Touch restrictions and player numbers during small-sided games in youth soccer players: Effects on physical and technical parameters

Abstract

This study examined how player numbers and touch restrictions influence physical and technical demands during small-sided games (SSGs) in youth soccer. Sixteen U15 players completed 3v3, 5v5, and 7v7 formats under one-touch (1 T), two-touch (2 T), and free play (FP) conditions on regular goals, with a constant relative pitch size, four 3-min bouts, and 3-min passive recovery between bouts. Physical variables (total distance [TD], high-intensity distance [HID >14.4 km/h], sprint distance [SpD >19.8 km/h], peak speed, sprint counts, number of accelerations [ACC > 2 m/s²] and decelerations [DEC > - 2 m/s²]) and technical actions (ball touches, releases, possessions, one-touch plays) were recorded using inertial measurement units.

Generalized linear mixed-effect models, with exponentiated regression coefficients, showed that ball touches increased from 3v3 to 5v5 across all touch conditions (30–39%), whereas ball releases were primarily elevated in larger formats under 2 T conditions (35–37%). Sprint frequency was higher in larger formats across touch conditions, while SpD decreased from 3v3 to 7v7 (−26% to −27%). Increases in DEC in 3v3 were observed only under 2 T (42%) and FP conditions (49%). Isolated increases in 5v5 compared to 7v7 under 2 T (+31%) were shown for HID.

Lower player numbers consistently increased DEC and individual ball involvement, especially under 2 T and FP conditions, whereas larger formats promoted sprint frequency and SpD. These findings suggest that adjusting player numbers is generally more effective than modifying touch restrictions, but combining both strategies can optimize specific physical and technical outcomes in youth players when employing SSGs.

Keywords

Association football high-intensity running load management performance analysis task constraint technical skills

Introduction

Small-sided games (SSGs) are widely used in soccer training, as they enable coaches to elicit specific physiological, physical, technical, and tactical adaptations within a game-based approach.^1–4 SSGs are particularly popular due to their organizational flexibility and adaptability to specific training objectives. Grounded in the principles of ecological dynamics, SSGs represent a form of representative learning design, in which players are exposed to realistic, perceptual-action couplings that closely reflect match conditions.^5,6

Within this framework, the constraint-led approach has been proposed to explain player behavior, which emerges from the interaction between task, individual, and environmental constraints.⁷ Consequently, the manipulation of these constraints, particularly task constraints within SSGs, has become a common and effective strategy for coaches to shape the affordance landscape (i.e., opportunities for action) available to players.⁸ This process promotes the emergence of diverse movement solutions at both the physical and technical levels.^9,10

Among the array of task constraints that can be modified, the number of players involved is one of the most extensively investigated.¹¹ Manipulating the number of players per team alters the frequency of technical involvements per player, thereby influencing overall task demands and the associated physical load (e.g., total distance, TD).¹² Thus, both physical and technical dimensions can be specifically targeted by modifying player numbers, while also shaping learning opportunities for through changes in action density.¹³ Evidence from meta-analytical data suggests that, when key confounders such as area per player (i.e., pitch width x length divided by the number of players) and length-to-width ratios are standardized, larger player numbers are associated with greater physical outputs (e.g., TD).¹² Furthermore, research in youth soccer indicates that increasing player numbers (e.g., 2v2 vs 4v4 vs 6v6) elevates several physical variables, including distance covered at moderate speeds (10–16 km/h), high-speed running (>14–21 km/h), and sprint frequency and duration.^14–17 In contrast, the number of accelerations (ACC; 1.5–2.75 m/s²) and decelerations (DEC; −1.5 to −2.75 m/s²) performed appears to decrease as player numbers increase, consistent with findings in youth soccer,¹⁸ and professional soccer (ACC: 0–4 m/s²).¹⁹

From a technical perspective, decreasing player numbers generally increases individual involvement and engagement,¹² which may enhance opportunities for skill acquisition through more frequent interactions with the ball.²⁰ For example, youth soccer players participating in 3v3 formats with a comparable relative playing area per player (50 vs 55 m²) performed a greater number of ball recoveries, ball receptions, and attacking actions than those in 6v6 formats. However, they also exhibited a higher number of ball losses, likely due to the increased frequency of technical involvements and the corresponding rise in error probability.²¹ A similar pattern was reported in U17 soccer players, where reducing team size by half (2v2 vs. 4v4), while maintaining a similar relative area per player (92–94 m²), resulted in a higher number of passes, accompanied by increased error rates (e.g., unsuccessful passes).²² Regarding ball possession, recent evidence indicates that 2v2 and 3v3 formats elicit more individual ball possessions than 4v4, further supporting the notion that smaller formats increase technical engagement due to the reduced number of players.¹² This may provide players with more frequent decision-making opportunities, albeit often under increased time pressure.²³

Within the range of task constraints commonly employed in SSGs, in addition to changes in player numbers, the restriction of ball touches represents a widely used coaching strategy.^8,24 This contextual variable includes different types of restriction patterns, primarily one-touch (1 T) and two-touch play (2 T), which are often compared to free-play (FP) conditions.²⁵ Reducing the time available on the ball may encourage players to attune to relevant informational cues and adjust their in-game behavior (e.g., spatial exploration, movement patterns), thereby shaping the affordance landscape available to them.²⁵

Offensively, 1 T play has been shown to enable players to bypass defensive pressure more rapidly and create scoring opportunities through quicker ball circulation,²⁶ reflecting its central integration into passing sequences during match play.²⁷ Furthermore, the implementation of a 2 T rule has been suggested to alter tactical patterns, such as defensive cooperation and the interconnectivity of the defending team.²⁸ Consequently, restricting the number of ball touches has become a popular task constraint in SSGs, with effects demonstrated on both physical and technical parameters.^24,29

Empirical studies in youth soccer indicate that FP conditions promote greater overall physical involvement, reflected in higher distance covered per minute and sustained moderate-intensity activity. In contrast, touch restrictions tend to increase walking (0–3.5 km/h) and reduce jogging (3.6–14.3 km/h), without consistently improving high-speed running (>14 km/h).²⁵ However, research in professional soccer shows a different pattern. 2 T play has been associated with greater distance covered (12–15 km/h), moderate (15.3–18 km/h), and high-intensity (13–17 km/h) running compared with free or 1 T conditions.^30,31 Conversely, 1 T constraints have been associated with greater TD, walking (6–8 km/h), high-intensity running (13–17 km/h), sprinting (>17 km/h), and ACC distance (>4 m/s²).^30–32 These contrasting findings likely reflect differences in technical proficiency, as restrictive rules may impose greater technical demands on youth players,²⁵ thereby limiting their physical output.

