specificity, the true negative rate indicating correct identification of non-winning matches, was notably higher at 88.5%. These performance metrics demonstrate the model’s effectiveness in distinguishing between wins and non-wins based on the selected performance variables. Furthermore, multicollinearity diagnostics yielded variance inflation factor (VIF) values ranging from 1.55 to 3.93, confirming that multicollinearity among predictors did not compromise the reliability of the coefficient estimates.

Following the binomial logistic regression analysis, a supplementary diagnostic analysis was performed using the DiagROC module in Jamovi to further evaluate the classification performance of key match performance variables (

Table 2. Diagnostic performance of individual key predictors for match outcome based on ROC curve analysis.

Predictor	AUC	95% CI	p	Cutoff	Sensitivity (%)	Specificity (%)	Accuracy (%)	PPV (%)	NPV (%)
Attempts at Goal on Target	0.74	0.65–0.84	0.001 *	≥3.5	80.0	60.7	67.7	53.9	84.1
Total Passes (Complete)	0.76	0.66–0.87	0.001 *	≥478.5	62.9	83.6	76.0	68.7	79.7
Receptions in Final Third	0.71	0.61–0.82	0.001 *	≥89.5	82.9	52.5	63.5	50.0	84.2
Defensive Pressures	0.70	0.59–0.81	0.001 *	≤245	77.1	57.4	64.6	50.9	81.4
Total Distance Covered (km)	0.58	0.47–0.70	0.169	≥105.9	88.6	31.2	52.1	42.5	82.6

Note: * p < 0.05.

and Figure 10). This analysis involved constructing Receiver Operating Characteristic (ROC) curves for selected predictors to assess their ability to discriminate between match outcomes (win vs. non-win). Optimal cutoff points were identified using Youden’s Index, and diagnostic accuracy measures including sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy were calculated. The ROC analysis complemented the regression results by providing practical thresholds and quantifying the discriminative power of individual performance metrics. It is worth highlighting that a threshold of four or more shots on target yielded 80% sensitivity and 60.7% specificity (AUC = 0.74), suggesting this is a practical benchmark for offensive productivity. Similarly, teams completing at least 478 passes were classified correctly in 76% of cases (AUC = 0.76), indicating that high passing volume is a strong indicator of winning performance. Receptions in the final third showed a cut-off point of 89.5, with high sensitivity (82.9%) but lower specificity (52.5%) and an AUC of 0.71, suggesting that while common in winning teams, this indicator may require combination with other metrics to improve predictive precision. Defensive pressures below a threshold of 245 were associated with winning outcomes, achieving 77.1% sensitivity and 57.4% specificity (AUC = 0.70), supporting the interpretation that lower defensive workload correlates with match success. Although total distance covered ≥105.9 km was frequently observed among winning teams and demonstrated high sensitivity (88.6%), meaning that 88.6% of winning teams exceeded this threshold, the corresponding low specificity (31.2%) indicated poor discriminatory capacity, as the majority of non-winning teams also met this criterion. In other words, while this threshold correctly identified most winners (true positives), it failed to exclude many non-winners (false positives). Moreover, the area under the ROC curve (AUC = 0.58) reflects limited overall predictive power, which is only marginally better than chance level (AUC = 0.50), suggesting that total distance alone is an insufficient indicator for reliably predicting match outcomes.

