Effect of Isometric Mid-Thigh Pull Asymmetry and Change of Direction Speed on Reactive Agility and in Young Football Players

Abstract

Background: Reactive agility (RA) and change of direction speed (CODs) are fundamental abilities that determine performance effectiveness in football. Biomechanical and motor factors, such as lower limb strength, asymmetry, and speed, may influence the level of these abilities. Moreover, reactive agility is a complex ability and may be partially dependent on the level of CODs. Methods: This study aimed to identify the key motor abilities responsible for shaping reactive agility in young football players. The study involved 55 boys aged $15.63\pm 1.56$ years. The following tests were used: the isometric mid-thigh pull (IMTP) for strength assessment, a 30 m sprint test, the 505 change of direction test, and reactive agility tests with and without a ball, utilizing the Skillcourt system. Results: Isometric strength and sprint speed were significantly correlated with the results of the 505 test. However, asymmetry in lower limb strength did not cause statistically significant changes in the analyzed parameters. Reactive agility without the ball showed significant correlations with speed, change of direction performance, and isometric strength. In the case of reactive agility with the ball, significant correlations were observed primarily with change of direction performance and reactive agility without the ball. Additionally, stepwise regression models revealed significant models for the 50 Random Run without the ball, the Star Run Random without the ball, and the 50 Random Run with the ball. Conclusions: Asymmetry in the isometric lower limb strength does not significantly affect the level of reactive agility or the ability to quickly CODs. Similarly, asymmetry in directional changes during running did not significantly impact the level of reactive agility. However, it may still be a contributing factor to increased injury risk in young football players.

1. Introduction

Football is the most popular sport in the world [1,2,3]. Football training is a multifaceted process that requires players to be comprehensively prepared in terms of physical fitness. During a match, players cover significant distances and perform a large number of passes, throws, shots on goal, accelerations, and decelerations [4]. The complexity of this discipline makes adequate motor and psychomotor preparation an essential component of a player’s development [5]. Modern training approaches increasingly emphasize the balanced development of young athletes [5]. One example of a contemporary approach is the focus on maintaining long-term motivation, which is closely tied to the athlete’s mental health and overall well-being [6]. Well-structured training forms the foundation for the future success of young players [7,8,9,10]. Training programs should be appropriately tailored to the individual needs of the athlete, taking into account a well-balanced amount of active and passive rest, as well as exercises aimed at injury prevention or addressing existing dysfunctions [11,12].

Among all abilities, agility plays a dominant role, although it depends on both sensory and motor components [13,14,15,16]. Agility is a multidimensional motor skill involving rapid changes in direction and speed, typically in response to external stimuli [17]. Unlike linear sprinting, it integrates neuromuscular coordination, perceptual processing, and decision-making under dynamic conditions [18,19,20,21]. A high level of agility primarily enables effective gameplay, regardless of the situation on the pitch [22,23]. Previous studies have shown that improving agility enhances players’ on-field performance by increasing passing accuracy and reducing the time required to complete specific tasks [13].

Another key skill during gameplay is change of direction speed (CODs), which enables players to execute both offensive and defensive actions effectively. CODs is the ability to change direction quickly, which does not require a reaction to external stimuli, and is usually referred to as preplanned agility and a closed skill [24,25]. During a match, players may perform more than a thousand changes of direction, which can be crucial for their overall effectiveness [26]. Previous studies suggest that CODs is a primary component of agility, thereby determining its level [27]. However, the ability to rapidly change direction alone may be insufficient in the context of improving agility, which also includes cognitive components [28,29].

A key factor influencing the abilities mentioned above and the movement efficiency of football players is lower limb strength [30]. A higher level of strength enables athletes to execute technical skills under physical pressure, thereby facilitating tactical implementation. [31]. Previous studies emphasize that plyometric training not only increases explosive strength but also enhances agility by improving the structural and functional characteristics of muscles [32]. In the context of assessing lower limb strength, not only is the overall level important, but also the asymmetry between limbs, especially during activities that require rapid changes in speed and direction [33]. Regular kicking with the dominant leg leads to specific motor adaptations, which may result in muscular imbalances between the lower limbs. Moreover, such asymmetries can increase the risk of injury [16,34]. The most common injuries include ankle and knee joint injuries, sprains, contusions, and strains, as well as inflammation of muscles, tendons, and ligaments [35]. These injuries can account for more than 60% of all cases observed in football [36].

Another ability that can significantly improve reactive agility (RA) is speed [37]. Reactive agility, also referred to as unplanned agility and classified as an open-ended skill, requires the involvement of perceptual and cognitive processes [23]. RA, compared to CODs, involves more complex motor reactions that integrate sensory inputs with motor outputs. RA thus requires adaptation to unpredictable patterns, whereas CODs involve executing a preplanned movement [38,39,40]. Therefore, both RA and CODs are distinct skills [41], and thus should not be combined within the same training programs. Analyzing RA is particularly important, as in most cases, a change in the direction of movement occurs in response to unpredictable external stimuli, such as the opponent’s actions [42]. Previous studies among team sport athletes suggest that performance in 10- and 20-m sprint tests may be indicative of both RA and CODs levels [43,44,45,46]. It is important to emphasize that speed may depend on the previously discussed level of strength [47,48,49].