Regarding technical behavior, youth studies indicate that touch restrictions increase passing frequency,^25,33 and influence passing directionality, such as a greater prevalence of backward passes,³³ but may simultaneously reduce pass accuracy, resulting in a higher number of unsuccessful passes.^25,34 When comparing 2 T play with FP, possession sequences were shorter in duration under 2 T conditions than in both free-form and four-pass play.³⁵ This constraint also reduced the total number of ball touches per sequence, while simultaneously increasing the number of passes per possession. Moreover, the ratio of players involved in each possession was greater under 2 T rules compared to FP.³⁵

Despite extensive research on the independent effects of player numbers and touch restrictions, evidence regarding their combined interaction remains limited. To date, no study has explicitly examined this interaction in youth soccer. Existing reviews addressing player numbers and touch restrictions suggest that their interaction is often acknowledged, particularly in professional contexts, but has not been systematically analyzed.^12,29 In youth soccer, this represents a critical gap, as both constraints constitute key task-related factors that coaches frequently manipulate during SSGs to develop specific technical skills as well as physical capacities.^36,37

Theoretically, formats with less players are expected to increase neuromuscular load and technical involvement due to greater ball engagement, while touch restrictions may further intensify technical demands by reducing decision-making time. However, the extent to which these constraints interact to influence physical output and technical actions remains unclear. Additionally, differences in the physical and technical profiles of youth players compared with adults may moderate the effects of both player-number and touch constraints. Given that game behaviors emerge from the interaction of multiple constraints,^6,7 understanding how player numbers and touch restrictions jointly shape physical and technical demands may provide valuable insights into their combined effects. Such knowledge could help optimize training load and support skill development in youth soccer.

Accordingly, this study aimed to examine how physical and technical variables vary across different player-number formats (3v3, 5v5, and 7v7) within three touch restriction conditions (1 T, 2 T, and FP) in youth soccer players. Based on the existing literature, we hypothesized that: (1) smaller formats would increase accelerative and decelerative demands while reducing TD and high-speed locomotor outputs, particularly under FP conditions; and (2) smaller formats combined with more restrictive touch conditions would increase technical involvement (e.g., ball touches), whereas larger formats would increase ball possession frequency.

Methods

Pilot study

Prior to the main study, a pilot study was conducted using the identical SSG protocol. Each player completed two sets for every Player number x Touch restriction condition (3v3, 5v5, 7v7 x FP, 1 T, 2 T), with the order of formats and touch restrictions randomized and counterbalanced across sessions. To ensure feasibility, games were conducted in parallel on multiple pitches, and temporary non-participation within single game blocks was rotated systematically so that all players completed all conditions twice.

Pilot data were analyzed using gamma log-link generalized linear mixed models (GLMMs), with player ID specified as a random intercept, consistent with the final analytical model. Using the 7v7 FP condition as the reference, proportional effect sizes (exp(B)) were calculated for each dependent variable.³⁸ A 20% effect (exp(B) = 1.20) was defined a priori as the smallest effect size of interest (SESOI), based on the variable ball touches, which exhibited the smallest observed effect and was therefore used for the power analysis.³⁹

Sample size estimation

An a priori power analysis was conducted based on the SESOI (exp(B) = 1.20) and variance components derived from the pilot GLMMs (random intercept SD = 0.1). Power simulations were conducted separately for each dependent variable, preserving the full planned fixed-effects structure (Player number, Touch, and their interaction) and the random intercept for player ID. Monte Carlo simulations (1.000 iterations) indicated that a minimum of 16 players would be required to achieve the targeted statistical power of 0.80⁴⁰ (actual power = 0.82) for detecting the Player number x Touch interaction. To account for multiple fixed-effect contrasts within each model, the significance threshold was adjusted a priori using Bonferroni correction, and this adjusted α-level was incorporated directly into the simulation.

Participants

A U15 youth soccer team (n = 19 players) from a German youth academy participated in the study. The academy competes in the highest regional youth league, with the senior team playing in the 2. Bundesliga. According to McKay et al.⁴¹ players were classified as highly trained/national level (Tier 3). All players trained four times per week (90–120 min per session) and participated in one official match per weekend during the observation period. Training routines typically included technical drills, SSGs, individual positional work, and team-based tactical exercises.

Inclusion criteria for the study participation were: (1) official registration with the club; (2) full participation in all training sessions during the study period; and (3) absence of injury throughout the study and during the preceding three months. Exclusion criteria comprised: (1) incomplete participation during study period; (2) injury or illness limiting full participation; and (3) playing position as goalkeeper, due to the distinct physical and tactical demands of this role.

Although the required sample size was 16 players, all available players (n = 19) were initially considered to account for potential dropouts. Three players sustained injuries before the start of data collection and were therefore not included in the experimental procedures. The final sample comprised 16 players (mean ± SD: age 15.4 ± 1.7 years; height 169.9 ± 7.3 cm; body mass 60.6 ± 8.9 kg; V30-15 IFT 20.1 ± 0.92 km/h; CMJ height 34.1 ± 4.3 cm), thereby meeting the a priori sample size requirement, with all players completing all study sessions. Ethical approval was obtained from the Ethics Committee of Exercise Science and Training, University of Würzburg (EV2026/2-2001), with additional approval from the club. Written informed consent was obtained from all players and their legal guardians, and all procedures complied with the Declaration of Helsinki.

Procedures

Data were collected over a six-week period, with two sessions per week scheduled at the same time of day (4:30–6:00 PM) during the mid-season phase to minimize circadian and seasonal variability. Sessions were scheduled relative to the upcoming match day (MD) and separated by standardised 48-h recovery periods, with no training between data collection sessions, to minimise fatigue-related confounding effects (i.e., MD-4 and MD-2). The duration of the observation and recovery periods was determined based on previous research demonstrating that the player numbers used in this study, which were identical to those in our design, influence physical performance outcomes.⁴²

A crossover design was used in which all players completed 3v3, 5v5, and 7v7 formats across the intervention period.⁴² Given the sample size (n = 16), team composition was adjusted for each format while ensuring systematic rotation of players across all sessions. In 3v3 games, two teams of three competed per bout, with the remaining players rotated systematically to ensure continuous inclusion of all participants across bouts. In 5v5 games, two teams of five competed while the remaining six players were rotated across bouts following a predefined substitution order to ensure balanced exposure. In 7v7 games, two teams of seven competed, with the remaining two players rotated into play to ensure equitable distribution of playing time.

