There has been a rise in the use of player performance statistics for the recruitment of players in modern football across professional leagues in Europe and North America [1
]. Individual performance statistics such as goals per game, key passes, pass completion rate, tackles won, turnovers, etc. have been used to measure, scout, and define talent. Technical, cognitive, and psychological skills and their contribution to positioning specific skill requirements have been extensively researched as indicators of performance. Athletic ability, coordination, and decision making are considered key performance aspects in both team and individual sports. Going one step further, position-specific skills and specializations have been studied using various on-field and analytical studies.
Skill requirements and degree of mastery vary from position to position. When it comes to football, certain positions require specific traits [1
]. Center midfielders (attacking and defensive) are required to pick up and process information from multiple sources since they spend most of the time at the central parts of the field and are required to have a 360-degree visual input for optimal performance [2
]. Midfielders are required to make quicker decisions in limited space and time on and off the ball compared to other positions.
Team sport involves complex dynamic situations, and variables in the form of players, conditions, tactics, etc. The aspect of the correct and high speed of decision making is present and is considered an indicator of performance across sports despite fundamental differences in both players and sports. There is accumulating evidence that skilled perception is a key aspect of decision making in sport performance [5
]. Studies have shown that professional athletes have enhanced abilities to rapidly learn complex and neutral dynamic visual input. Football involves 22 players in a constantly changing dynamic environment [6
]. Expert football players show an enhanced ability to recognize, process, interpret, and act upon visual input compared to their novice counterparts. The anticipation of play progression seems to be the key. Expert players show a significantly better understanding of “when” and “where” to look in an ever-changing in-game environment [8
]. Even if less skilled players know “when” and “where” to look, they might be forced to keep their eye on the ball due to limited technical skills [9
]. Higher skilled players showed a greater level of automation compared to their less skilled counterparts in soccer-specific technical abilities, enabling them to direct attention and visual resources to the secondary task of visual exploration [2
“Visual search strategy” is the way in which performers continually move their eyes to focus on important features on a screen enabling them to filter relevant information and act upon it [11
]. A set of previous studies have examined visual search strategy mainly in laboratory settings using screens of different sizes and eye-tracking equipment. It has been shown that the number of fixations on the screen is higher in experienced soccer players, while the fixation duration is shorter compared to their novice counterparts [5
]. Skills important to midfielders such as playing a forward pass, passing to an area opposite to where the ball was received from, turning with the ball, and playing a one-touch pass all showed associations with visual exploration [12
Studies conducted on the field or using video analysis are restricted to specific situations in which the observed midfielders are receiving the ball from the defender [13
]. Only restricted research in very specific game situations has been performed on the visual search strategy in a real game situation with a statistically significant number of players [13
]. Previous studies were done in a situation only when the concerned players are in a certain situation on the field viz. a midfielder receiving the ball form the defender.
Visual exploratory frequency (VEF), an important variable in visual search strategy is the frequency of body and/or head rotational movement prior to receiving the ball, engaged in perceiving information away from the ball [14
]. The current study examined a normally distributed set of players over a period of a month and examined their VEF when their team was in possession of the ball. Furthermore, it examined the correlation of scan rates with the individual performance statistics of the respective players.
In the present study, we correlated VEF with individual performance statistics of midfielders from teams that participated in the knockout stages of the Euro 2016 championship. We hypothesized that there would be a positive correlation between VEF and average pass completion rate. Along with a negative correlation between VEF and turnover rate.
The first result of the present study was that the scanning rate of a player was positively correlated with his pass completion rate. Although a considerable amount of exceptions exists, a higher pass completion showed tendencies of higher levels of success in the 2014 FIFA world cup [16
]. It has been shown in multiple studies that a high pass completion rate in individuals and in teams indicates higher individual and team performance [17
]. Passing statistics, especially completion rate, may be a key indicator of individual player performance according to the results in the current study. VEF might play a role in reaching high pass completion rates. The present study shows a significant positive correlation between the VEF (total scans and scans) and pass completion rates of midfielders and may be an indicator that average VEF may contribute to individual passing performance in a real game situation, although the
values shown in correlations were very low (3%–4%). Such differences may still play an important role in determining the ability of an elite player. As in the case of elite youth players, these small differences may result in larger differences over time with deliberate training [19
The second result of the present study was that the scan rate while receiving the ball was negatively correlated with his turnover rate. Possession in soccer seems to be one of the key factors in deciding match outcomes [18
]. The team with players who cause fewer turnovers has a higher possession of the ball. A recent study showed a significant negative correlation between transition scan rate and number of turnovers per minute. However, scan rate per second was not correlated with turnover rate [9
]. These results are in accordance with the results of the present study and suggest that the transition scan performed by the player is more important to keep the ball safe (i.e., not causing turnovers) compared to the other two VEF types. The visual search of the player in concern, right before receiving the ball, seems to be crucial. It may provide further evidence confirming the prediction that a player of higher technical ability may need to allocate less visual attention on the ball and allocate more to visual exploratory searches [9
]. The low
values indicated high variance, which may suggest that although VEF is a piece of puzzle in modeling passing and ball retention performance, it is only a very small piece of the entire model. The current study provides further evidence on the positive relationship of VEF on various performance indices as stated in previous studies [4
One possible source of error may be the lack of split screen footage and the spatial resolution of the videos. Thus, it was sometimes difficult to clearly define a scan, especially when the player was on the far side of the field with respect to the camera. This might also explain why interrater reliability failed to be excellent despite a seemingly easy observation task. A more reliable and objective approach would be to measure the head movement using inertial measurements units (IMU) [12
]. However, in the present study, we aimed at analyzing the top-level players in real game situations, which makes it impossible to use IMU technology. Another approach using video analysis would be to use split screen footage and use machine learning algorithms to increase quality, reliability and control for interrater subjectivity. Furthermore, eye movement as a factor for visual exploration was ignored, and thus, it can be that in some players, we underestimated VEF. Finally, it must be acknowledged that even though correlations turned out to be significant, R2
values were low, questioning their practical relevance. However, as the present study was performed during a competition including the world’s top-class midfielders, even small effects can be key.
The study results confirmed a positive correlation between VEF and average pass completion rates, as well as a negative correlation between VEF and turnovers per minute for transition scans. Scans and transition scans can thus be used as indicators of a player to complete a high percentage of passes. A higher transition scan rate specifically predicts the ability of a player to not lose the ball, i.e., cause fewer turnovers. It seems to be evident from the results that transition scans may be a more specific parameter for measuring ball retention performance compared to scans when it comes to in-game situations.
The current findings may provide perspectives for talent identification and improved coaching methods by focusing on increasing VEF in line with previous research. Focus on improving transition scan rate may provide an improvement in passing and ball retention capabilities of midfield players. The method used to determine the VEF in the current study can be used to study other performance indices as it assures, for the most part, statistically significant normal distribution of points in random scenarios of the game. The method can be further used in future studies along with machine learning algorithms, high-quality footage, and big data to provide us with models that allow a more accurate prediction of passing performance.