The soccer season: performance variations and evolutionary trends

Rationale for the review

Player’s physical performance is one of the relevant performance domains and so, understanding the dynamic nature of adaptations throughout the season is of relevance for the population of soccer players. However, despite that there are important reviews concerning physiological characteristics of soccer players (Shephard, 1999; Stolen et al., 2005; Svensson & Drust, 2005), soccer biomechanics (Lees et al., 2010; Lees & Nolan, 1998), determinants of players’ performance (Bangsbo, Iaia & Krustrup, 2007; Bangsbo, Mohr & Krustrup, 2006; Reilly, Drust & Clarke, 2008), specific training-induced effects (Hill-Haas et al., 2011; Hoff, 2005; Hoff & Helgerud, 2004; Iaia, Rampinini & Bangsbo, 2009; Silva, Nassis & Rebelo, 2015), and development of soccer fatigue and kinetics of recovery (Hader et al., 2019; Mohr, Krustrup & Bangsbo, 2005; Nedelec et al., 2012; Nedelec et al., 2013; Reilly, Drust & Clarke, 2008; Reilly & Ekblom, 2005; Silva et al., 2018), an understanding of seasonal adaptations and evolutionary trends on players’ physical fitness is still required.

Intended audience and organization

Understanding the different variables that affect the dynamic nature of adaptations during the season may allow coaches, medical departments, and researchers to improve training periodization and monitoring. We target this review for students (e.g., exercise physiology and strength and conditioning), researchers, and all practitioners (coaches and medical department related staff) to whom the knowledge about the physiological and functional characteristics of the players is a matter of undeniable interest and understanding the different variables that affect the dynamic nature of adaptations occurring within the training season may allow informed decisions on training periodization and monitoring.

In this review we examined adaptations in body composition (body mass, body fat and lean body mass) and different neuromuscular qualities of professional soccer players (force production, jump, sprint and change of direction abilities). Subsequently, we analyzed seasonal alterations in endurance-related physiological and performance parameters as well as in competition measures of professional soccer players. Moreover, we will reference how different training stimulus and situational variables (e.g., competition exposure) affect the physiological and performance parameters of highly trained players. Further, we will explore aspects relevant for training monitoring of professional players.

Literature search strategy

For the search for relevant scientific literature, a review was performed using the PubMed and SportDiscus databases multiple times until June 2022. Additionally, Google Scholar and bibliographic searches of relevant articles were also completed. The description of seasonal trends comprises papers from January 2000 to April 2022 (Tables 1 and 2). The search strategy included the following search terms and Boolean operators using the term “soccer” AND “seasonal alterations”, OR “performance analysis”, OR “competition”, OR “physiology”, OR “body composition”, OR “strength training”, OR “neuromuscular performance”, OR “fatigue”, OR “field tests”, OR “intermittent endurance”, OR “muscular power”, OR “jump ability”, OR “sprint ability”, OR “agility”, OR “change of direction”, OR “training period”, OR “detraining”, OR “off season”, OR “in season”, OR “preseason” OR “competition period”.

Analysis and interpretation of the results

Studies were included if they: (i) investigated adults (>19 years) soccer players described has professional or elite player (ii) measured at least two season time points; specifically, the preparation period (PPS), beginning of the competitive period (BCP), middle (MCP) or end of competition period (ECP).

The mean and standard deviation for each measurement was extracted. In the case the necessary statistics were represented in figures and graphs their value was extrapolated using a specific software for the purpose (webPlotDigitizer; https://automeris.io/WebPlotDigitizer/). To evaluate the magnitude of the effects, percent change was calculated for each dependent variable for each study using the procedures defined elsewhere (Silva et al., 2018). Using the procedures defined in Schmitz et al. (2018) we compute a global mean (by time-point and variation between moments) based on the reported means of the individual studie for each outcome. We apply this procedure assuming that the players from each research within the same group belong to the same population and that their test results were extract from the same normal distribution (Schmitz et al., 2018). Each global mean was computed as weighted mean of the individual reported mean, with weights built by the number of subjects per investigation (Schmitz et al., 2018). Effect size (ES) were computed to present standardized values on the outcome variables (Cohen, 1998). The different ES within individual studies were calculated with Cohen’s d, by dividing the raw ES (difference in means) by the pooled standard deviations, as proposed (Cohen, 1998). To account for possible overestimation of the true population ESs were corrected accounting for the magnitude of the sample size of each study (Lakens, 2013). Therefore, a correction factor was calculated as proposed by Hedges & Olkin (1985). Threshold values for g were defined as trivial (<0.2), small (0.2–0.6), moderate (0.6–1.2), large (1.2–2.0) and very large (>2.0) (Cohen, 1998).