Touch restriction conditions (1 T, 2 T, FP) were systematically counterbalanced across sessions and formats, such that each condition occurred equally often in the first, second, and third position across the intervention period. Although full within-session counterbalancing was constrained by the fixed number of sessions per format, the rotation design ensured distributed exposure to all player number x touch restriction combinations. A standardized 30-min passive recovery period was implemented between conditions to reduce acute fatigue and mitigate carry-over effects. The experimental schedule is provided in Table 1.

Table 1.

Experimental schedule. FP = Free play; 1 T = One-touch; 2 T = Two-touch; P = Player.

Week	Session	Format	Touch restrictions	Starting players	Rotation between bouts
1	1	3v3	2 T → FP → 1T	P1-P6	Shift to P7-P12 → P13-P16 + P1-P2
1	2	5v5	FP → 1 T → 2T	P7-P16	Shift to P3-P12 → P9-P16
2	1	7v7	1 T → 2 T → FP	P1-P14	Bench rotation: P15-16 → P1-2 → P13-14
2	2	3v3	FP → 2 T → 1T	P7-P12	Shift to P13-P16 + P1-P2 → P3-P8
3	1	5v5	2 T → 1 T → FP	P1-P10	Shift to P5-P14 → P11-P16 + P1-P4
3	2	7v7	1 T → FP → 2T	P3-P16	Bench rotation: P1-2 → P13-14 → P5-6
4	1	3v3	FP → 1 T → 2T	P13-P16 + P1-P2	Shift to P3-P8 → P9-P14
4	2	5v5	2 T → FP → 1T	P5-P14	Shift to P9-P16 → P1-P8
5	1	7v7	FP → 2 T → 1T	P1-P14	Bench rotation: P15-16 → P7-8 → P9-10
5	2	3v3	1 T → 2 T → FP	P3-P8	Shift to P9-P14 → P15-P16 + P1-P2
6	1	5v5	1 T → FP → 2T	P9-P16	Shift to P1-P10 → P5-P14
6	2	7v7	2 T → 1 T → FP	P3-P16	Bench rotation: P1-2 → P11-12 → P13-14

Participants adhered to standardized activity routines throughout the observation period, including consistent training modalities and daily schedules. To ensure procedural consistency, all sessions were supervised by the same coaching staff, each holding at least a UEFA B-license, and followed a predetermined schedule specifying drill sequences, durations, and rest periods. The research team continuously monitored adherence and verified that session conditions (e.g., warm-up structure, equipment setup) were replicated precisely across all data collection days.

Before each SSG session, players completed a standardized 10–15-min warm-up on 3G artificial turf consisting of running and change-of-direction drills, passing sequences, and static and dynamic stretching exercises, aligned with the FIFA 11 + protocol. Each SSG format consisted of four 3-min bouts with 3-min passive recovery periods between bouts. All games were conducted with two regulation goals (7.32 × 2.44 m) on standardized pitches designed to maintain a consistent relative area of 215–216 m² per player across all formats similar to previous research,⁴² with a fixed length-to-width ratio of ∼1.5:1. FIFA-approved match balls (410–450 g; 68–70 cm circumference) were used, with additional balls positioned around the field to facilitate quick restarts. Detailed configurations of each SSG format are presented in Figure 1.

Figure 1.

Small-sided games employed: a) 3v3, b) 5v5, c) 7v7.

Inertial measurement units

Physical and technical parameters were recorded using inertial measurement units (IMUs; PlayerMaker™, Tel Aviv, Israel) equipped with triaxial 16 g accelerometers and 2000°/s gyroscopes (MPU-9150, InvenSense, California, USA), recording data at a sampling frequency of 10 Hz. The IMUs were securely mounted on the lateral side of each player's boots and activated 20 min prior to data collection. To minimize inter-unit variability, each player used the same device throughout the study. Following each monitored SSG, data were synchronized and uploaded to the manufacturer's cloud platform and subsequently exported to Microsoft Excel (Version 2108, Microsoft Corporation, Redmond, USA) for further analysis.

Physical variables

Physical parameters included TD, high-intensity distance (HID; >14.4 km/h), and sprint distance (SpD; >19.8 km/h). Additionally, sprint counts, peak speed, and the number of ACC (>2 m/s²) and DEC (>-2 m/s²), were recorded. Good inter-unit reliability and concurrent validity for distance-based metrics (e.g., TD), velocity measures (e.g., peak velocity), and various ACC/DEC bands when compared with GPS systems.^43–45 All physical variables were selected in alignment with previous research conducted in youth academy settings.^45–47

Technical variables

For technical analysis, the following variables were recorded: ball touches (all foot-to-ball contacts), ball releases (passes, kicks, or shots), one-touch plays (single touches where the ball is received and immediately released), and individual ball possessions, defined as either a release action, a sequence of ≥3 touches, or ball movement covering ≥6 m. As ball possessions are constructed from touches and releases, both components have previously been validated, demonstrating high concurrent validity (proportion of agreement: 95.1–97.6%) and high intra-unit reliability (CV: 1–14.35%).^48,49 All technical measures were selected based on previous research in youth soccer.^46,47

Statistical analysis

Descriptive statistics for all dependent variables are presented as means ± standard deviations. Data were screened for outliers, and distributions were visually inspected using Q-Q plots to inform model specification. Factor levels were explicitly coded for touch restriction conditions (1 T, 2 T, FP) and player numbers (3v3, 5v5, 7v7). GLMMs with a Gamma distribution and log link (lme4 package) were fitted for each dependent variable.⁵⁰ This approach was selected after evaluating distributions and link functions for the continuous dependent variables; both normal and gamma distributions with identity and log links were tested, and model fit was assessed using the Akaike Information Criterion (AIC), with the model showing the lowest AIC retained.⁵⁰

Final models included touch restriction condition, player numbers, and their interaction as fixed effects, with a random intercept for player ID to account for within-subject variability. Model fit was further evaluated using marginal (R²_m) and conditional (R²c) coefficients of determination, representing variance explained by the fixed effects alone and by the full model (fixed and random effects), respectively.⁵¹ Significance of main and interaction effects was assessed using Type-3 Wald tests.