4. Discussion

4.1. Impact of Scoring the First Goal and Temporal Patterns of Goal Scoring

The present analysis confirmed the significant influence of scoring the first goal on match outcomes in the FIFA Club World Cup, with teams that scored first winning 62.5% of matches and losing only 10.4%. This finding aligns with an established body of literature demonstrating the competitive advantage conferred by gaining an early lead [3,4,5,6,7,8,9]. Prior studies reported comparable or higher win rates following the first goal in domestic and international contexts. For example, in the Chinese Super League, Liu et al. [3] found a 66.3% win rate and an 87.01% unbeaten rate for teams scoring first, while in European women’s leagues, Sánchez-Murillo et al. [4] observed that 75.9% of match winners had scored first. Similar trends were documented in the K-League [6], UEFA Champions League [5], and across major European men’s leagues [10]. On the international stage, Martínez and González-García [7] found that in FIFA World Cup and UEFA Euro knockout phases, teams scoring first won 78.46% of the time, although this advantage diminished in extra time (62.5%) and during penalty shootouts (57.14%). In the UEFA Champions League, a more stratified tournament, Parim et al. [5] reported that teams scoring first had a 92% win rate when facing weaker opponents, 71% against balanced teams, and 59% against stronger ones. The UEFA Euro 2024 analysis revealed a significant association (χ² = 8.167, p = 0.017) between scoring first and positive outcomes, with only four of thirty-six teams losing after scoring first [8]. Similarly, Lago-Peñas et al. [10] observed that home teams scoring first in Europe’s top leagues secured 84.85% of available points, while away teams scoring first still won 76.25% of matches. These findings collectively confirmed that scoring first offered a substantial competitive advantage across contexts. However, the current study contributed uniquely by situating this advantage within the specific framework of the FIFA Club World Cup. Unlike domestic leagues or continental tournaments, the Club World Cup comprised champions from six confederations with disparate tactical systems, seasonal calendars, and resource levels. The tournament’s short, knockout-based format, often played at a neutral venue, reduced contextual familiarity and diminished traditional advantages such as home support and environmental adaptation. As a result, although the win rate after scoring first (62.5%) remained considerable, it was notably lower than in more structurally stable settings (e.g., 78–92% in [4,5,6,7]). This moderation likely reflected the inherent unpredictability of intercontinental encounters, where dominant ball possession or tactical superiority could not always be converted into control after an early goal. Thus, this study revealed that while scoring first was an important predictor, its influence may have been less deterministic in global competitions where contextual asymmetries and limited preparatory time reduced the capacity to consolidate psychological and tactical momentum.

Interestingly, temporal distribution of goals was balanced between the first and second halves, with no significant difference overall, although offensive peaks were observed near the end of each half (30–45 and 75–90 min). This pattern is consistent with previous research showing heightened goal-scoring activity toward the close of halves, potentially driven by tactical adjustments or player fatigue [13,14]. These findings therefore emphasized the necessity for sustained tactical vigilance throughout the entire match and highlight critical periods in which intensified offensive pressure may yield decisive scoring opportunities [4,15]. The roughly equal split of goals in the first and second halves (49.3% and 50.7%, respectively) further supports the need for consistent focus across the full match duration.

4.2. Offensive and Defensive Performance Indicators and Physical and Transition Metrics

Consistent with existing literature emphasizing the value of possession and offensive precision [1,2,8,11], winning teams in our study exhibited significantly greater ball possession percentages (M = 53.3%, SD = 12.8, 95% CI [48.9, 57.7]) compared to draws (M = 45.6%, SD = 10.0, 95% CI [41.5, 49.6]) and losses (M = 39.1%, SD = 12.4, 95% CI [34.9, 43.4]). Winning teams also scored significantly more goals (M = 2.91, SD = 1.9, 95% CI [2.28, 3.55]) relative to draws (M = 0.92, SD = 1.2, 95% CI [0.45, 1.39]) and losses (M = 0.51, SD = 0.7, 95% CI [0.26, 0.77]). Similarly, attempts at goal on target were greater in wins (M = 6.17, SD = 3.4, 95% CI [4.99, 7.35]) than losses (M = 3.17, SD = 2.3, 95% CI [2.38, 3.96]).

These data confirm previous findings that possession dominance facilitates offensive control and creates more scoring and winning opportunities [25,47,48,49]. Additionally, winning teams demonstrated superior passing performance in total completed passes (M = 519.69, SD = 170.6, 95% CI [461.10, 578.28]), highlighting effective ball circulation as a critical component of successful play [16,17]. This pattern supports the notion that high-quality passing, more than sheer passing volume, is pivotal in advancing play and maintaining offensive pressure [1,2]. Taken together, these results reinforced the concept that possession-based, precise offensive strategies strongly correlate with match success. In line with this, De-la-Cruz-Torres et al. [50] found that parabolic shots, typically emerging from well-structured attacking phases, were the most frequent shot type in both men’s (59.86%) and women’s (67.12%) European Championship matches, further underscoring the importance of sustained possession and coordinated build-up play in generating technically demanding and strategically advantageous shooting opportunities. Complementing this view, Mitrotasios et al. [51] demonstrated that one-touch finishing, particularly within the penalty area and following assists from crosses or cutbacks, significantly increased goal-scoring probability, suggesting that decisive execution in the final third is a critical component of offensive efficiency and should be a focal point in tactical and technical training.