Body composition

Studies investigating seasonal changes in anthropometric variables, such as body mass (BM, n = 507) (Aziz, Tan & Teh, 2005; Bunc, Hráský & Skalská, 2015; Casajus, 2001; Clemente et al., 2021; D’Ascenzi et al., 2013; Edwards, Macfadyen & Clark, 2003; Fessi et al., 2016; Kalapotharakos, Ziogas & Tokmakidis, 2011; Koundourakis et al., 2014; Lago-Peñas et al., 2013; Meckel et al., 2018; Metaxas et al., 2006; Michalczyk et al., 2008; Ostojic, 2003; Ostojic et al., 2009; Reinke et al., 2009; Silva et al., 2011; Suda et al., 2012), body fat (BF, n = 579) (Aziz, Tan & Teh, 2005; Bonuccelli et al., 2012; Bunc, Hráský & Skalská, 2015; Casajus, 2001; Clemente et al., 2021; D’Ascenzi et al., 2013; Devlin et al., 2017; Fessi et al., 2016; Iga et al., 2014; Kalapotharakos, Ziogas & Tokmakidis, 2011; Koundourakis et al., 2014; Lago-Peñas et al., 2013; Los Arcos et al., 2015; Meckel et al., 2018; Metaxas et al., 2006; Ostojic, 2003; Ostojic et al., 2009; Owen et al., 2018; Papadakis, Patras & Georgouli, 2015; Reinke et al., 2009; Silva et al., 2011; Suda et al., 2012) and lean body mass (LBM, n = 226) (Bonuccelli et al., 2012; Bunc, Hráský & Skalská, 2015; Casajus, 2001; D’Ascenzi et al., 2013; Devlin et al., 2017; Metaxas et al., 2006; Ostojic, 2003; Owen et al., 2018; Reinke et al., 2009; Suda et al., 2012) are presented in Tables 1 and 2 and Figs. 1–3.

The overall analysis of a reasonable number of investigations seems to suggest that players’ BM (Fig. 1) is stable during the season; trivial effects from PPS to BCP (Δ = −0.79%, ES = −0.07), MCP (Δ = −0.85%, ES = −0.04) and ECP (Δ = −1.33%, ES = −0.12) are examined by average.

Generally, both the absolute and relative BF decreases during the season (Figs. 1 and 2). From the observed studies, we might conclude that alterations of small magnitude are examined from PPS to BCP (Δ = −9.6%, ES = −0.54), MCP (Δ = −8.2%, ES = −0.57) and ECP (Δ = −8.7%, ES = −0.39) in absolute BF. In this line of study, relative BF may decrease by a small magnitude in BCP (Δ = 8.9%, ES = 0.45), MCP (Δ = 9.9%, ES = 0.43) and ECP (Δ = 12%, ES = 0.53). Interestingly, at BCP 88% (16 in 18), at 94% (MCP 17 in 18) and at ECP 100% (12 in 12) of the ES reported are negative and so pointing on a decrease in absolute BF. Moreover, there are reports of decrements by moderate magnitude at BCP (Clemente et al., 2021; D’Ascenzi et al., 2013; Devlin et al., 2017; Meckel et al., 2018; Ostojic, 2003), MCP (D’Ascenzi et al., 2013; Devlin et al., 2017; Kalapotharakos, Ziogas & Tokmakidis, 2011; Meckel et al., 2018) and ECP (D’Ascenzi et al., 2013; Koundourakis et al., 2014; Ostojic, 2003; Papadakis, Patras & Georgouli, 2015). Although on average trivial changes in BF (absolute and relative) may occur during in-season (BCP vs MCP and MCP vs ECP, Δ = −2.4% and −1.8%, ES = −0.06 and −0.07, respectively), within the 16 studies that monitored in-season changes, both substantial decrements (Casajus, 2001; D’Ascenzi et al., 2013; Fessi et al., 2016; Kalapotharakos, Ziogas & Tokmakidis, 2011; Koundourakis et al., 2014; Ostojic, 2003; Papadakis, Patras & Georgouli, 2015; Suda et al., 2012) and increments (Devlin et al., 2017; Papadakis, Patras & Georgouli, 2015; Suda et al., 2012) are reported.

In this line of evidence towards a more positive body composition profile during season, the overall analysis of the studies suggest that players may substantially increase LBM during in-season. Our analyses reveal increases of small magnitude at BCP (Δ = 1.3%, ES = 0.23), MCP (Δ = 1.8%, ES = 0.33) and ECP (Δ = 3.1%, ES = 0.37) concerning PPS. Importantly, there are no reports of substantial decreases in LBM within the in-season assessments (BCP vs MCP and MCP vs ECP). On average increments of trivial magnitude are observed from BCP to MCP (Δ = 0.67%, ES = 0.15) and MCP to ECP (Δ = 1.2%, ES = 0.12). Curiously, variations in body composition seems to not be associated with the players’ participation time (combined training and match exposure time) and did not differ across seasons (Carling & Orhant, 2010) and are independent of players position (Milanese et al., 2015). In summary, the general picture (Fig. 1) may suggest that professional players may maintain their BM after starting the training period through decreases in BF and increases in LBM. Although off-season detraining seems to reverse these anthropometric adaptations, with alterations of small magnitude in BM (Δ = 1.9%, ES = 0.2), BF (Δ = 1.6%, ES = 0.5) and decrements of moderate magnitude in LBM (Δ = 5%, ES = 0.9) (Silva et al., 2016) they may return to “optimal” initial values for competition after the preparation period. Factors related to training (e.g., the type of strength training), competition fixtures (e.g., extent of the pre-season and/or in-season period, mid-season breaks) and diet (e.g., a Mediterranean diet) (Ostojic, 2003) may, among other factors, may explain part of the observed variability throughout the season (e.g., BF). Nevertheless, the computed values for the different BM, BF and LBM were derived from diverse assessment methods that have different measurements and precision errors associated (Mills, De Ste Croix & Cooper, 2017). Moreover, only a general picture has been provided and so, not capable to characterize the different body regions and associated seasonal variations.

