INTRODUCTION

Much of the scientific literature on football focuses on aspects of physical fitness. Variables such as distance covered by players or speed are easy to measure [1,2]. Although these data are useful, it is difficult to understand such a complex reality through fragmented and isolated physical phenomena. Therefore, the game needs to be examined using other variables, such as technical-tactical variables (goals scored/received, scoring opportunities, shots on target, crosses into the box, etc.) [3].

Football team performance has traditionally been evaluated based on scored goals. Yet, because football is a low-scoring sport, such an approach leads to a fragmented and incomplete evaluation [4-6]. Scoring opportunities, however, are three times more frequent and provide a more accurate statistical reflection of team performance [4]. Opportunities also allow identifying competition differences across various countries [8]. In the same way, shots on target are linked to scoring opportunities and can be performed and/or received by a team during a match. They are also better at explaining team performance than a mere count of received or scored goals. In this sense, metrics such as the “Expected goal” (xG) or “Expected goal on target” (xGOT) have been used to qualify (and not only quantify) a match to understand team performance [9]. Nevertheless, while this data type enables establishing a working hypothesis, it does not allow coaches to solve team performance problems. A technical staff member could be warned against an opponent’s high “expected goal” value, but this parameter will not explain why a team is displaying a weak defense. Therefore, it is important to grasp the original tactics followed and the reasons behind the outcome of each match.

Moreover, crosses into the box represent a technical-tactical action that can explain football performance: indeed, crosses are often used in the completion phase of an attack. Specialised literature has used crosses into the box in different ways to explain a game. Gai et al. [10] identify the technical differences between domestic and foreign football players according to playing-positions based on the crosses. In this line, Zhou et al. [11] and Errekagorri et al. [12] included them as an indicator to verify how the game model evolved over several years. Notable is the study by Fernández-Navarro et al. [13] who found 8 attack styles based on a factor analysis of 97 matches of different leagues. Two of those 8 styles were linked to crosses into the box.

Regarding the impact of crosses into the box on match outcome, McHale et al. [14] studied the Premier League and found that crosses triggered 52% of shots on target. Similarly, Mitrotasios et al. [8] concluded that almost 20% of goal opportunities in European leagues were generated after a cross into the box. The conversion rate of crosses into goals ranges between 3% and 7% depending on different factors [15], but as stressed above, goals, as single events, are not sufficient to explain football team performance.

Therefore, the impact of crosses into the box on team offensive performance raises questions such as: What is the average number of crosses into the box in football compared to other events in the game? Do the TFEFL (Top Five European Football Leagues) present differences regarding crosses into the box? How are crosses into the box linked to other technical-tactical variables? The study objective was therefore to evaluate crosses into the box as a performance indicator compared to other performance indices of the Top Five European Football Leagues.

MATERIAL AND METHODS

RESULTS

Figure 1 compares the average number of crosses into the box per game with technical-tactical actions. During the 2020-2021 season, an average of 24.36 crosses into the box were found per game in the TFEFL. This average exceeded that of other technical-tactical means to finish an attack (Figure 1): key passes (13.84), shots on target (8.80), total shots (22.13), opportunities (10.40) and goals (2.81).

A differentiated analysis between the leagues shows that the SLL presented the highest average number of crosses into the box, with 25.66 per match. Conversely, GL displayed the smallest average number of crosses into the box with 24.03 per match (Table 1). Since the samples did not follow a normal distribution, the Kruskal Wallis test was used to quantify differences among the 5 leagues. The results showed differences (p<0.001) between the tournaments. Additionally, a pairs comparative analysis only reflected differences between two tournaments, SLL and BG (p<0.005) as well as between SLL and FLL (p<0.001). The size of the differences was small in both cases (ES = 0.29 and 0.38, respectively).

We analysed the evolution of the crosses into the box of the TFEFL over 5 seasons, obtaining an average of 25.76 crosses per game (Figure 2). The ISA obtained the highest average (29.95 crosses) during the 2018-2019 season and GBL the lowest average (22.11 crosses) during the 2016-2017 season. Within this range of values, the TFEFL total average tended to drop slightly over the last three seasons. Subsequently, we compared the total average of the 5 competitions across the 5 seasons. The Kruskal Wallis test revealed significant differences among the crosses into the box totals of the 5 major leagues over the last 5 seasons (p<0.001). The 2018–2019 season (26. 93) presented higher values (p<0.05) compared with three other seasons (2016–2017, 2019–2020 and 2020–2021). The 2020–21 season showed lower values compared with the other four seasons (2016–2017, 2017–2018, 2018–19 and 2019–2020). The size of the differences among the seasons ranged from small to moderate (ES = 0.02–0.35). The greatest difference was found between the seasons 2018-2019 and 2020-2021 (ES = 0.35).

Finally, we examined TFEFL crosses into the box focusing on their links with different game sequences: attack “finishing”, ball “circulation”, and “transition” (Table 3). Direct links were found in most cases. Regarding attack “finishing” variables, a moderate relationship was found between the TFEFL crosses and shots on target (r=0.341; p<0.001). This correlation was even stronger in the case of the GB (r=0.401; p<0.001). Most ball circulation variables were also found to be directly linked with crosses into the box (possession time: r=0.442; p<0.001; possession percentage: r=0.484; p<0.001 and average time of each possession: r=0.328; p<0.001). Worthy of note, the GL (Sg: r=0.537; p<0.001; %: r=0.5464) and FLL (Sg: r=0.517; p<0.001; %: r=0.541 p<0.001) revealed strong correlations between crosses into the box and possession measured in seconds and percentage. Finally, some game transition variables presented links such as total ball recoveries (r=0.524; p<0.001) and total lost ball (r=-0.496; p<0.001). An analysis differentiated by leagues showed a close transition link (ball recovery) in the case of the GL.