Although the present study found only a marginal effect of match outcome on total set plays, with no statistically significant differences between winning, drawing, and losing teams, this does not diminish their tactical relevance (drawing teams: M = 34.31 ± 7.23, 95% CI [31.39, 37.23]) and winning teams: M = 33.97 ± 6.93, 95% CI [31.59, 36.35]) executed slightly more set plays than losing teams: M = 30.49 ± 6.65, 95% CI [28.20, 32.77], though these differences did not reach statistical significance in pairwise comparisons). As highlighted by Sarmento et al. [52], set pieces, particularly corners and penalties, are critical phases of play that can influence match outcomes, and their effectiveness depends on contextual, technical, and organizational factors that may not be fully captured by frequency alone. Supporting this, recent evidence from Plakias et al., [53] demonstrated that while the overall goal conversion rate from corners remains relatively low (3.09%), outswinging deliveries significantly increase the likelihood of a final attempt (OR = 0.79, p = 0.02), even though scoring rates do not differ between delivery types. These findings underscore that strategic delivery choices in set plays may enhance offensive productivity and mitigate defensive vulnerabilities such as counterattacks. Consequently, future research should delve deeper into delivery zones, marking systems, and opponent strategies to fully evaluate the functional contribution of set pieces to performance outcomes.

Regarding defensive performance and pressure indicators, the present study revealed an inverse relationship between defensive pressures and match success. Losing teams faced higher defensive pressures, with a mean of 283.20 (SD = 66.9, 95% CI [260.23, 306.17]) compared to winning teams (M = 209.86, SD = 57.5, 95% CI [190.10, 229.61]). This is consistent with prior research indicating that teams subjected to sustained pressure struggle to maintain possession and dictate play, resulting in reactive rather than proactive defensive postures [19]. Furthermore, winning teams executed significantly more completed line breaks (M = 118, SD = 28.6, 95% CI [105.0, 124.7]) compared to losses (M = 81.0, SD = 29.1, 95% CI [73.1, 93.1]) and draws (M = 92.0, SD = 19.2, 95% CI [81.6, 97.1]), indicating superior ability to penetrate opponent defensive lines and create attacking opportunities, a finding aligned with literature highlighting the importance of defensive line penetration in elite performance [11,12]. This is further reinforced by Armatas [54], who demonstrated that transitions beginning in advanced areas and targeting disorganized defensive structures significantly increase the likelihood of offensive sector entry, thereby highlighting the vulnerability of teams that apply high defensive pressure without maintaining structural compactness. The results of the present study are further supported by Harrop and Nevill [1], who found that losing teams performed more clearances, tackles, and blocks in the defensive third, suggesting defensive overload rather than controlled strategy. Moreover, recent tactical analyses emphasized the value of organized, proactive defending. During the FIFA World Cup 2022, successful teams were not characterized by higher volumes of defensive actions, but by their ability to reduce available playing space and maintain advanced defensive lines [20]. Similarly, in top-tier European leagues, bottom-tier teams exhibited more last-resort defensive actions, which correlated with poorer outcomes due to reduced technical and transitional control [22].