Post-hoc analyses were conducted using estimated marginal means (emmeans package). Pairwise contrasts were calculated to compare formats within each Touch condition. Effect sizes were expressed as exponentiated regression coefficients (exp(B)), interpreted as proportional changes in the outcome variable.³⁸ Bonferroni corrections were applied to adjust for multiple comparisons, with statistical significance set at p < 0.05.

To evaluate practical meaningfulness, pairwise contrasts were additionally assessed using two one-sided tests (TOST) for equivalence.^52,53 To reduce the risk of Type I errors associated with subjective threshold benchmarks, the SESOI was defined based on values reported in related studies.⁵² Although the interaction between player numbers and ball touches has not previously been investigated in youth soccer, we used our pilot study to inform the SESOI and provide an estimate of effect magnitude. Effects were considered practically meaningful if their 90% confidence intervals (TOST) exceeded the SESOI threshold. All analyses were conducted in RStudio (Version 2026.01.1 + 403).

Results

Descriptive statistics for physical and technical variables are presented in Figures 2 and 3, with model outputs summarized in Table 2. Conditional R² values indicated moderate to high model fit (0.12–0.73), while random effects were small (SD = 0.00–0.04) and intraclass correlation coefficients (ICCs) were low (0.00–0.06). A complete overview of the full model characteristics is provided in the supplementary material.

Figure 2.

Box plots of the physical variables analyzed. TD = Total distance; HID = High-intensity distance; SpD = Sprint distance; ACC = Acceleration; DEC = Deceleration.

Figure 3.

Box plots of the technical variables analyzed.

Table 2.

Fixed effects, random components, and model fit indices of generalized linear mixed models. TD = Total distance; HID = High-intensity distance; SpD = Sprint distance; ACC = Acceleration; DEC = Deceleration; AIC = Akaike's Information Criterion; SD = Standard deviation; ICC = Intra-Class Correlation.

Fixed effects						Random components		Model fit
Variable	Effect	Sum Sq	Mean Sq	F value	p	SD	ICC	R²_marginal	R²_conditional	AIC
TD	Touches	0.02	0.01	4.91	0.008	0.01	0.06	0.69	0.71	−1077.66
	Player number	0.76	0.38	204.17	<.001
	Touches x Player number	0.03	0.01	4.58	0.112
HID	Touches	0.42	0.21	37.83	<.001	0.000	0.000	0.64	0.65	−729.48
	Player number	0.48	0.24	42.80	<.001
	Touches x Player number	1.56	0.39	69.83	<.001
	Touches	1.13	0.57	16.89	<.001
SpD	Touches	1.81	0.91	112.18	<.001	0.000	0.000	0.64	0.65	−606.31
	Player number	3.20	1.50	197.66	<.001
	Touches x Player number	0.02	0.02	0.72	0.028
Sprints	Touches	4.31	2.16	96.39	<.001	0.04	0.06	0.71	0.73	−252.16
	Player number	13.75	6.87	307.16	<.001
	Touches x Player number	0.39	0.09	4.35	0.002
Peak speed	Touches	0.45	0.22	10.80	<.001	0.04	0.06	0.13	0.18	−275.93
	Player number	0.59	0.29	14.16	<.001
	Touches x Player number	0.13	0.03	1.59	0.177
ACC	Touches	1.13	0.57	16.89	<0.001	0.04	0.06	0.26	0.29	−121.45
	Player number	2.27	1.13	33.80	<.001
	Touches x Player number	0.11	0.03	0.80	0.526
DEC	Touches	0.43	0.22	4.21	0.016	0.000	0.000	0.24	0.25	100.53
	Player number	4.43	2.22	43.03	<.001
	Touches x Player number	0.15	0.04	0.74	0.022
Ball touches	Touches	0.97	0.48	38.59	<.001	0.000	0.000	0.60	0.62	−460.89
	Player number	4.49	2.24	179.35	<.001
	Touches x Player number	0.11	0.03	2.18	0.041
Ball releases	Touches	1.84	0.92	45.28	<.001	0.03	0.05	0.47	0.50	−287.03
	Player number	2.12	1.06	52.31	<.001
	Touches x Player number	0.96	0.24	11.85	<.001
Ball possession	Touches	1.55	0.78	55.46	<.001	0.000	0.000	0.34	0.36	−423.58
	Player number	1.15	0.58	41.10	<.001
	Touches x Player number	0.06	0.02	1.05	0.380
One-touch play	Touches	0.67	0.33	20.30	<.001	0.000	0.000	0.39	0.41	−370.06
	Player number	2.32	1.16	70.62	<.001
	Touches x Player number	0.57	0.14	8.64	0.498

Significant main effects of player numbers were observed for all variables (p < .001), while touch restriction significantly affected most outcomes (p < .05). Significant player numbers x touch restriction interactions were identified for HID, Sprints, SpD, DEC, ball releases, and ball touches (p < .05).

For physical variables, Sprints exceeded the SESOI in larger formats across all touch restriction conditions for the 3v3-7v7 comparison, as well as for 3v3-5v5 under 1 T and FP, and for 5v5-7v7 under 2 T. SpD decreased from 3v3 to 7v7 (−26% to −27%), while DEC increased meaningfully in 3v3-7v7 under 2 T and FP (2T: 42%, 90% CI: 23–63%; FP: 49%, 90% CI: 26–75%). HID showed mostly trivial effects, with meaningful increases observed only in isolated contrasts (e.g., HID 2 T 5v5-7v7: 31%) (Figure 4).

Figure 4.

Interaction effects of player number and touch restriction conditions on physical variables. Data are presented as percent change (Δ %) derived from exponentiated regression coefficients with 90% confidence intervals. Dashed lines indicate the smallest effect size of interest. TD = Total distance; HID = High-intensity distance; SpD = Sprint distance; ACC = Acceleration; DEC = Deceleration.