Force production

Longitudinal studies examining changes in the force production capacity of specific muscle groups in professional players mainly relied on isokinetic assessments, despite the discrepancy in the angular velocities analyzed (Table 2) (Eniseler et al., 2012; Malliou et al., 2003; Silva et al., 2011). Seasonal alterations in force production capabilities of specific muscles groups at angular velocities of 60°/s⁻¹ (Eniseler et al., 2012; Malliou et al., 2003), 90°/s⁻¹ (Silva et al., 2011), 180°/s⁻¹ (Malliou et al., 2003), 300°/s⁻¹ and 500°/s⁻¹ (Eniseler et al., 2012) have been analyzed. Off-season induces alterations of small magnitude in knee extensors force production capacity at moderate (180°/s⁻¹) angular velocities (KE, Δ = 3.9%, ES = 0.37); no substantial alterations were observed at low angular velocities (60°/s⁻¹, Δ = −0.8%, ES = −0.07) (Malliou et al., 2003). During preseason, trivial effects are by average observed in KE at angular velocities of 60°/s (ranging from 227–272 and 222–229 N^.m, respectively at PPS and BCP) (Malliou et al., 2003), 90°/s⁻¹ (ranging from 239–242 and 241 N^.m, respectively at PPS and BCP) (Silva et al., 2011) and 180°/s (ranging from 150–155 and 157–158 N^.m, respectively at PPS and BCP) (Malliou et al., 2003). The same was observed for KF at 90°/s⁻¹ (ranging from 129–131 and 129–132 N^.m, respectively at PPS and BCP) (Silva et al., 2011). In this line of evidence, effects of trivial magnitude are by average observed from PPS to MCP for KE and KF, respectively when evaluated at 90°/s (KE, ranging from 241 N^.m and KF ranging from 133–135 N^.m at MCP) (Silva et al., 2011). Interestingly, when profiling adaptation in the force-velocity continuum perspective from PPS to ECP, small decrements at low (ranging from 272–273 and 251–253 N^.m, respectively) (Eniseler et al., 2012), changes of trivial magnitude at moderate (ranging from 239–242 and 244 N^.m, respectively) (Silva et al., 2011) and very large alterations at high angular velocities (ranging from 74–80 and 136–150 N^.m, respectively) (Eniseler et al., 2012) for KE strength have been reported. Interestingly, a consistent substantial increment is KF force production from PPS to ECP seems to take place independently of the angular velocity evaluated. Specifically, from small magnitudes at low (<60°/s⁻¹_, ranging from 148–150 and 159–178 N^.m, respectively), moderate at moderate angular velocities (90°/s⁻¹, ranging from 129–131 and 134–138 N^.m, respectively) and moderate and at high (300°/s⁻¹, ranging from 97–107 N^.m, respectively) and very large at very high angular velocities (500°/s⁻¹, ranging from 148–150 and 159–178 N^.m, respectively), respectively (Eniseler et al., 2012; Silva et al., 2011). This is particularly interesting, since is well documented that soccer-related injuries likely occur under rapid movement perturbations or actions requiring rapid force development and are more prevalent in hamstring muscles group (Hagglund, Walden & Ekstrand, 2005; Walden et al., 2015).

Jump ability

Seasonal changes in jump ability (15 studies, 390 players, Tables 1 and 2, Figs. 4–6) have frequently investigated the performance on single non-countermovement jump (Non-CMJ, SJ and SJWAS, Fig. 7) (Aziz, Tan & Teh, 2005; Casajus, 2001; Koundourakis et al., 2014; Lago-Peñas et al., 2013; Malliou et al., 2003) and single (CMJ_Based, CMJ and CMJWAS; Fig. 8) (Casajus, 2001; Clark et al., 2008; Eliakim et al., 2018; Fessi et al., 2016; Koundourakis et al., 2014; Lago-Peñas et al., 2013; Los Arcos et al., 2015; Malliou et al., 2003; Meckel et al., 2018; Ostojic et al., 2009; Papadakis, Patras & Georgouli, 2015; Requena et al., 2017; Silva et al., 2011), and repeated countermovement jumps (Casajus, 2001; Clark et al., 2008).

Sprint ability

Insights from training

Maximal oxygen consumption

Although with obvious limitations, e.g., just one study involves a longitudinal, inter-seasonal examination of soccer players (12 seasons, 1,545 players), it seems that among professional players, VO_2max is not improving over time and perhaps has the tendency to decrease (players tested from 2006–2012 showed 3.2% lower values than those tested from 2000–2006) (Tonnessen et al., 2013).

Seasonal alterations in VO_2max have been extensively analyzed (Table 2, Figs. 9 and 10, n = 393) (Aziz, Tan & Teh, 2005; Bunc, Hráský & Skalská, 2015; Casajus, 2001; Castagna et al., 2013; Clark et al., 2008; Edwards, Macfadyen & Clark, 2003; Eliakim et al., 2018; Haritodinis et al., 2004; Kalapotharakos, Ziogas & Tokmakidis, 2011; Koundourakis et al., 2014; Lago-Peñas et al., 2013; Manzi et al., 2013; Meckel et al., 2018; Metaxas et al., 2009; Michalczyk et al., 2008; Mohr, Krustrup & Bangsbo, 2002). Generally, during the pre-season, professional players appear to regain their oxygen capacity and maintain it throughout the season as off-season seems to induce a large impairment in this physiological parameter (Δ = 4.4%, ES = 1.4) (Silva et al., 2016). Studies with players from different backgrounds exposed a large magnitude of improvements in VO_2max from the PPS (ranging from 52.2–62.7 ml^.min^−1.kg⁻¹) to BCP (Δ = 7.3%, ES = 1.3, ranging from 54.8–66.5 ml^.min^−1.kg⁻¹). Additionally, improvements of moderate magnitude were by average observed in the MCP (Δ = 6.4%, ES = 1.0, ranging from 55.5–66.8 ml^.min^−1.kg⁻¹) and ECP (Δ = 4.2%, ES = 0.8 ranging from 52.7–64.1 ml^.min^−1.kg⁻¹) compared with the PPS. Moreover, increases in VO_2max (from PPS to BCP, MCP and ECP) seem to be independent of the position role (Metaxas et al., 2006). Interestingly, just one of the 14 ES did not confirm the substantial improvements at MCP and four on 13 at ECP compared to PPS assessments. Within the competitive phase they are observed by average trivial changes from BCP to MCP (Δ = 0.5%, ES = 0.1) and a small decrement from MCP to ECP (Δ = −2.3%, ES = −0.28).