In football, many commonly used physical fitness tests still assess only CODs, rather than true agility [50]. However, it should be noted that effective gameplay depends primarily on a player’s ability to respond to specific match situations, anticipate actions, and make accurate decisions—factors that CODs tests do not account for [25].

Moreover, no universally accepted definition comprehensively describes reactive agility and its components [29]. While some modern tests incorporate an additional stimulus to introduce unpredictability, they still fail to replicate real match conditions fully [50]. However, some systems provide a comprehensive assessment of motor–cognitive performance. One such device is Skillcourt, which enables the evaluation of agility and cognitive abilities in a more ecologically valid environment [51,52,53]. This research device consists of a 5 × 5 m board divided into nine zones. During the trials, players observe a monitor that displays all tasks (movement concepts GmbH, Schweinfurt, Germany). Friebe et al. [51] suggest that, compared to other computer-based tests, Skillcourt better reflects the motor-cognitive demands of football.

Therefore, the main objective of this study is to investigate the relationship between the change of direction, reactive agility, lower limb isometric strength, and speed among young football players. Additionally, the analysis will assess the relationship between lower limb asymmetry in the isometric strength test and the results of the CODs test, as well as the reactive agility assessment.

2. Materials and Methods

2.1. Materials

The study involved 55 young male football players aged $15.63\pm 1.56$ years (

Table 1. Characteristics of the study group.

Variable	$\overline{\mathit{x}}$	Me	sd	Min	Max
Age (years)	15.63	15.90	1.56	13.10	18.50
Body weight (kg)	60.57	59.30	11.71	37.10	83.60
TBW (kg)	39.70	39.60	7.59	24.50	54.00
FAT (kg)	6.31	6.10	2.49	1.40	13.90
FFM (kg)	54.26	53.80	10.40	33.50	73.80
SMM (kg)	30.37	30.00	6.31	17.80	41.60
FAT (%)	10.32	10.30	3.50	3.00	25.60
Body height (cm)	174.67	176.50	10.57	151.0	195.60

TBW—total body water, FAT—body fat mass, FFM—fat-free body mass, SMM—skeletal muscle mass, $\overline{x}$—mean value, Me—median, sd—standard deviation, min—minimum value, max—maximum value.

). The selected study group falls within the period of adolescence, which is a critical phase for the development of motor skills, including agility [54]. Moreover, during this stage, young football players often transition from basic to more advanced levels of training [55]. The participants attend the School of Sports Mastery in Rzeszów, where they engage in daily football training and one motor training session per week. Each participant declared that they had no health issues or injuries within the past six months. All participants and their legal guardians provided written informed consent to participate in the study.

The study was exempted from the requirement for review and approval by a bioethics committee because it does not violate the fundamental ethical principles governing research involving human subjects. All procedures involving human participants were conducted following the ethical standards of the University of Rzeszów Ethics Committee and the 1964 Declaration of Helsinki and its later amendments.

2.2. Methods

The study was conducted at the University Athletics Center—Center for Innovative Sports Research in Rzeszów. Initially, the procedures and protocols for each test were explained. The participants were informed that they could withdraw from individual tasks or the entire study at any time without facing any consequences. Participation in the study was entirely voluntary. After completing all documentation and the introductory briefing, the next stage involved a comprehensive warm-up. This included raising the temperature of all muscle groups, with particular emphasis on the lower limbs.

Strength

Maximum isometric strength was evaluated using the Isometric Mid-Thigh Pull (IMTP) test, conducted with two Gamma force platforms. At the outset of the procedure, each participant individually adjusted the barbell height to align with the mid-thigh level. Following this, participants positioned themselves on the platforms, assuming the initial stance with feet parallel, lower limbs flexed, torso upright, and head aligned with the spine. The deadlift was executed with an overhand grip under isometric conditions [56]. The contraction endured for approximately five seconds, and the test was conducted two times. Verbal encouragement was offered during each attempt. The rest interval between trials was three minutes. The final assessment consisted of the maximum force measured in newtons and the symmetry index ($SI$) between the right and left lower limbs, calculated using the formula outlined by Sato et al. [57].

$$SI={\displaystyle \frac{(\mathrm{higher}\mathrm{value}-\mathrm{lower}\mathrm{value})}{\mathrm{total}\mathrm{value}}}·100\%$$

Player speed was assessed over distances of 5 m, 10 m, 20 m, and 30 m using SMART-SPEED timing gates, which record data with an accuracy of 0.01 s. Participants positioned themselves 30 cm behind the first gate in a standing start position. Upon receiving the signal, they commenced the trial and were instructed to run the specified distance at their maximum attainable speed. Each participant conducted the test once. A repeat trial was allowed only if the participant stumbled or failed to attain maximum running velocity for any reason.