Significant interactions exceeding the SESOI for technical variables are illustrated in Figure 5. Post-hoc tests indicated that the variable ball touches increased from 3v3 to 5v5 across all touch restriction conditions (1T: 30%, 90% CI: 22–39%; 2T: 31%, 90% CI: 21–41%; FP: 39%, 90% CI: 32–46%). Ball releases increased primarily under 2 T (3v3-5v5: 35%, 90% CI: 22–49%; 3v3-7v7: 37%, 90% CI: 25–50%).

Figure 5.

Interaction effects of player number and touch restriction conditions on technical variables. Data are presented as percent change (Δ %) derived from exponentiated regression coefficients with 90% confidence intervals. Dashed lines indicate the smallest effect size of interest.

No significant interaction effects were found for 1 T plays, ball possessions, TD, peak speed, and ACC. Simple post-hoc tests from the main effects revealed that ACC increased from 3v3 to 7v7 (+30%, 90% CI: 20–41%), whereas all other technical and physical variables showed trivial differences for the main effects of touch restrictions and player numbers.

Discussion

The present study examined how SSGs with varying player numbers (3v3, 5v5, and 7v7) touch restriction sequences (1 T, 2 T, and FP) are associated with physical and technical demands in youth soccer. Overall, reducing player numbers were associated with increased DEC under 2 T and FP conditions, whereas the number of sprints and SpD were tended to be higher in formats with more players involved, particularly when comparing 7v7 to 3v3 across all touch restriction conditions. HID was also higher in 5v5 compared with 7v7 under 2 T conditions. These findings partially support our first hypothesis, showing that smaller formats increase decelerative demands, especially when interacting with touch restrictions (i.e., 2 T), while high-speed locomotor outputs tend to increase with larger player numbers. Regarding technical outcomes, smaller formats combined with all touch restrictions were associated with more ball touches, and 2 T conditions increased ball releases; however, these effects were heterogeneous, particularly for ball touches, contrasting with the initial expectation that more restrictive touch conditions (e.g., 2 T) would increase technical involvement.

TD is among the most frequently examined physical outcomes in research investigating the effects of player numbers and touch restrictions within SSGs.^12,24,29 Interestingly, our study did not reveal a significant interaction effect, suggesting that the influence of touch restrictions and player numbers on TD may operate in an independent manner within the examined formats and age group. Moreover, main effect post-hoc comparisons did not exceed the SESOI, contrasting with some previous findings. For instance, Li et al.¹⁵ reported consistently higher locomotor outputs with moderate effects sizes for TD (d = 0.55–0.71) in 5v5 and 8v8 formats compared with 3v3, while Coutinho et al.²⁵ demonstrated that limiting ball touches reduced TD in youth players.

Conversely, the touch restriction x player number interaction indicated that HID was greater in the 5v5 format compared with the 7v7 under the 2 T restriction. The 2 T constraint has been associated with increased exploration of both depth and width on the field, which may have contributed to the observed increase in HID.²⁸ In SSGs, such as 5v5, the reduced team size may result in lower levels of collective organization while also increasing distributional variability.⁵⁴ This combination of factors can create more open space along both the horizontal and longitudinal axes, affording players greater opportunities to cover HID.

With regard to speed-oriented variables, SpD and sprint count increased meaningfully with higher player numbers. This effect was consistent across all touch conditions employed, with most comparisons between 7v7 and 3v3 exceeding the SESOI. This pattern is aligns with previous research showing greater high-speed running in larger formats, with elevated distances reported in high-velocity bands exceeding 19 km/h⁵⁵ and > 21 km/h,¹⁵ comparable to the threshold applied in the present study (> 19.8 km/h). It can be speculated that larger team sizes result in on-pitch structures that more closely resemble those observed in official match play. In this context, our findings suggest that the use of 1 T, 2 T, or FP conditions in larger formats may be appropriate to elicit elevated high-speed demands. Positional analyses suggest that increased player numbers are associated with higher degrees of structural regularity, such as greater inter-player distances and increased team centroid spacing, facilitating more coordinated team behavior compared to smaller-sided formats.¹³ Although tactical measurements were not conducted in the present study, it is plausible that this heightened structural stability affords offensive players greater opportunities to disrupt organized defensive patterns through high-speed runs, thereby contributing to the observed increases in SpD and sprint counts.⁵⁶

However, interaction effects revealed an inconsistent pattern in the combined influence of touch restrictions and player numbers on sprint count. In some cases, the 1 T and FP conditions were associated with higher sprint counts in 5v5 compared with 3v3, whereas the 2 T condition elicited higher sprint counts in 7v7 compared with both 3v3 and 5v5. In smaller formats such as 3v3, constant player involvement may limit opportunities for longer sprinting actions irrespective of the imposed touch condition. This interpretation is partly supported by the trend observed in the 2 T condition, where the confidence interval nearly exceeded the SESOI threshold, potentially reflecting the limitations of lower numbered SSGs when the objective is to increase sprinting actions.⁵⁷ In such congested formats, play may remain spatially concentrated, while opportunities for longitudinal exploration occur less frequently.¹²

Previous studies that isolated the effect of touch restrictions have reported no meaningful differences in sprint-related variables among youth soccer players of comparable age.^34,58 This may suggest that, contrary to analyses focusing on a single task constraint, the combined manipulation of player numbers and touch restrictions could influence sprint counts. Nevertheless, this tendency should be interpreted with caution and requires further validation, as the post-hoc comparison between 5v5 and 7v7 in the FP condition for sprint frequency also approached the SESOI threshold.

Interaction effects exceeding the SESOI were observed for DEC actions when comparing 3v3 and 7v7 under 2 T and FP conditions. Previous research in U14 and U16 players has similarly shown that reducing player numbers while maintaining a comparable area per player, and without imposing touch restrictions, increases DEC occurrences around the −2 m/s² threshold, which aligns with the present findings.¹⁸ The parallel increase in DEC actions observed under the 2 T condition is consistent with the results of Mahlangu et al.,³⁴ who reported no differences between 2 T and FP formats in low (−1 to −2 m/s²), moderate (−2 to −3 m/s²), or high (←3 m/s²) DEC counts. This similarity suggests that the level of restriction imposed by a 2 T rule may not substantially alter players’ DEC demands compared with FP conditions. Overall, the results indicate that reductions in player numbers, combined with either FP or 2 T constraints, are associated with increased eccentric load as reflected by DEC frequency.