Anaerobic threshold

Studies examining changes in physiological parameters at sub-maximal intensities are presented in Table 2 (13 studies, n = 249, Fig. 9) (Casajus, 2001; Castagna et al., 2011; Clark et al., 2008; Dunbar, 2002; Edwards, Macfadyen & Clark, 2003; Kalapotharakos, Ziogas & Tokmakidis, 2011; Los Arcos et al., 2015; Manzi et al., 2013; Meckel et al., 2018; Mohr, Krustrup & Bangsbo, 2002; Papadakis, Patras & Georgouli, 2015; Zoppi et al., 2006).

From the large variety of parameters examined, some were shown to be sensitive in one but not in other studies that used players of similar standards. Nevertheless, enhancements in the ability to cope with sub-maximal internal and external loads regarding PPS performances were by average detected by different parameters, as follows:

1. the percentage of VO_2max (76%VO_2max) and percentage of maximal heart rate (87%HR_max) at a lactate concentration of 4 mmol⁻¹ at BCP (ES = 0.62 and 0.71, 78% and 89%) and MCP (ES = 0.89 and 0.91, 78% and 89%), respectively) (Kalapotharakos, Ziogas & Tokmakidis, 2011);

2. oxygen consumption at the LT (ES = 0.5 at ECP, ranging from 51.4–53.5 ml^.min^−1.kg⁻¹) (Edwards, Macfadyen & Clark, 2003) and VT (ES = 0.85 and 0.41 at BCP (ranging from 50.2-52.7 ml^.min^−1.kg⁻¹) and ECP (52.9 ml^.min^−1.kg⁻¹), respectively) (Casajus, 2001; Edwards, Macfadyen & Clark, 2003; Manzi et al., 2013);

3. heart rate measures at speeds of 14-km/h (ES = 2.7), 16-km/h (ES = 2.6), and 18-km/h (ES = 2.0) at MCP (Mohr, Krustrup & Bangsbo, 2002);

4. the speed at a fixed lactate concentration (Fig. 9) of: (a) 2 mmol⁻¹ (ES = 0.67, 0.66 and 0.68, at BCP (ranging from 11.4–14.5-km/h), MCP (ranging from 10.5–14.8-km/h), and ECP (ranging from 10.8–13.9-km/h) regarding PPS (ranging from 9.5–14.3-km/h) respectively) (Castagna et al., 2011; Castagna et al., 2013; Dunbar, 2002; Kalapotharakos, Ziogas & Tokmakidis, 2011; Manzi et al., 2013; Papadakis, Patras & Georgouli, 2015); (b) 3 mmol⁻¹ (ES = 0.52, 0.20 and −0,27 at BCP (ranging from 12.7–15.4-km/h), MCP (15.7 km/h), and ECP (15.0-km/h) regarding PPS (ranging from 12.2–15.4-km/h), respectively) (Dunbar, 2002; Los Arcos et al., 2015); (c) 4 mmol⁻¹ (ES = 1.0, 1.41 and 1.27 at BCP (ranging from 13.6–14.9-km/h), MCP (ranging from 13.6–14.4-km/h) and ECP (ranging from 13.5–14.3-km/h), regarding PPS (ranging from 12.3–13.9-km/h), respectively) (Castagna et al., 2011; Castagna et al., 2013; Kalapotharakos, Ziogas & Tokmakidis, 2011; Manzi et al., 2013; Papadakis, Patras & Georgouli, 2015);

5. The speed at the LT (ES = 1.9 at BCP, ranging from 10.5–13.8-km/h, respectively) (Zoppi et al., 2006) and VT (ES = 0.57 and 1.1, at BCP (ranging from 11.6–12.2-km/h) and MCP (12.8-km/h), respectively) (Meckel et al., 2018).

Interesting, although a wide variety of submaximal parameters have been measured, substantial improvements between BCP and MCP are consistently reported within the analyzed studies in some form of physiological parameter (Casajus, 2001; Dunbar, 2002; Kalapotharakos, Ziogas & Tokmakidis, 2011; Meckel et al., 2018; Papadakis, Patras & Georgouli, 2015). Nevertheless, alterations of trivial magnitude have also been examined (Kalapotharakos, Ziogas & Tokmakidis, 2011; Papadakis, Patras & Georgouli, 2015). From the MCP to ECP distinct alterations have been observed with both reports of trivial (Papadakis, Patras & Georgouli, 2015), small improvements (Papadakis, Patras & Georgouli, 2015) and impairments (Dunbar, 2002).

Interestingly, although trivial alterations in VO_2max are by average observed from BCP to MCP, an improvement of small magnitude (ES = 0.29) is observed between these time points, which suggest that further improvement in sub-maximal exercise performance (e.g., LT), but not in VO_2max, are likely related to a faster restoration or improvement of central factors (i.e., VO_2max) than peripheral factors (i.e., muscle oxidative enzymes) (Impellizzeri et al., 2006). Furthermore, although adaptations in RE being dependent on multi-dimensional factors (e.g., mechanical, and neuromuscular skills) they may had occurred further in season, and so determinant for improving running performance (Foster & Lucia, 2007); RE can better discriminate soccer players of different standards with similar VO_2max values (Ziogas et al., 2011).