Speed

Player speed was assessed over distances of 5 m, 10 m, 20 m, and 30 m employing SMART-SPEED timing gates, which record data with a precision of 0.01 s. Participants positioned themselves 30 cm behind the initial gate in a standing start stance. Upon receiving the signal, participants initiated the trial, during which they were instructed to cover the designated distance at their maximal attainable speed. Each participant performed the test once. A repeat trial was permitted solely in cases where the participant stumbled or did not attain maximum running speed for any reason.

Change of Direction Speed

Planned agility was assessed using the 505 change of direction speed (CODs) test. The measurement device used was the SMART-SPEED timing gate system (VALD Performance, Brisbane, Australia), which records data with an accuracy of 0.01 s. Participants positioned themselves 30 cm behind the starting line in a standing start position. On their signal, they began the trial. The nearest timing gate was placed 10 m from the starting line, serving as the run-up distance. Crossing the first gate triggered the timing. Participants then had to run 5 m, perform a 180° turn, and return to the starting point by running the same 5 m distance. The test was performed twice: once with a turn using the right foot and once with a turn using the left foot. A repeat trial was allowed only if the participant stumbled or failed to reach maximum running speed [58].

Reactive Agility Test

The final component was the assessment of reactive agility using two tests performed with the Skillcourt device (movement concepts GmbH, Schweinfurt, Germany). This research apparatus consists of a 5 × 5 m board divided into nine zones. During the trials, players observe a monitor that displays all tasks.

• Random Run 50 with and without the ball—At the start, players positioned themselves on the central field of the board. Once ready, fields on the monitor lit up randomly, and the players were required to reach each target as quickly as possible. The total distance to be covered was 50 m [59]. After completing the entire course, the time was recorded with an accuracy of 0.001 s. The test was conducted under two distinct conditions: without the ball and with the ball. A trial was repeated if the player lost control of the ball and it left the boundaries of the board.
• Random Star Run with and without the ball—At the beginning, players positioned themselves on the central field of the board. Once ready, eight fields on the monitor lit up sequentially in a random order. Players had to run to each indicated field as quickly as possible and then return to the central field after each run. Throughout the entire test, players were required to face the monitor at all times [53]. Upon successful completion of the test, the total time was recorded with an accuracy of 0.001 s. The test was performed under two different conditions: with and without the ball. A trial was repeated if the player lost control of the ball and it left the boundaries of the board.

2.3. Statistical Method

Basic statistical measures (arithmetic mean, median, standard deviation, minimum, and maximum values) were used in the study to characterize the group and evaluate the level of selected motor abilities. The normality of distribution was analyzed using the Shapiro–Wilk test. Since the variables deviated from a normal distribution, Spearman’s rank correlation was used to assess the relationships [60]. The analysis was conducted using Statistica 13 software. Additionally, stepwise regression models were performed using RStudio 2023.06.0 (R version 4.5.3). The following indicators were used in the stepwise regression models: Unstandardized Beta ($\beta$), Standard Error (SE), Residual Standard Error (RSE), the ratio of the mean regression sum of squares divided by the mean error sum of squares (F), Adjusted R-squared (Adj. R²), statistical probability (p), and effect size analysis based on Cohen’s (f²) [61].

3. Results

The analysis presented in

Table 2. Numerical characteristics of the motor skills of the studied athletes.

Variable	$\overline{\mathit{x}}$	Me	sd	Min	Max
Load on the left lower limb (kg)	90.06	94.38	23.95	43.20	144.30
Load on the right lower limb (kg)	93.38	93.90	22.25	56.00	138.10
Strength (N/100)	11.50	11.30	3.26	6.00	18.50
Strength (N/kg)	18.82	19.12	3.13	12.03	28.47
$SI$ IMTP (%)	4.85	4.54	3.70	0.15	14.03
Run 5 m (s)	1.15	1.14	0.08	0.99	1.35
Run 10 m (s)	1.91	1.89	0.11	1.73	2.22
Run 20 m (s)	3.24	3.22	0.21	2.92	3.81
Run 30 m (s)	4.51	4.51	0.32	4.03	5.35
CODs 505 L (s)	2.40	2.40	0.15	2.12	2.87
CODs 505 R (s)	2.41	2.42	0.14	2.09	2.75
$SI$ CODs (%)	1.59	1.25	1.26	0.10	6.54
50 Random Run without ball (s)	23.81	22.93	2.63	19.58	30.85
Star Run Random without ball (s)	19.75	19.42	1.81	16.78	25.85
50 Random with ball (s)	35.72	35.87	4.99	26.70	49.77
Star Run Random with ball (s)	31.25	30.68	3.38	24.80	39.30