Regarding technical actions, interaction effects indicated meaningful differences in ball releases when comparing 3v3 with 5v5 and 7v7 under the 2 T condition, with effects observed in the 3v3 format. Reducing player numbers is well established as an effective task constraint to increase technical involvement.^59,60 In 3v3, players are less constrained by fixed positional roles, allowing more fluid participation across offensive phases of play (e.g., build-up, progression, and finishing).⁶⁰ This interpretation is supported by findings from Silva et al.,²³ who reported more aggressive offensive tactical principles, such as penetration and depth mobility, in 3v3, whereas larger formats (e.g., 6v6) promoted offensive unity, characterized by coordinated support from deeper defensive units. Our results suggest that, when combined with reduced player numbers, the application of touch restrictions, particularly 2 T, may further enhance ball releases, as previous research has shown increases in the frequency of release actions under such condition.⁵⁸ This finding is plausible, as the task constraint itself encourages players to release the ball quickly in order to avoid potential turnovers, thereby spending less time on the ball. In this context, the smaller format may further increase individual involvement and the frequency of ball interactions.^46,61

Our technical analysis also indicated that ball touches were higher when player numbers were lower, with meaningful effect sizes observed when comparing 3v3 with 5v5 across all touch restriction conditions. This suggests that, irrespective of whether a 1 T, 2 T, or FP condition was applied, smaller formats are associated with increased player-ball interactions. For comparison, a 4v4 ball-possession game reported more ball touches under 2 T restrictions than under 1 T conditions.⁵⁸ Although differences in game design and sample characteristics limit direct comparisons, the reduction in player numbers in the present study likely represents a central factor explaining the increase in ball touches across all touch conditions. Moreover, responses in technical variables such as ball touches and releases may have been influenced by the sequencing of touch restriction conditions within the same session. Residual behavioral adaptations from previously imposed constraints cannot be fully excluded and should be considered when interpreting the results.

For ball possessions, all effects remained within the SESOI range, suggesting limited meaningful variation across player numbers and touch restriction conditions. This may indicate that, under the employed touch restriction conditions, players’ ability to maintain possession and execute quick ball actions may be relatively robust to changes in game format, whether in 3v3, 5v5, or 7v7 settings. For the variable one-touch plays, we observed a favorable trend toward increased 1 T plays when fewer players were involved (i.e., 3v3 compared with 5v5) under the FP condition; however, the lower bound of the confidence interval crossed the SESOI threshold, indicating that the effect cannot be interpreted as meaningful. Interestingly, higher values were not observed in the 1 T condition across formats. This may be attributed to players’ tendency to perform 1 T actions even during 2 T or FP conditions across different game formats. This assumption is supported by previous evidence showing that, within a 2 T SSG, players still frequently executed passes with their first touch.³³ In this context, 1 T play may function as a technical strategy to maintain faster ball circulation (e.g., by avoiding duels) and to facilitate progressive offensive combinations aimed at reaching the opponent's goal.^26,30 However, the present analysis did not include a qualitative assessment of 1 T actions, which may be particularly relevant given the technical difficulty of these actions and the level of expertise of the players in our sample.²⁴

Limitations and future research

While this study contributes to the limited scientific literature examining the influence of touch restrictions and player numbers during SSGs in youth soccer, several limitations should be acknowledged. Firstly, we did not assess additional physical metrics (e.g., metabolic power) and zones (e.g., ACC and DEC). Future research should therefore incorporate a broader range of physical parameters and bands. Additionally, tactical metrics (e.g., collective behaviors) and physiological measures (e.g., heart rate responses) should be examined to provide a more comprehensive understanding of the effects of player numbers and touch restrictions. In addition, the number of technical actions was limited by the IMU system, and incorporating video-based technical analysis could complement this approach. Qualitative assessment of technical actions may also strengthen understanding, which might be of particular interest in youth soccer.

Secondly, while this study is the first to examine the interaction between player numbers and ball touches in youth soccer, the sample consisted of players from a single youth academy and team. This relatively homogeneous sample may have limited between-player variability, as reflected in the low ICC, indicating that most variability occurred within players across sessions rather than between players. It further suggests that residual factors, such as day-to-day fluctuations or situational influences, contributed to the observed outcomes. Therefore, future research should explore additional contextual variables (e.g., physiological, tactical) and examine whether the observed trends hold across different academies and age groups to strengthen external validity.

Thirdly, with regard to the study design, a cyclic rotation procedure was implemented to systematically distribute playing roles and enhance exposure balance across sessions. However, practical constraints inherent to the SSG formats limited the feasibility of a fully counterbalanced design. In particular, unequal team sizes in certain formats (notably 5v5 and 7v7) resulted in discrepancies in the number of individual exposures across players and sessions. Consequently, some variability in exposure dosage and sequencing may have occurred, introducing potential residual order effects that could not be fully controlled. Future research should address this limitation by employing fully balanced or counterbalanced allocation strategies where the format structure allows.

In addition, although touch restriction conditions were counterbalanced across sessions to reduce systematic bias, multiple experimental conditions were still administered within the same session. This may have introduced carry-over effects, as players could have retained tactical or behavioral adaptations from preceding conditions. While passive recovery intervals were included between conditions, these may not have been sufficient to fully eliminate short-term adaptation effects. To improve internal validity and allow for a more isolated examination of task constraints in SSG research, future studies should consider incorporating longer wash-out periods between conditions, particularly when multiple constraints are manipulated.

Lastly, the findings are specific to male youth soccer players; therefore, caution is warranted when generalizing to female players or different competitive levels. Future research should further examine the interaction between player numbers and touch restrictions across diverse populations and competitive contexts to enhance the external validity of these findings.

Conclusion

This study aimed to investigate how manipulating player numbers and touch restrictions alters the physical and technical demands of SSGs in youth soccer. Based on the findings, both task constraints appear to interact in shaping the distribution of physical load. Peak speed and SpD were more likely to be attained in formats with more players involved across all touch restriction conditions. Reductions in player number differences produced additional interaction effects, with FP being the only condition associated with elevated sprint counts in higher player number formats, while other touch restriction conditions showed inconsistent patterns. Under 2 T and FP conditions, meaningful differences were observed between 3v3 and 7v7 formats, whereas other interactions for physical variables (e.g., TD, ACC) were not meaningful.

On the technical side, formats with less players involved (e.g., 3v3 vs. 5v5 or 7v7) led to more ball touches (all touch restriction conditions) and releases, with the latter effect evident only under the 2 T condition. This reinforces the idea that technical involvement can be increased by manipulating player numbers and applying touch restriction conditions, particularly the 2 T rule. In contrast, ball possessions and one-touch actions showed no clear meaningful differences across conditions.