In summary, these physiological determinants of endurance performance, improve during the first part of the season (4–8 weeks) and generally remain stable throughout the season. Generally, improvements in VO_2max occurred after a relatively short period of time (e.g., pre-season training), while no significant further in-season increases are observed. Moreover, no increase was examined in the VO_2max when players already possessed values of approximately 61–62 ml/kg/min. In fact, the increases in VO_2max found in different standard of players during the in-season period (Caldwell & Peters, 2009; Jensen et al., 2009; Magal et al., 2009) occurred under this threshold and in players of a lower standard. Additionally, when professional players began the competitive season with values above this threshold (61–62 ml/kg/min), no improvements in the VO_2max throughout the season were by average reported (Clark et al., 2008; Edwards, Macfadyen & Clark, 2003). This may be related with soccer training-specific constrains and/or demands, such as, the limited time for fitness training due to the high density of in-season match commitments. Our analysis seems to corroborate the observations of others (Tonnessen et al., 2013), indicating that VO_2max values of approximately 62–64 ml/min/kg may fulfill the general demands for aerobic capacity in male professional soccer players; nevertheless, characteristics related to the specific demands of different positional roles should be considered, as reported values reflect team averages and large inter-individual variations can be observed.

Maximal aerobic speed

MAS (Figs. 9 and 11) (n = 143) (Boullosa et al., 2013; Bunc, Hráský & Skalská, 2015; Fessi et al., 2016; Kalapotharakos, Ziogas & Tokmakidis, 2011; Lago-Peñas et al., 2013; Requena et al., 2017) reflects the maximum aerobic capacity and combines VO_2max and RE into a single factor (Billat & Koralsztein, 1996). As such, MAS is a good indicator of aerobic performance (Billat & Koralsztein, 1996), and the determination of MAS gives a practical assessment of the aerobic demands during running performance (Kalapotharakos, Ziogas & Tokmakidis, 2011). Off-season break may induce a decrement of moderate magnitude in MAS (Δ = 4.6%, ES = 0.61) (Requena et al., 2017). Preseason training restores MAS (Δ = 5%, ES = 1.3, ranging from 18.1–19.7-km/h), with substantial improvements still evident at MCP (Δ = 4.3%, ES = 1.3, ranging from 17.4–19.6-km/h) and at ECP (Δ = 4.9%, ES = 1.05, ranging from 17.3–18.4-km/h) regarding the PPS values (ranging from 16.5–19.2-km/h). Although, by average no substantial improvements take place from BCP to MCP (Δ = −0.7%, ES = −0.11) and from MCP to ECP (Δ = 0.6%, ES = −0.09), there are contradictory observations between BCP and MCP, with both trivial (Δ = −0.4%, ES = −0.09) (Lago-Peñas et al., 2013), moderate impairments (Δ = −3.3%, ES = −0.82) (Fessi et al., 2016) and improvements of small magnitude (Δ = 1.7%, ES = 0.59) (Kalapotharakos, Ziogas & Tokmakidis, 2011) reported. Interestingly, Boullosa et al. (2013) did not observed changes in the MAS (18.1 to 18.2-km/h) in professional players after pre-season. The different findings are, at least in part, associated with the dissimilar baseline MAS that were reported and the applied protocols (Dupont, Akakpo & Berthoin, 2004; Fessi et al., 2016; Kalapotharakos, Ziogas & Tokmakidis, 2011; Wong et al., 2010). We would like to highlight that in this narrative review we discussed the velocity at VO_2max (vVO_2max) (Kalapotharakos, Ziogas & Tokmakidis, 2011), and final velocity reached (Vam-eval and Gacon test) as one parameter (Boullosa et al., 2013; Bunc, Hráský & Skalská, 2015; Fessi et al., 2016; Lago-Peñas et al., 2013; Requena et al., 2017). Although they are highly correlated, with the two terms being often used interchangeably, they refer to different physiological entities (Buchheit, 2010) with MAS maybe 10–15% greater than the vVO_2max (Berthon & Fellmann, 2002).

In summary, despite the scarcity of research monitoring these performance parameters, MAS increase after pre-season training and remain stable throughout the season. The magnitude of alterations (MAS) may be associated with the baseline training status of players at the time of intervention (Boullosa et al., 2013).