IMTP—Isometric Mid-Thigh Pull, CODs—change of direction speed, SI—symmetry index, R—right lower limb, L—left lower limb, $\overline{x}$—arithmetic mean, Me—median, sd—standard deviation, min—minimum value, max—maximum value.

showed that young football players demonstrated a strength level of $11.50\pm 3.26$ N/100. Strength per kilogram of body weight averaged at $18.82\pm 3.13$ N/kg. The participants tended to load their right lower limb slightly more, with an asymmetry index of $4.85\pm 0.15$%. The shortest 5 m sprint time was 0.99 s, while the longest was 1.35 s. The 30 m sprint time was $4.51\pm 0.32$ s. CODs were similar for both the left and right legs, with a CODs asymmetry index of $1.59\pm 0.10$%. Reactive agility was assessed using the 50 Random Run and Star Run Random tests without the ball. Players required more time to complete the 50 Random Run without the ball ($23.81\pm 2.63$ s) than the Star Run Random without the ball ($19.75\pm 1.81$ s). Similarly, during the assessment of sport-specific agility, players needed more time in the 50 Random Run with the ball ($35.72\pm 4.99$ s) than in the Star Run Random with the ball ($31.25\pm 3.38$ s).

The relationship between strength, speed, CODs, reactive agility, and sport-specific agility is presented in

Table 3. Correlation matrix between motor skills and reactive agility with ball.

Variable	50 Random Run	Star Run Random
	with Ball (s)	with Ball (s)
Load on the left lower limb (kg)	−0.19	−0.10
Load on the right lower limb (kg)	−0.24	−0.15
$SI$ IMTP (%)	−0.07	−0.20
Strength (N/100)	−0.18	−0.11
Strength (N/kg)	−0.15	−0.10
Run 5 m (s)	0.22	0.24
Run 10 m (s)	0.24	0.26
Run 20 m (s)	0.21	0.24
Run 30 m (s)	0.22	0.24
CODs 505 L (s)	0.40 *	0.20
CODs 505 R (s)	0.28 *	0.22
$SI$ CODs (%)	0.01	0.09
50 Random Run without ball (s)	0.29 *	0.16
Star Run Random without ball (s)	0.43 *	0.23

IMTP—Isometric Mid-Thigh Pull, CODs—change of direction speed, SI—symmetry index, R—right lower limb, L—left lower limb, *—statistical significance.

. Statistically significant correlation coefficients were observed between the 50 Random Run with the ball and CODs for both the left and right legs. The positive correlation coefficients indicated that longer times required to change direction were associated with longer times recorded in the 50 Random Run with the ball. Furthermore, a positive correlation was found between the 50 Random Run with the ball and reactive agility, as assessed by the 50 Random Run and Star Run Random without the ball. In contrast, no statistically significant relationships were found between motor abilities and the Star Run Random with the ball.

The relationship between the analyzed motor abilities and reactive agility without the ball is presented in

Table 4. Correlation matrix between motor skills and reactive agility without ball.

Variable	50 Random Run	Star Run Random
	Without Ball (s)	Without Ball (s)
Load on the left lower limb (kg)	−0.42 *	−0.21
Load on the right lower limb (kg)	−0.49 *	−0.18
Strength (N/100)	−0.45 *	−0.18
Strength (N/kg)	−0.10	−0.35 *
$SI$ IMTP (%)	−0.12	0.00
Run 5 m (s)	0.55 *	0.27 *
Run 10 m (s)	0.57 *	0.30 *
Run 20 m (s)	0.59 *	0.35 *
Run 30 m (s)	0.60 *	0.35 *
CODs 505 L (s)	0.25	0.58 *
CODs 505 R (s)	0.10	0.47 *
$SI$ CODs (%)	−0.08	−0.25

IMTP—Isometric Mid-Thigh Pull, CODs—change of direction speed, SI—symmetry index, R—right lower limb, L—left lower limb, *—statistical significance.

. It was shown that strength, as well as the load on both the left and right lower limbs in the IMTP test, were significantly correlated with the 50 Random Run without the ball test. The results indicated that the greater the strength in the IMTP test, the shorter the time in this agility test. Moreover, it was observed that the higher the strength relative to body weight, the shorter the time in the Star Run Random test without the ball. Furthermore, statistically significant correlation coefficients were found between speed over all analyzed distances and the 50 Random Run and Star Run Random without the ball tests. The correlation values indicate that the shorter the sprint time, the shorter the time in the reactive agility tests. Additionally, significant correlations were observed, showing that the shorter the change of direction time for both the left and right legs, the shorter the time in the Star Run Random without the ball test.

Table 5. Correlation matrix between motor skills and change of direction speed.