Overall, these findings suggest that the combined manipulation of player numbers and touch restrictions can be used to shape the physical and technical demands of SSGs in youth soccer. Practitioners may consider these constraints as tools to modulate physical output and technical engagement; however, the observed effects should be interpreted within the specific training context (e.g., task constraints and player population). Furthermore, the results should be considered in light of the study design employed (e.g., sequencing of touch restriction conditions) when extrapolating them to applied settings.

Footnotes

ORCID iDs

Johannes Jäger

Michael C. Rumpf

Benedikt Meixner

Ethical considerations

The study was approved by the Ethics Committee of Exercise Science and Training at the University of Würzburg (EV2026/2-2001), with additional approval obtained from the club. All procedures were conducted in accordance with the 1975 Declaration of Helsinki and its subsequent amendments.

Consent to participate

Written informed consent to participate was obtained from all participants and their legal guardians.

Consent for publication

Written informed consent for publication was obtained from all participants and their legal guardians.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability

The data of this study are available from the corresponding author upon reasonable request.

References

Hill-Haas

Dawson

Impellizzeri

, et al. Physiology of small-sided games training in football: a systematic review. Sports Med 2011; 41: 199–220.

Renshaw

Chow

Davids

, et al.

A constraints-led perspective to understanding skill acquisition and game play: a basis for integration of motor learning theory and physical education praxis?

Phys Educ Sport Peda 2010; 15: 117–137.

Clemente

Sarmento

. The effects of small-sided soccer games on technical actions and skills: a systematic review. Hum Mov 2020; 21: 100–119.

Clemente

Afonso

Castillo

, et al. The effects of small-sided soccer games on tactical behavior and collective dynamics: a systematic review. Chaos Solition Fract 2020; 134: 1–16.

Pinder

Davids

Renshaw

, et al. Representative learning design and functionality of research and practice in sport. J Sport Exercise Psy 2011; 33: 146–155.

Davids

Araújo

Correia

, et al. How small-sided and conditioned games enhance acquisition of movement and decision-making skills. Exerc Sport Sci Rev 2013; 41: 154–161.

Renshaw

Chow

. A constraint-led approach to sport and physical education pedagogy. Phys Educ Sport Peda 2018; 24: 1–14.

Sarmento

Clemente

Harper

, et al. Small sided games in soccer – a systematic review. Int J Perform Anal Sport 2018; 18: 693–749.

Borges

Praça

Figueiredo

, et al. Promoting tactical-technical actions during small-sided soccer games: a narrative review on constraints’ manipulation within ecological teaching models. Retos 2022; 45: 566–575.

10.

Praça

Chagas

Bredt

, et al. Small-sided soccer games with larger relative areas result in higher physical and physiological responses: a systematic and meta-analytical review. J Hum Kinet 2022; 81: 163–176.

11.

Bujalance-Moreno

Latorre-Román

García-Pinillos

. A systematic review on small-sided games in football players: acute and chronic adaptations. J Sports Sci 2019; 37: 921–949.

12.

Clemente

Martinho

Silva

, et al. Physiological, physical and technical demands during sided soccer game formats: a review. Int J Sports Med 2025; 46: 710–741.

13.

Aguiar

Gonçalves

Botelho

, et al. Footballers’ movement behaviour during 2-, 3-, 4- and 5-a-side small-sided games. J Sports Sci 2015; 33: 1259–1266.

14.

Hill-Haas

Dawson

Coutts

, et al. Physiological responses and time–motion characteristics of various small-sided soccer games in youth players. J Sports Sci 2009; 27: 1–8.

15.

Mao

Krustrup

, et al. Internal and external load during 8 v 8, 5 v 5 and 3 v 3 in Chinese elite youth male football players. Biol Sport 2022; 39: 1065–1071.

16.

Aguiar

MVD

Botelho

GMA

Gonçalves

BSV

, et al. Physiological responses and activity profiles of football small-sided games. J Strength Cond Res 2013; 27: 1287–1294.

17.

Brandes

Heitmann

Müller

. Physical responses of different small-sided game formats in elite youth soccer players. J Strength Cond Res 2011; 26: 1353–1360.

18.

López-Fernández

Sánchez-Sánchez

García-Unanue

, et al. Physical and physiological responses of U-14, U-16, and U-18 soccer players on different small-sided games. Sports 2020; 8: 1–16.

19.

Giménez

Del-Coso

Leicht

, et al. Comparison of the movement patterns between small- and large-sided game training and competition in professional soccer players. J Sports Med Phys Fitness 2018; 58: 1383–1389.

20.

Clemente

Ramirez-Campillo

Sarmento

, et al. Effects of small-sided game interventions on the technical execution and tactical behaviors of young and youth team sports players: a systematic review and meta-analysis. Front Psychol 2021; 12: 1–13.

21.

Clemente

Chen

Bezerra

, et al. Between-format differences and variability of technical actions during small-sided soccer games played by young players. Hum Mov 2019; 19: 114–120.

22.

Wang

Xue

. Do different playing formats and aerobic capacity influence variances in psychophysiological demands and technical performance in small-sided games? A study among under-17 male soccer players. J Sports Sci Med 2024; 23: 619–627.

23.

Silva

Garganta

Santos

, et al. Comparing tactical behaviour of soccer players in 3 vs. 3 and 6 vs. 6 small-sided games. J Hum Kinet 2014; 41: 191–202.

24.

Rumpf

Jäger

Altmann

, et al. Touch restriction during small-sided games in soccer—effects on physiological, physical and technical and tactical performance. Front Sports Act Living 2025; 7: 1–5.

25.

Coutinho

Gonçalves

Santos

, et al. Exploring how limiting the number of ball touches during small-sided games affects youth football players’ performance across different age groups. Int J Sports Sci Coach 2022; 17: 545–557.

26.

Lemoine

Jullien

Ahmaidi

. Technical and tactical analysis of one-touch playing in soccer - study of the production of information. Int J Perform Anal Sport 2005; 5: 83–103.

27.

Carling

. Influence of opposition team formation on physical and skill-related performance in a professional soccer team. Eur J Sport Sci 2011; 11: 155–164.

28.