The positive association between total distance covered and match success observed in our study supports the established link between higher work rate and favorable competitive outcomes [8,18]. Winning teams covered on average 111.56 km (SD = 5.5, 95% CI [109.68, 113.43]), compared to draws (107.55 km, SD = 4.9, 95% CI [105.56, 109.53]) and losses (111.59 km, SD = 5.9, 95% CI [109.56, 113.63]). Although higher physical output was advantageous, receptions in the final third negatively predicted winning, with winning teams averaging 165.37 receptions (SD = 102.7, 95% CI [130.09, 200.66]); however, this high reception volume may reflect territory control without effective execution. This result may reflect sterile possession in advanced zones, lacking effective penetration or outcome-oriented actions necessary to translate possession into scoring opportunities. Similarly, Kite and Nevill [2], analyzing match data from an English League One professional soccer team across the 2012/13, 2013/14 and 2014/15 seasons, found that a higher number of final third entries did not correspond to superior performance. Specifically, during Season One (S1), the team achieved greater goal output despite significantly fewer final third entries compared to Season Three (S3), where 13% of entries led to a shot and only 6% to a shot on target, both rates approximately 5% lower than in S1, indicating that offensive efficiency, rather than quantity of entries, is more determinative of success. This supports the interpretation that higher receptions in the final third may reflect ineffective or non-threatening possession, especially when not coupled with efficient shot creation or conversion, as also noted by Kite and Nevill [2]. Moreover, S1 had fewer penalty box entries yet more goals, implying that direct play and early shooting may have been more effective than extended build-up. These patterns underscore that productive actions in the final third, such as shot creation and conversion, are more impactful than territorial occupation alone. Our findings thus emphasized the need to align physical output with tactical execution, particularly within offensive phases. Supporting this, Michailidis et al. [55] demonstrated that running performance in elite youth players varied significantly by position within a 1–4–3–3 formation, highlighting the importance of tactical specificity in training. Likewise, Vardakis et al. [56] observed strong correlations between external load and physiological adaptation markers (e.g., VO₂max, metabolic zones), irrespective of positional physical capacity. Michailidis et al. [57] further cautioned against non-functional overload in congested microcycles, where increased high-speed and deceleration loads failed to enhance match output. Finally, Kanaras et al. [58] emphasized that while total distance is positively associated with match outcome, high-intensity metrics (e.g., sprints, accelerations) must be calibrated with training-to-match ratios to ensure competitive transfer. These insights are further supported by recent findings from Pan et al. [59], who emphasized that match-running performance is shaped by the interaction between possession status and possession percentage, rather than by physical metrics alone. Higher possession negatively affected running output during in-possession phases but enhanced it during out-of-possession play. Thresholds were identified (e.g., 36% for total distance) beyond which in-possession values surpassed out-of-possession outputs, with positional differences, such as forwards requiring over 60% possession to observe this shift. These findings reinforce the need to interpret physical performance within tactical and contextual frameworks. This is further corroborated by Lorenzo-Martínez et al. [60] who found that players, especially attackers, in teams with higher possession percentages covered less total distance, particularly at low and medium speeds. This supports the notion that greater ball retention may reduce running demands, depending on positional roles and tactical approach. In further support of this view Modric et al. [61] observed significant declines in running performance across 15 min periods during UEFA Champions League matches, attributing the reduction not solely to fatigue but potentially to pacing strategies or tactical decisions. Furthermore, tactical formation significantly influenced positional demands, with players operating in systems with three defensive players showing higher acceleration and deceleration loads, particularly among midfielders [62]. This reinforced the importance of tailoring physical preparation to specific tactical configurations. Similarly, Modric et al. [63] highlighted that contextual factors such as match location and outcome influenced match running performance, yet team and opponent quality did not significantly alter external load metrics, underscoring the primacy of situational over structural determinants. Lastly, recent findings by Modric et al. [64] suggested that the most elite levels of soccer play are characterized by greater running efforts during the defensive phase, especially among players from top-tier leagues, with reduced efforts in offensive phases. These patterns further emphasized that total distance or high-intensity output alone do not guarantee success unless integrated with strategic match dynamics. Accordingly, training and match analysis should incorporate not only volume and intensity metrics, but also contextualized performance indicators aligned with tactical objectives.

Analysis of in-possession phases revealed that winning teams performed more build-up plays under opposition pressure (M = 16.46, SD = 3.6, 95% CI [15.21, 17.71]) and relied less on long ball tactics (M = 2.60, SD = 2.55, 95% CI [1.73, 3.48]) compared to losing teams (build-up plays: M = 14.49, SD = 3.3, 95% CI [13.37, 15.61]; long balls: M = 4.43, SD = 3.30, 95% CI [3.30, 5.56]). These tactical preferences align with findings from recent research advocating possession-oriented styles and progressive passing as optimal for maintaining offensive momentum and control [21,22,25]. Furthermore, winning teams exhibited higher frequencies of high pressing, with a mean of 7.40 (SD = 3.96, 95% CI [6.04, 8.76]), reflecting an aggressive approach to regaining possession and disrupting opposition build-up, tactics known to increase turnover opportunities and reduce opponent efficacy [18,19]. Such sophisticated tactical elements extend understanding beyond isolated performance metrics, highlighting the multi-faceted nature of match success which encompasses technical, physical, and strategic dimensions.

4.3. Logistic Regression Insights and Predictive Value of Key Performance Variables

The logistic regression models identified attempts at goal on target as the strongest positive predictor of winning, emphasizing offensive accuracy as decisive in determining match success. This finding resonates with prior studies underscoring shots on target as critical success indicators [2,8,30]. Specifically, in the present study, each additional attempt on target increased the odds of winning by 40% (OR = 1.40, 95% CI [1.10, 1.75], p = 0.005). Additionally, completed passes and total distance covered also positively influenced match outcomes, reflecting the combined necessity of technical skill and physical endurance [16,18]. Conversely, increased defensive pressures correlated negatively with winning, supporting the idea that pressure intensity imposed by opponents may compromise possession and reduce scoring opportunities [19]. The robust predictive accuracy of our models, as evidenced by AUC of 0.85, may suggest their potential utility in practical performance monitoring and strategic planning for coaching staff.