Variable	CODs 505 L (s)	CODs 505 R (s)	$\mathit{S}\mathit{I}$ CODs (%)
Load on the left lower limb (kg)	−0.36 *	−0.33 *	0.45 *
Load on the right lower limb (kg)	−0.35 *	−0.32 *	0.55 *
Strength (N/100)	−0.38 *	−0.37 *	0.55 *
Strength (N/kg)	−0.38 *	−0.36 *	0.47 *
$SI$ IMTP (%)	0.04	0.01	−0.01
Run 5 m (s)	0.52 *	0.52 *	−0.33 *
Run 10 m (s)	0.63 *	0.59 *	−0.29 *
Run 20 m (s)	0.63 *	0.60 *	−0.34 *
Run 30 m (s)	0.66 *	0.62 *	−0.35 *

IMTP—Isometric Mid-Thigh Pull, CODs—change of direction speed, R—right lower limb, L—left lower limb, SI—symmetry index, *—statistical significance.

presents the relationship between the examined motor abilities and CODs. It was observed that CODs for both the left and right legs were significantly correlated with strength, as well as with the load on the left and right limbs in the IMTP test. The correlation coefficients were negative, indicating that lower strength was associated with longer change of direction times. Moreover, it was found that a higher CODs asymmetry index was associated with greater strength in the IMTP test. Similar statistically significant relationships were observed between strength and body weight. Additionally, the results showed that the shorter the sprint time over each distance, the shorter the CODs time for both the left and right legs. In contrast, for the CODs asymmetry index, it was observed that a higher asymmetry index was associated with shorter sprint times across all distances. Regarding the relationship between reactive agility and CODs, a statistically significant positive correlation was found between the Star Run Random without the ball and CODs for both the left and right legs.

Table 6. Stepwise regression models for reactive agility.

Variable	Intercept		Right Load (kg)		Strength (N/100)		Strength (N/kg)		Run 10 m (s)		Run 20 m (s)		CODs 505 L (s)		CODs 505 R (s)
	$\mathit{\beta }$	SE	$\mathit{\beta }$	SE	$\mathit{\beta }$	SE	$\mathit{\beta }$	SE	$\mathit{\beta }$	SE	$\mathit{\beta }$	SE	$\mathit{\beta }$	SE
50 RR (s)	7.04	8.88	0.13	0.07	−1.14 *	0.53	0.48 *	0.22	−26.18	15.00	18.29 *	8.55	7.62	3.86	−7.81	3.91
SRR (s)	−3.21	9.94											16.27 ***	4.14
50 RRWB (s)	9.56 *	3.94	−0.02 *	0.01									5.19 ***	1.48
SRRWB (s)	18.77 *	7.25											5.26	3.02

$\beta$—Unstandardized Beta, SE—Standard Error, CODs—change of direction speed, R—right lower limb, L—left lower limb, * p < 0.05, *** p < 0.001, 50 RR—50 Random Run without ball, SRR—Star Run Random without ball, 50 RRWB—50 Random Run with ball, SRRWB—Star Run Random with ball.

presents the stepwise regression models for reactive agility without and with the ball. The analysis showed that overall strength in the IMTP test was a significant predictor in the 50 Random Run without ball test. It was observed that higher strength was associated with shorter test completion times ($\beta$ = −1.14, p < 0.05), while relative strength (Strength N/kg) was also retained as a significant predictor ($\beta$ = 0.48, p < 0.05). Additionally, the 20-m sprint time was also a statistically significant predictor ($\beta$ = 18.29, p < 0.05). The model also included variables such as right leg load, 10-m sprint, CODs 505 L, and CODs 505 R; however, none of these reached statistical significance (p > 0.05). In the Star Run Random without ball test, only one predictive variable was included in the stepwise regression model: CODs 505 L. The analysis showed that shorter change of direction time was associated with better performance in the test ($\beta$ = 16.27, p < 0.001). For the 50 Random Run with ball, significant predictors included right leg load in the IMTP tests ($\beta$ = −0.02, p < 0.05) and CODs 505 L ($\beta$ = 5.19, p < 0.001). In the Star Run Random with ball test, the model included only CODs 505 L, but it did not reach statistical significance (p > 0.05). The analysis also showed that the intercept was statistically significant in the models for 50 Random Run with ball and Star Run Random with ball. It is worth noting that none of the stepwise regression models included the following variables—$SI$ IMTP, 5-m sprint, 30-m sprint, or $SI$ CODs—suggesting limited predictive value in the context of the analyzed agility tests.

The evaluation of the quality of the stepwise regression models is presented in

Table 7. Quality assessment of stepwise regression models.

Variable	RSE	Adj. R2	F	p	f2
50 RR (s)	2.38	0.17	2.55	0.0265 *	0.39
SRR (s) (s)	4.43	0.21	15.47	0.0003 *	0.30
50 RRWB (s)	1.51	0.31	12.74	0.0001 *	0.50
SRRWB (s)	3.23	0.04	3.04	0.0874	0.06

RSE—Residual Standard Error, Adj. R²—Adjusted R-squared, F—the ratio of the mean regression sum of squares divided by the mean error sum of squares, p—test probability, *—statistical significance, f²—Effect size (Cohen’s f²), 50 RR—50 Random Run without ball, SRR—Star Run Random without ball, 50 RRWB—50 Random Run with ball, SRRWB—Star Run Random with ball.