Brito E Sousa

Bredt

SDGT

Greco

, et al. Influence of limiting the number of ball touches on players’ tactical behaviour and network properties during football small-sided games. Int J Perform Anal Sport 2019; 19: 999–1010.

29.

Lauria

Santo

LCE

Bicalho

CCF

, et al. Influence the number of touches on physical and physiological responses during small-sided soccer games. J Phys Educ 2024; 35: 1–13.

30.

Dellal

Lago-Penas

Wong

, et al. Effect of the number of ball contacts within bouts of 4 vs. 4 small-sided soccer games. Int J Sports Physiol Perform 2011; 6: 322–333.

31.

Giménez

Liu

Lipińska

, et al. Physical responses of professional soccer players during 4 vs. 4 small-sided games with mini-goals according to rule changes. Biol Sport 2017; 35: 75–81.

32.

Dellal

Karim

Lee

, et al. Influence of technical instructions on the physiological and physical demands of small-sided soccer games. Eur J Sport Sci 2011; 11: 341–346.

33.

Rosten

Aune

Dalen

. The impact of touch restrictions in small-sided games on soccer Players’ passing, receiving, and ball-touch temporal intervals. J Hum Kinet 2025; 98: 107–116.

34.

Mahlangu

Kubayi

Toriola

, et al. Effects of the number of ball touches among elite youth players during small-sided soccer games. J Sports Med Phys Fitness 2024; 65: 225–229.

35.

Almeida

Ferreira

Volossovitch

. Manipulating task constraints in small-sided soccer games: performance analysis and practical implications. Open Sports Sci J 2012; 5: 174–180.

36.

Nunes

Santos

Coutinho

, et al. Coaches’ perspectives of the use of small-sided games in the professional soccer training environment. J Kinesiol Exerc Sci 2024; 34: 27–39.

37.

Clemente

Afonso

Silva

, et al. Contemporary practices of Portuguese and Brazilian soccer coaches in designing and applying small-sided games. Biol Sport 2024; 41: 185–199.

38.

Ananth

Kleinbaum

. Regression models for ordinal responses: a review of methods and applications. Int J Epidemiol 1997; 26: 1323–1333.

39.

Lakens

. Sample size justification. Collabra Psychol 2022; 8: 1–28.

40.

Cohen

. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York: Routledge, 1988.

41.

McKay

AKA

Stellingwerff

Smith

, et al. Defining training and performance caliber: a participant classification framework. Int J Sports Physiol Perform 2022; 17: 317–331.

42.

Castellano

Casamichana

Dellal

. Influence of game format and number of players on heart rate responses and physical demands in small-sided soccer games. J Strength Cond Res 2013; 27: 1295–1303.

43.

Sandmæl

Dalen

. Comparison of GPS and IMU systems for total distance, velocity, acceleration and deceleration measurements during small-sided games in soccer. Cogent Soc Sci 2023; 9: 1–11.

44.

Waldron

Harding

Barrett

, et al. A new foot-mounted inertial measurement system in soccer: reliability and comparison to global positioning systems for velocity measurements during team sport actions. J Hum Kinet 2021; 77: 37–50.

45.

Sandmæl

van den Tillaar

Dalen

. Validity and reliability of Polar Team Pro and Playermaker for estimating running distance and speed in indoor and outdoor conditions. Sensors 2023; 23: 1–12.

46.

King

Barrett

McLellan

, et al.

Does team size affect Scottish male academy soccer player technical, locomotor and psychosocial outcomes during age and maturity bio-banded small-sided games?

J Sports Sci 2025. Epub ahead of print 4 February 2025. DOI: 10.1080/02640414.2025.2456408.

47.

Salter

Forsdyke

Arenas

, et al. Differences in physical and technical performance characteristics between 11 v 11 chronological and bio-banded soccer match-play. J Sci Med Sport 2025; 29: 296–307.

48.

Marris

Barrett

Abt

, et al. Quantifying technical actions in professional soccer using foot-mounted inertial measurement units. Sci Med Football 2022; 6: 203–214.

49.

Lewis

Towlson

Roversi

, et al. Quantifying volume and high-speed technical actions of professional soccer players using foot-mounted inertial measurement units. PLoS One 2022; 17: 1–13.

50.

De Melo

Daldegan-Bueno

Menezes Oliveira

, et al. Beyond ANOVA and MANOVA for repeated measures: advantages of generalized estimated equations and generalized linear mixed models and its use in neuroscience research. Eur J Neurosci 2022; 56: 6089–6098.

51.

Nakagawa

Schielzeth

. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods Ecol Evol 2013; 4: 133–142.

52.

Lakens

Scheel

Isager

. Equivalence testing for psychological research: a tutorial. Adv Methods Pract Psychol Sci 2018; 1: 259–269.

53.

Lakens

. Equivalence tests. Soc Psychol Personal Sci 2017; 8: 355–362.

54.

Brito

Roriz

Garganta

. Positioning and displacement patterns of young players during 5v5, 7v7, 9v9, and 11v11 soccer matches. J Hum Sport Exerc 2019; 15: 904–917.

55.

Modena

Schena

. External training load and performance recovery after small-sided games in soccer: insights for return-to-play management. AIMS Public Health 2024; 11: 315–329.

56.

Vilar

Araújo

Davids

, et al. The role of ecological dynamics in analysing performance in team sports. Sports Med 2012; 42: 1–10.

57.

Clemente

. The threats of small-sided soccer games: a discussion about their differences with the match external load demands and their variability levels. Strength Cond J 2020; 42: 100–105.

58.

Kubayi

Barrett

. The effect of ball touches on the physical and technical outputs of professional female soccer players in ball possession-based games. Int J Sports Sci Coach 2025. Epub ahead of print 30 December 2025. DOI: 10.1177/17479541251407844.

59.

Bergmann

Meier

Braksiek

. Involvement and performance of U9 soccer players in 7v7 and 5v5 matches during competition. Ger J Exerc Sport Res 2022; 52: 125–134.

60.

Clemente

Wong

Martins

FML

, et al. Acute effects of the number of players and scoring method on physiological, physical, and technical performance in small-sided soccer games. Res Sports Med 2014; 22: 380–397.

61.

Alcântara

Teixeira

, et al. Manipulation of number of players and bouts duration in small-sided games in youth soccer players. Sport Sci Health 2021; 17: 597–605.