High-intensity intermittent exercise

A summary of studies examining changes in high intensity intermittent exercise (IE) tests is presented in Tables 1 and 2 and Figs. 9 and 12 (Boullosa et al., 2013; Bradley et al., 2010; Campos-Vazquez et al., 2016; Castagna et al., 2013; Iaia et al., 2009b; Krustrup et al., 2003; Krustrup et al., 2006; Manzi et al., 2013; Silva et al., 2011). Off-season seems to result in decrements of moderate and very large magnitude in IE performance (Δ = 27.8% and 10%, ES = 1.0 and 2.2 for YYIE2 and YYIR2, respectively). However, preseason phase by average induces large improvements IE (Δ_overall = 32.4%, ES = 1.8). Specifically, improvements of 56%, 60%, 18% and 5%, and effect sizes of 4.1, 2.4, 1.1 and 1.25 for YYIR2 (ranging from 742–780-m and 1,033–1,160-m), YYIE2 (ranging from 1,120–2,171-m and 2,250–2,411-m), YYIR1 (ranging from 1,760–2,475-m and 2,211–2,600-m) and 30–15 (20.1 to 21.1 km/h), respectively. These performance improvements are extended to MCP (Δ_overall = 18.9%, ES = 1.5). Precisely, increases of 43.9%, and 17.9%, with magnitudes of 2.4 and 0.7 for YYIR2 (ranging from 742–780-m) and YYIE2 (ranging from 742–780-m). Interestingly, the magnitude of alterations is lower from PPS to ECP (Δ_overall = 22.5%, ES = 1.0). Specifically, increments of 11.9%, 29.7% and 19.5% with magnitudes of 0.51, 0.96 and 1.56, for YYIR2 (873-m), YYIE2 (ranging from 1,640–2,381-m), YYIR1 (2,103-m), are examined. Within the season, the ability to perform IE is by average impaired to a small extent from BCP to MCP (Δ_overall = −2.4%, ES = −0.23 (Δ = −7.2% and 6.1%, ES = −0.47 and 0.24 for YYIR2 and YYIE2, respectively)) and from MCP to ECP (Δ_overall = −7%, ES = −0.3). We would like to highlight again, that within each team, a great inter-individual ability to perform repeated intense exercise can be observed throughout the season, with some players improving, others decreasing and/or maintaining their performance (Bangsbo, Iaia & Krustrup, 2008).

Interestingly, Boullosa et al. (2013) did not report substantial changes in YYIR1 from PPS to BCP. It should be observed that in this study, players started the season with a high YYIR1 performance (2,475-m), which may be related to the performance of an off-season program (5 weeks/21 sessions). This previous evidence, at least in part, highly indicate the benefits of performing a structured training program during the off-season (Silva et al., 2016). Moreover, it should be noted that despite no significant improvements in YYIR1, the authors reported important changes in certain indices of cardiac autonomic adaptations (e.g., short heart-rate recovery) after this period of intensified training.

Repeated sprint ability

Impellizzeri et al. (2008) observed that elite players improved different parameters in RSSA test performance throughout the season. Namely, the mean time of the sprints (RSSA_mean) improved to a moderate extent from PPS to BCP (Δ = 2.2%, ES = 1.14), MCP (Δ = 1.4%, ES = 0.74) and ECP (Δ = 1.6%, ES = 0.29). The fatigue index improved to a small magnitude from PPS to BCP (Δ = 20.4%, ES = 0.56), and in a moderate extent from PPS to MCP (Δ = 22.2%, ES = 0.62) and ECP (Δ = 25.9%, ES = 0.71). The lower fatigability during repeated sprints performed during MCP and ECP vs PPS as also been verified when monitoring U20 elite players using the Bangsbo sprint test (Jorge, Garrafoli & Cal Abad, 2020). Nevertheless, a small deterioration of the RSSA_mean occurred from the BCP to MCP (Δ = 0.84%, ES = 0.41) with trivial changes been observed from MCP to ECP and for the fatigue index within these specific in-season moments.

We intended to characterize the general ability of performing repeated intense exercise and with this purpose we combined results of different specific IE tests that are widely used in professional settings. We acknowledge the differences between protocols of each individual test and that they might evaluate slightly different physical capacities (Buchheit & Rabbani, 2014). As example, YYIR1 leads to a maximal activation of the aerobic system, whereas YYIR2 determines an individual’s ability to recover from repeated exercise with a high contribution from the anaerobic system (Bangsbo, Iaia & Krustrup, 2008). Nevertheless, their sensitivity to training is almost certainly similar (30-15 vs YYIR1) (Buchheit & Rabbani, 2014) and given the very large correlations between tests (YYIR1 vs YYIR2) practitioners have been advised to consider using only one of the Yo-Yo tests and a RSA test in a general soccer-specific field test protocol (Ingebrigtsen et al., 2012; Ingebrigtsen et al., 2013a).

Sub-maximal intermittent field exercise

It has been observed that soccer players %HR_max at the 6-min point of the YYIR1 decreased from the PPS to the middle of the pre-season, BCP and ECP (Krustrup et al., 2003). Rago et al. (2020) when applying the same protocol during the in-season period (four assessment moments from MCP to ECP) observed a continuous moderate improvement in heart rate measurements towards ECP. Moreover, others observed that even though professional players may show a decline in VO_2max from the preparation period to the end of the season, their heart rate responses during the sub-maximal version of the YYIE2 were not altered during five time-points of a soccer season (from 14 days pre-season to ECP) (Heisterberg et al., 2012).

Game-related physical parameters

Match analysis is a widely used instrument in professional soccer to study technical, tactical, and physical performances of players (Abt & Lovell, 2009). These instruments allow careful analysis of player match performance, dependent of a large number of factors (e.g., training status, field position, age) and allows for the investigation of seasonal changes in game-related physical performance (Helgerud et al., 2001; Impellizzeri et al., 2006; Morgans et al., 2014; Padron-Cabo et al., 2018; Rampinini et al., 2007b; Silva et al., 2013b) and study evolutionary trends over consecutive seasons (Akyildiz et al., 2022; Barnes et al., 2014; Bradley et al., 2016; Bush et al., 2014; Pons et al., 2021; Vigne et al., 2012).