. It was observed that all models were statistically significant, except for the model for the Star Run Random with ball (p = 0.0874). The 50 Random Run without ball model explained 17% of the variance in performance ($Adj.R^2=0.17$), with a Residual Standard Error (RSE) of 2.38 s and a large effect size (f² = 0.39). In the case of the Star Run Random without ball, the model showed a 21% level of fit, with a prediction error of 4.43 s. The 50 Random Run with ball model explained 31% of the variance ($Adj.R^2=0.31$), and its Residual Standard Error was the lowest at 1.51 s. The effect size analysis indicated that the most well-fitted regression model was the “50 Random Run with ball” model, which suggests a strong relationship between the predictors and the dependent variable (f² = 0.50). In contrast, the model for “Star Run Random with ball” showed a weak model fit (f² = 0.06).

Table 8. Diagnostic tests and 5-fold cross-validation results for the reactive agility regression models.

Model	Shapiro–Wilk (p-Value)	Breusch–Pagan (p-Value)	${\mathit{R}}_{\mathit{CV}}^2$	${\mathit{RMSE}}_{\mathit{CV}}$
50 RR (s)	0.068	0.085	0.107	2.574
SRR (s)	0.003	0.807	0.319	1.599
50 RRWB (s)	0.655	0.062	0.218	4.481
SRRWB (s)	0.107	0.013	0.144	3.316

50 RR—50 Random Run without ball, SRR—Star Run Random without ball, 50 RRWB—50 Random Run with ball, SRRWB—Star Run Random with ball, $R_{CV}^2$—Cross-Validated Coefficient of Determination, $RMS{E}_{CV}$—Cross-Validated Root Mean Squared Error.

presents the residual diagnostic test results and the parameters of the 5-fold cross-validation for the four analyzed reactive agility regression models. The Shapiro–Wilk test confirmed the normality of the residual distribution for the full 50 RR model ($\mathrm{W}=0.960;p=0.068$), the 50 RRWB model ($W=0.983;p=0.655$), and the SRRWB model ($\mathrm{W}=0.964;p=0.107$). A statistically significant deviation from a normal residual distribution was observed in the SRR model ($W=0.928;p=0.003$). The Breusch–Pagan test confirmed the homoscedasticity of residuals in the full 50 RR model ($BP=12.498;p=0.085$), the SRR model ($BP=0.060;p=0.807$), and the 50 RRWB model ($BP=5.571;p=0.062$). Statistically significant heteroscedasticity of the residuals was demonstrated in the SRRWB model ($BP=6.152;p=0.013$). In the 5-fold cross-validation procedure, the highest out-of-sample coefficient of determination was recorded for the SRR model ($R_{\mathrm{CV}}^2=0.319$; $RMS{E}_{\mathrm{CV}}=1.599\mathrm{s}$). For the remaining models, this indicator reached the following levels, respectively: $R_{\mathrm{CV}}^2=0.218$ ($RMS{E}_{\mathrm{CV}}=4.481\mathrm{s}$) for the 50 RRWB model, $R_{\mathrm{CV}}^2=0.144$ ($RMS{E}_{\mathrm{CV}}=3.316\mathrm{s}$) for the SRRWB model, and $R_{\mathrm{CV}}^2=0.107$ ($RMS{E}_{\mathrm{CV}}=2.574\mathrm{s}$) for the full 50 RR model.

4. Discussion

The main objective of the research was to examine the association between selected motor skills and sport-specific agility, as well as change of direction speed (CODs), in football. The analysis encompassed speed, reactive agility, change of direction speed, sport-specific agility, lower limb strength, and asymmetry between the left and right limbs, quantified using the symmetry index ($SI$).

The observed relationships between sport-specific agility and change of direction speed confirm the findings of other authors. For example, Matlák et al. [62] demonstrated low values of shared variance (r = 0.03–0.18) between reactive agility (RA) and CODs, which may suggest that cognitive factors play a dominant role as the main determinants of RA. Similar conclusions were presented by Čoh et al. [63], who emphasized that CODs and RA are distinct motor abilities, even when similar movement patterns are used. Moreover, this thesis is supported by the results of our research, which show that correlations in ball tests were significantly weaker. This may indicate that in conditions requiring ball control, technical and perceptual–decisional components play a greater role. No significant determinants were identified for the Star Run Random with the ball test. Regarding reactive agility assessed through the 50 Random Run without the ball, a significant relation was observed with strength and speed. For the Star Run without the ball, notable associations were observed with speed and CODs. These findings imply that strength, speed, and CODs may partially influence the level of reactive agility. Conversely, fewer significant correlations were detected for reactive agility with the ball. Stepwise regression analysis confirms that performance in strength, speed, and CODs contributes to reactive agility, although these are not the sole determinants. The research indicates that reactive agility—both with and without the ball—is influenced not only by motor abilities but also by cognitive and technical factors. [23,64]. For example, elite athletes, in contrast to sub-elite players, are more effective at interpreting changes in body posture, such as hip flexion or lower limb positioning, treating them as crucial visual cues for anticipating opponents’ actions [43,65,66]. This indicates that a higher level of technical skill is associated with better decision-making abilities [43].