Table 4 summarises the three predictive models obtained. The first model (attack finishing) explains 21% of the “Crosses into the box” variance as a dependent variable. In this model, the “Shots” is the predictor variable that best explains the model with a standardised coefficient of 0.478 (p<0.001). The second model is based on “ball possession” variables and explains 23.1% of the variance. In this case, the “possession time” (seconds) variable is the one that best explains the model with a standardised coefficient of 1.160 (p<0.001). Finally, the third model (game transition) obtains an R2 which accounts for 33.7% of the variance. In the last case, ‘Recovery Ball’ is the best interacting variable with a standardised coefficient of 0.415 (p=0.000). In all three models, the Durwin-Watson obtained values that rule out autocorrelation. An absence of multicollinearity was verified in all models.

DISCUSSION

The study objective was to analyse the crosses into the box as a performance indicator comparing it with other performance indices of the Top Five European Football Leagues. To do this, a descriptive analysis, correlational analysis and contrast tests were performed. The data obtained showed that crosses into the box could be considered as highly frequent technical-tactical events (occurring almost 25 times per game). This frequency is significant from a tactical perspective because it represents a skill used in the attack finishing phase and therefore directly influences match score. In this way, this quantitative average presents an added value because it is more frequent than other finishing events (goals: 2.81; opportunities: 10.50; key passes:13.84; shots: 22.13; shot on target: 8.80). Therefore, crosses into the box can be described as high-frequency and qualitatively influential events in TFEFL.

Moreover, to meet the study objective, it was necessary to perform a comparative analysis of the TFEFL. Significant crosses in the box differences were found among the 5 tournaments during the 2020-2021 season. The samples did not follow a normal distribution and the differences obtained may owe to the fact that the Kruskal Wallis test does not compare averages, only entire distributions. In any event, an additional peer-reviewed analysis strategy of the different tournaments showed significant differences only between SLL and GL, and between SLL and FLL, with a low effect size. That is, only two out of 10 tournament pairing possibilities reflected significant differences. On the other hand, studies that use multiple European league samples rarely compare them [13, 17-20]. One exception is the study by Mitrotasios et al. [7], in which the authors compared major European leagues. They found significant differences among the four leagues’ tactical variables, except when an "individual play" was performed in the penultimate action, finishing inside the "score pentagon" and the "goal conversion rate". The tactical variables included crosses into box (p<0.001) which triggered goal-scoring opportunities.

Studies on how game models historically evolve reveal that short passes have increased while direct play has declined [11, 21-23]. It was thus necessary to explore crosses into the box over time as this technical-tactical action can be used in any attack style (combination, direct or counterattack). As shown in this paper, however, crosses are one of the most widely used finishing actions in major European leagues. Nevertheless, the results showed significant differences among the last 5 seasons. Crosses cannot be defined as an indicator of game model changes because the range encountered was narrow: between 29.95 (2018-2019 at ISA) and 22.11 crosses per game (2016-2017 in GBL). The low technical-tactical significance of these differences can be explained not only by the non-parametric nature of the data, but also by the small effect size found between the season/league pairs – in which statistically significant differences were found. In this line, Errekagorri et al. [12] did not find differences either in the SLL’s crosses into the box over the period under study.

Finally, based on a third strategy of exploration of crosses into the box, direct links were found between the crosses into the box and game sequences such as attack “finishing”, ball “circulation” and attack-defence “transition”. But these relationships did not show any cause-and-effect relationship between the crosses and the different correlated game sequences. Future observational studies could seek to explain the direct associations found and determine cause-and-effect relationships. In the case of the transition phase, observational designs should explore whether the close link found between crosses and transition variables results from crosses cleared by opponent defences. The same question applies to the relationship with ball possession variables. Another relevant fact is the “average time” indicator of each possession. This indicator presented a moderate relationship with crosses into the box. It better explains the link between crosses into the box and possession from a qualitative viewpoint, since team % of possession sometimes corresponds to numerous short-duration possessions. These types of “ping-pong” possessions distort data interpretation if we do not focus on each average possession time. Fernández-Navarro [13] also found a link between crosses into the box and possession in the final third of the field, and even conceptualised it as an attacking style of play. Finally, in the case of the attack “finishing” phase, the moderate relationship found between shots and scoring opportunities in the present work was supported by the literature in other contexts [8, 24-26]. In the same way, although the conversion rate of crosses into goals is very low, half the crosses are necessary to complete the goal during the second ball after the cross [15]. Based on the regression models obtained, we interpreted that for each unit point increase of the variables “shots”, “ball possession” (seconds) and “ball recoveries”, the crosses would increase by 0.478, 1.160 and 0.415 points, respectively.

This study presented some limitations. Several contextual factors that potentially influence performance were not evaluated, such as the time of the match and team performance level or ranking. Moreover, we did not consider certain anthropometric characteristics that can potentially significantly influence how these actions unfold in the game.

CONCLUSION

Regarding the TFEFL total, a moderate correlation was found between crosses into the box and other ball possession variables such as time, percentage, and average time of each possession. Identical correlations were found for game transition variables such as total ball recoveries and ball recoveries from opponent.

The data obtained in the present work, however, only establishes a working hypothesis. Future studies should explore this data further and address a number of questions: From which location do crosses into the box generate most scoring opportunities? From which location in the box are the crosses most effectively finished? Does numerical superiority in the box have an impact? Does the provisional match score have an effect? Can any sequential play patterns be identified before or after a cross? While the conversation rate of crosses into goals is clearly low [8, 15, 27], crosses should be examined in terms of how they influence second balls and the game cycle, that is, going beyond goals and shots.