Seasonal variations

Seasonal alterations in distance covered in different speed zones during the game is presented in Figs. 13–15 (Link & de Lorenzo, 2016; Morgans et al., 2014; Padron-Cabo et al., 2018; Rampinini et al., 2007b; Silva et al., 2013b). There are by average trivial changes in the TD (Fig. 14) from BCP to MCP (Δ = −1.09%, ES = 0.03, ranging from 9,150–10,513-m and 9,350–10,722-m, respectively) with a small increment between BCP and ECP (Δ = 1.63%, ES = 0.22, ranging from 9,600–10,921-m). A small increase in TD seems to occur from MCP to ECP (Δ = 1.5%, ES = 0.47). Interestingly, a clear variability exists, with both increments (Mohr, Krustrup & Bangsbo, 2003; Rampinini et al., 2007b; Silva et al., 2013b) and decrements (Link & de Lorenzo, 2016; Mohr, Krustrup & Bangsbo, 2003; Padron-Cabo et al., 2018) between these time-points. Importantly, within the season the variation ranged from −2.4% to 5.9% that are below for the reference value (10–15%) (Oliva-Lozano et al., 2021a) that can establish a practical significance considering the high match-to match variability (Gregson et al., 2010; Oliva-Lozano et al., 2021a).

The distance covered in HSR (~>14.4–15 km/h; Figs. 13 and 15) during the match has been proposed to be of great importance for performance in elite soccer because clearly distinguishes players of different standards (Mohr, Krustrup & Bangsbo, 2003; Saeterbakken et al., 2019). However, these observations of HSR proficiency being associated with player standards have not been unanimously confirmed (Di Salvo et al., 2012). The amount of HSR by average decreases with a trivial magnitude from BCP to MCP (Δ = 2.1%, ES = −0.14, ranging from 1,350–2,450-m and 1,270–2,544-m, respectively) and increase to a moderate extent to ECP (Δ = 22.5%, ES = 0.75, ranging from 1,900–2,738-m) and from MCP to ECP (Δ = 25.9%, ES = 0.92). Interestingly all the studies, although reporting different magnitudes (small to large), observed substantial alterations between these two time-points (Mohr, Krustrup & Bangsbo, 2003; Rampinini et al., 2007b; Silva et al., 2013b). Moreover, the amount of HSR performed in the last fifteen-minute period of each half, indicative of the ability to maintain performance during the game (Krustrup et al., 2005), was reported to be higher towards the ECP (Silva et al., 2013b). Additionally, in the ECP, a greater distance in HSR was covered in the peak and in the lowest fifteen-minute periods of the match than in the corresponding fifteen-minute periods at other season time points (Silva et al., 2013b). Furthermore, Silva et al. (2013b) observed that professional players were more engaged in high-intensity activities and had higher peak 5-min periods of HSR during the matches towards the last quarter of the season (ECP).

Very-high speed running (>19.8–21 km/h; VHSR; Fig. 15) is stable from the BCP to MCP (Δ = 3.8%, ES = 0.10, ranging from 465–916-m and 485–829-m, respectively) and ECP (Δ = 4.5%, ES = 0.16, ranging from 481–977-m). Small increments in VHSR may take place from MCP to ECP (Δ = 0.74%, ES = 0.29). Importantly, within the different season moments the variation ranged from −18% to 20% that are below for the reference value of 60–64% informing that a real change take place (Gregson et al., 2010; Oliva-Lozano et al., 2021a).

The sprint distance performed during the match (>24–30 km/h; Fig. 15) (Morgans et al., 2014; Padron-Cabo et al., 2018; Silva et al., 2013a) is by average stable from BCP to MCP (Δ = 4.5%, ES = 0.19, ranging from 98–201-m and 111–225-m, respectively) and increase with moderate and small magnitude from BCP to ECP (Δ = 11%, ES = 0.69, ranging from 192–234-m) and MCP to ECP (Δ = 6.1%, ES = 0.55), respectively. However, within the three studies analyzed, two (Morgans et al., 2014; Padron-Cabo et al., 2018) consistently reported trivial changes between these time-points. It should be again highlighted, that a large match-to-match variability in game-physical parameters of elite players may occur, suggesting that only large sample sizes may allow the clarification of systematic changes and that the “training stimulus” provided by the match is largely variable (Gregson et al., 2010; Oliva-Lozano et al., 2021a).

Insights from training

High intensity training (HIT) comprises different modes of high-intensity exercise, namely high-intensity aerobic training (HIA), speed endurance training (SE) and repeated sprint ability training (RSA) (Bishop, Girard & Mendez-Villanueva, 2011; Mohr et al., 2022; Spencer et al., 2005). Generally, the common factors between modes are the high degree of physiological stress and the sharing of some similar physiological and functional training-induced adaptations imposed by the acute and chronic effects of the high-intensity bouts. HIT is a useful training method, providing a high training stimulus (Bangsbo et al., 2009; Buchheit & Laursen, 2013a; Christensen et al., 2011; Iaia & Bangsbo, 2010; Iaia et al., 2009a) on both the cardiopulmonary (Buchheit & Laursen, 2013a) and neuromuscular levels (Buchheit & Laursen, 2013b), thereby promoting physiological and performance adaptations that allow players to more successfully cope with the match and training demands (Castagna et al., 2009; Girard, Mendez-Villanueva & Bishop, 2011; Gunnarsson et al., 2012; Helgerud et al., 2001; Iaia, Rampinini & Bangsbo, 2009; Impellizzeri et al., 2006; Ingebrigtsen et al., 2013b; Krustrup et al., 2005; Mohr, Krustrup & Bangsbo, 2003; Rampinini et al., 2007a).