According to previous findings, decision-making time is strongly correlated with RA [14,65]. This relationship has also been corroborated in other sports disciplines. In table tennis, the response time to visual stimuli exhibits a significant correlation with RA test outcomes [39]. In basketball, cognitive abilities are integral to enhancing game performance; players demonstrate improved observational skills regarding both opponents and teammates, resulting in more effective passing and decision-making on the court [67].

Pojskic et al. [22] similarly reported a moderate correlation between CODs and RA. In their study, the sole distinction between the two assessments was the players’ awareness of the direction of movement. In contrast, the configuration of the timing gates remained constant. Matlák et al. [62], on the other hand, emphasized that CODs and RA are two distinct motor abilities. Interestingly, the authors employed the SpeedCourt system and comparable movement patterns; however, their results did not substantiate a significant relationship between rapid direction changes and RA. Similar conclusions were drawn by Rauter et al. [68], who also argued that CODs and RA are separate motor skills. Furthermore, they emphasized that these abilities ought to be evaluated separately employing various testing protocols. It is noteworthy, however, that both the studies conducted by Rauter et al. [68] and Matlák et al. [62] were conducted without the ball, which may have contributed to the lack of significant relationships between the analyzed abilities.

It was additionally observed that CODs, a constituent of planned agility, is related to lower limb strength, speed, and the Star Run Random test without the ball (reactive agility). The findings indicated that greater levels of isometric lower limb strength are associated with shorter CODs durations. Similar relationships were reported by Thomas et al. [69], who also analyzed the relationship between these parameters. However, in their case, the correlation was marginally more pronounced. Their study encompassed netball athletes and incorporated a testing protocol that included full-body isometric strength measurements, squat jumps, countermovement jumps, 10-m sprints, and the 505 CODs tests. The correlation between isometric strength and CODs performance was significant, at –0.66. [70]. Earlier research by Thomas et al. [70] focused on collegiate athletes and examined isometric muscle strength, 20-m sprint time, and the 505 CODs test. The authors recommend systematic isometric strength assessments, as they are associated with athletic performance [56]. A high degree of strategic agility allows players to swiftly and efficiently alter their direction while sprinting during gameplay. [71]. It is worth noting that the IMTP test is a reliable tool for assessing athletes’ fitness levels [72,73], as it allows for a one-time evaluation that correlates with various dynamic sports performance tests [74]. Marco et al. reported varied findings in their study concerning young female football players. According to their results, increased isometric strength correlated with reduced sprint times; however, it did not exhibit a significant association with the ability to change direction rapidly [75].

Shorter 30-m sprints and Star Run Random times were associated with faster change of direction performance. The study by Bayraktar et al. focused on female handball players and found that straight-line sprint time significantly influenced CODs test performance [76]. Pereira et al. [77] examined athletes of both sexes and observed moderate-to-substantial correlations between the results of two CODs tests and the ability to generate high speed and power. In contrast, the study by Robbins and Daniel [78] reported inconsistent results, with correlation strength ranging from small to moderate. The correlation between the analyzed variables largely depends on the sprint distance [15]. Moreover, numerous publications confirm the relationship between CODs test results and acceleration speed over distances up to 40 m [15]. It has also been emphasized that acceleration, maximum speed, and agility without the ball share common physiological and biomechanical determinants [79]. In the case of change of direction asymmetry, it was found that a higher asymmetry index was associated with greater lower limb strength and shorter 30-m sprint times. These results suggest that higher levels of speed and strength abilities expose players to a higher asymmetry index. Menzel et al. [80] point out that asymmetry may result from compensatory mechanisms, which include movement technique and posture. It is worth noting that the $SI$ asymmetry index obtained in the conducted studies was 4.85%, which is significantly lower than the 15% threshold commonly accepted in the literature as the safe limit in terms of injury risk and lack of negative impact on agility abilities [80,81,82]. The studies by Espada et al. [83] showed that professional soccer players had LSI indices of 92.57% for the countermovement jump (CMJ) and 89.06% for the drop jump (DJ), corresponding to asymmetries of 7.43% and 10.94%, respectively. These results suggest that the young athletes included in this study exhibited relatively better functional symmetry despite a lower level of sports advancement. However, other studies suggest that lower limb strength asymmetry may reduce the effectiveness of abilities such as changing direction while running or jumping [84,85]. However, research by Kalata et al. [55] suggests that functional asymmetry in young age may decrease with age and training experience, thereby limiting its role in the further motor development of soccer players.