4.4. Practical Implications

The results of this study offer important practical implications for football coaches and performance analysts aiming to optimize match outcomes in elite competitions such as the FIFA Club World Cup. The significant advantage of scoring the first goal highlighted the need for tactical emphasis on rapid offensive transitions to secure early leads, while the temporal distribution of goals near the end of each half suggests conditioning programs should prepare players to maintain high performance during these critical periods. Winning teams exhibited superior possession metrics, including higher ball possession, total passes, pass completion, and completed passes, indicating that training should focus on technical skills and coordinated ball circulation under pressure. Furthermore, attempts at goal on target emerged as a strong predictor of victory, underscoring the importance of shooting accuracy and decision-making in the final third. The inverse relationship between defensive pressures and winning suggests that teams subjected to intense pressing require enhanced resilience and strategies to counteract opposition pressure, such as effective build-up play. Additionally, greater total distance covered by players in winning matches emphasizes the importance of physical conditioning and sustained work rate. The finding that increased receptions in the final third negatively associated with winning indicates that quality of possession and effective penetration are more decisive than quantity of receptions alone, guiding coaches to prioritize efficient final-third actions. Defensive and transition phases, notably high pressing and counter-pressing, were more prominent in winning teams, suggesting training should incorporate aggressive defensive organization and rapid ball recovery. Finally, the diagnostic thresholds derived from ROC analyses, such as a minimum of 3.5 attempts on target and 478.5 completed passes per match, provide measurable performance benchmarks for practitioners to set objectives and monitor team effectiveness. Collectively, these insights advocate for an integrated match strategy combining offensive precision, possession control, physical endurance, and dynamic defensive pressing to enhance competitive success through informed training and tactical decision-making.

4.5. Limitations and Future Research

Despite the valuable results provided by this study, several limitations should be acknowledged to contextualize the findings and guide future research. First, the analysis focused exclusively on group-stage matches of the FIFA Club World Cup 2025, with knockout-stage fixtures deliberately excluded to avoid inconsistencies arising from extra-time periods. While this decision enhanced data consistency, it limited the generalizability of the findings to the full competition structure, particularly the high-pressure and tactically distinct knockout rounds. Second, although a wide range of technical, tactical, and physical performance indicators were examined, the analysis did not incorporate contextual, tactical, psychological, or environmental factors that may significantly influence match outcomes. For instance, home advantage, the presence or absence of fans, weather conditions, opponent strength, and travel demands were not considered, despite prior research suggesting their impact on team performance. Furthermore, team-specific tactical configurations, such as formation choices, in-game adjustments, or strategic pressing schemes, were not captured by the available data. Psychological variables such as player motivation, team cohesion, and leadership dynamics were also excluded, although these factors may be particularly influential in closely contested matches. In addition, the decision to dichotomize match outcomes into win versus non-win categories, while analytically practical, may have obscured significant differences between draws and losses. Finally, although the application of ROC curve analysis introduced a novel methodological dimension to football performance modeling, future research could expand this approach by incorporating advanced predictive techniques, such as machine learning algorithms, or adopting mixed-methods frameworks that integrate both quantitative and qualitative data. Such advancements would enhance the ecological validity, interpretability, and predictive capacity of performance models in elite football contexts.

Future studies may benefit from conducting comparative analyses across club-level and international competitions, while also integrating gender-based perspectives to enrich understanding of contextual performance influences. The inclusion of player-level psychological metrics, such as decision-making efficiency, stress responses, mental fatigue, or team dynamics, alongside situational variables (e.g., home vs. away status, match importance, or climatic conditions), would offer a more holistic view of the determinants of success. Moreover, longitudinal designs following teams across multiple seasons or tournaments could shed light on how key performance indicators evolve in response to team restructuring, managerial strategies, or player development trajectories.

5. Conclusions

The analysis of the FIFA Club World Cup 2025 data revealed a strong influence of scoring the first goal on match outcomes, with teams scoring first significantly more likely to secure victory. Performance indicators differed notably between winning, drawing, and losing teams across technical, tactical, and physical dimensions. Winning teams demonstrated superior ball possession, higher frequencies of goals and attempts on target, greater passing accuracy, volume, and completed line breaks, all of which were critical contributors to success. Conversely, losing teams showed higher defensive pressures, indicating reactive play under opposition dominance. Temporal analysis of goal scoring showed relatively balanced distribution across match halves and intervals, with minor peaks before halftime and toward match end, suggesting key strategic moments for offensive efforts. Logistic regression modeling identified key predictive performance variables for match outcomes, notably attempts at goal on target, total completed passes, defensive pressures, total distance covered, and receptions in the final third, highlighting the combined influence of offensive precision, efficient ball circulation, physical effort, and defensive dynamics on winning probability. These findings emphasized the multifaceted nature of soccer success, where offensive effectiveness and tactical control synergize with physical performance and defensive resilience.