In the study by Read et al. [86], the authors noted that there is no universal threshold value for asymmetry—it depends on the type of physical ability being assessed in a particular test. Furthermore, their analyses revealed that asymmetry exhibited considerable variability, ranging from 5.2% to 14.5%. Nevertheless, this did not influence overall performance in lower limb strength assessments measured through jumping tasks. Comparable findings were documented by Dos’Santos et al. [87], who observed statistically significant differences between the dominant and non-dominant limbs in the IMTP and 505 tests. However, among university students participating in team sports, no statistically significant correlations were identified between limb asymmetry as measured in both tests. Similar findings were reported by Young et al. [88] and Rouissi et al. [89], who concluded that lower limb strength asymmetry measured using the IMTP test does not influence the identification of the dominant side during CODs tests. Maloney et al. [90] reported comparable findings concerning jump height in the drop jump test for the left and right limbs, as well as asymmetry observed in CODs. Among 18 healthy, physically active male participants, they identified a significant correlation between asymmetries measured in both tests; however, these asymmetries did not influence the identification of the dominant limb during change of direction speed.

Strength asymmetry, however, may significantly increase the risk of injury [91,92,93]. Previous studies have shown that even minor asymmetry can contribute to injuries of the joints, tendons, and ligaments in the lower limbs [91]. Research by Croisier et al. [94] demonstrated that regular training aimed at minimizing strength asymmetry significantly reduced injury risk among professional football players. Among young footballers aged 11 to 16, ground reaction force asymmetry measured during landing was associated with an increased risk of injury [95]. Moreover, asymmetry values exceeding 15% were hazardous and were characteristic of individuals with a high frequency of injuries [95]. In the present study, the asymmetry level in the IMTP test was 4.85%, and 1.59% in the 505 test. However, the observed maximum values were 14% and 6%, respectively, which may pose an increased risk of injury.

The results suggest that reactive agility, particularly when combined with ball control, depends on numerous factors, making it challenging to identify its primary determinants. To enhance unplanned agility in football, training sessions should emphasize the recognition of dynamic visual cues, such as changes in an opponent’s body posture [96]. In this context, subsequent research could utilize innovative methods like stroboscopic glasses to assess and develop visual information processing under stimulus-limited conditions [97]. Furthermore, integrating neurocognitive assessments that target brain executive functions—specifically working memory and attentional control—remains a vital future direction, as higher-level players consistently demonstrate superior performance in executive control tasks [98,99]. To fully understand these shifting motor and psychomotor relationships over time, future studies should adopt longitudinal designs utilizing a broader, multi-disciplinary testing battery.

However, it must be acknowledged as a key limitation that while reactive agility inherently demands high levels of perception, anticipation, and rapid decision-making [96], direct cognitive metrics were not quantified in the present study. Although the Skillcourt system imposes substantial visual–spatial cognitive loads on youth players during displacement, the absence of independent, standardized simple reaction-time or choice-response velocity tests represents a baseline gap. Integrating these objective perceptual–cognitive measures directly alongside motor profiles should be considered a critical priority for subsequent predictive modeling to capture the unmeasured variance in multifaceted agility tasks. Additionally, as presented in the diagnostic analysis, the SRR model exhibited a significant deviation from the normality of residuals, whereas the SRRWB model demonstrated significant heteroscedasticity. Consequently, these specific models should be interpreted with caution, despite demonstrating satisfactory performance in cross-validation.

Furthermore, a notable limitation of this study is that physical and mechanical predictors explained a relatively small percentage of the variance during the ball-dribbling conditions (50 RRWB and SRRWB models). This statistical outcome explicitly demonstrates that when sport-specific ball control is introduced, raw physical capacity carries diminished predictive weight, being heavily superseded by technical skill proficiency and sport-specific dribbling mechanics. Consequently, integrating high-fidelity technical skill metrics alongside the aforementioned perceptual–cognitive variables remains an absolute priority for future work attempting to model ball-involved reactive performance in elite youth football cohorts.

5. Conclusions

The study did not show that the degree of lower limb strength asymmetry is a significant predictor of RA level among young soccer players. Moreover, no significant relationship was observed between CODs asymmetry in the 505 test and sport-specific agility. However, the results indicate that the level of lower limb isometric strength and speed may determine the ability to change direction quickly and the extent of its asymmetry. Significantly, a higher level of strength and speed was associated with greater asymmetry in the CODs test. It should be noted that this asymmetry did not influence RA, as assessed by the 50 Random Run with ball and Star Run Random with ball tests. Among all the agility components, only the CODs test results showed significant correlations with RA. These findings suggest that training aimed at improving sport-specific agility should primarily focus on developing CODs ability, as it was the only factor significantly correlated with RA. It should also be noted that RA is composed of many factors, including those related to psychomotor abilities. A higher level of speed and isometric strength in young football players may contribute to greater functional asymmetry, which may slightly improve CODs performance. However, monitoring the degree of asymmetry appears essential in minimizing injury risk and optimizing motor development in young athletes.