Research

Research in soccer uncovers the complexity of interactions established between the different performance dimensions and the factors that are intrinsic to each player and team. However, the paucity of in-season data on specific anaerobic/neuromuscular qualities (e.g., anaerobic power, relative force, rate force development, maximal speed) and physiological and endurance-related parameters (e.g., RE, VO₂ kinetics, cardiac autonomic adaptations; short heart-rate recovery) that may be relevant in improving running capacity, should be investigated to allow for a better understanding of seasonal variations in physical fitness, more robustly, through the in-season phases. As example, overall systematic analyses of the data revealed better scores in multi-joint, power-based, dynamic efforts during in-season periods. In part, these observations may lead to the following proposals: (i) neuromuscular adaptations affecting SSC mechanisms (phase analysis) may occur throughout the in-season period; and (ii) a composite score of power-based efforts may be more relevant for tracking the training status of professional players than a single measure, per se. Future research should also aim to understand seasonal changes in force capabilities during various velocities conditions and during specific motor tasks (jumping and sprinting) (Morin, 2019); efforts are already being developed in this direction (Haugen, 2018; Jimenez-Reyes et al., 2022). Moreover, studies aiming in improve the understanding of acute and chronic neuromuscular and endurance adaptations of professional players triggered by different in-season concurrent training modes (e.g., two instead of one: build power and endurance at the same time) is necessary. Research examining the effect of match exposure throughout the season on the performance adaptation kinetics of professional players is warranted; match-playing time may influence adaptations of specific and non-specific endurance and neuromuscular parameters during the season (Hader et al., 2019; Morgans, Di Michele & Drust, 2017; Silva et al., 2011; Sporis et al., 2011). Furthermore, understand how the distinct internal and external load parameters (Level 1, 2 and 3 metrics) experienced by each individual player during the optimization of the distinct performance dimensions (e.g., tactical) impact players fitness status will be key for optimize the full spectrum of the physical potential of the players (mechanical and metabolic). Studies characterizing the periodization of training loads (overall) during the pre-season and in-season periods of professional players are necessary. Moreover, considering the off-season detraining effects, a “reorganization” of the periodization during the transition period is necessary (Silva et al., 2016). In fact, these findings could lead one to question what the usefulness of such a loss of individual (collective) performance potential during off-season? We recently, made a call to action to understand how the prescription of off-season individualized training programs may influence seasonal performance (Silva et al., 2016). We highlighted that this period should be viewed as a ‘window of opportunity’ for players to recover and to ‘rebuild’ for the following season (Silva et al., 2016). ‘Rebuild’ for a more efficient and consistent in-season performance.

The perceptible increase in HSR towards the end-of-season period can be influenced, at least in part, by an improvement of pacing strategies in some form by professional players. As such, the development and improvement of conscious and/or sub-conscious pacing strategies (Carling & Bloomfield, 2010; Edwards & Noakes, 2009; Mugglestone et al., 2012) that seems to take place during matches cannot be excluded; there are contradictions regarding the concept of team sport players pacing their effort throughout the game (Aughey, 2010). This fact seems consistent with the higher physical performance in games towards the end of the season (Mohr, Krustrup & Bangsbo, 2003; Rampinini et al., 2007b; Silva et al., 2013b) and in other football codes (Aughey, 2011), without improvements in the majority of physiological and functional parameters; evidence of increases in certain stress biomarkers have also been reported (Handziski et al., 2006; Heisterberg et al., 2012; Kraemer et al., 2004; Meyer & Meister, 2011; Reinke et al., 2009; Silva et al., 2014; Suda et al., 2012). However, the well know context of the final stage of the competitive season (e.g, definition of team rank and contract renewal) as obvious impact in players “motivation” to perform. In these specific periods there is no space to “error”, and most likely “Mind will prevail over Muscle” (Marcora & Staiano, 2010; Pessiglione et al., 2007). As so, caution is needed when estimating players “readiness” from overall match activity profile. Research on these factors is necessary.

A better understanding of roles and tactics of team organization and an improvement in decision-making during season matches should be taken in account as central variables that may impact performance throughout the season (Vigne et al., 2012). Interestingly, data on longitudinal changes in match activity throughout the season seem to suggest an increased match efficiency (ranging from 2.6–6%) during the in-season period (efficiency = percentage of the total distance performed in high-intensity categories) (Mohr, Krustrup & Bangsbo, 2003; Rampinini et al., 2007b; Silva et al., 2013b). Another interesting factor is that high-level soccer players seem to exhibit superior anticipation capacity accompanied by more effective search behaviors and elaborative thought processes (Casanova et al., 2013). Nevertheless, the state of research regarding improvements in perceptual-cognitive processes in highly trained players and the influence of pacing and match activity remains very scarce. Curiously, elite players with long-term careers, parallelly to a annual gradual decrease in match-related physical output (0.56–1.8% by year) improve technical–tactical skills with increasing age (Rey et al., 2022). ‘Integrated’ approachs that contextualizes physical demands in relation to key tactical activities for each position and collectively for the team are warranted; understanding the physical performance in relation to the tactical roles (Bradley & Ade, 2018). In fact, is not the match running performance alone that is important for achieving success, but rather its relation to technical/tactical skills (Hoppe et al., 2015). Finally, the causative factors of the observed long-term changes (evolutionary trends) are not known and can be related, among others, with processes of players selection (e.g., towards more “highly impulsive” players), improvements in facilities and equipment’s (e.g., grass conditions) and training-related processes (e.g., better physical conditioning, training monitorization and players nutrition and recovery support). Research into the previous components is necessary.

Training

Although one normally expects than within the season (from BCP to MCP or MCP to ECP) the consistent training of the physical, tactical, and technical dimensions of performance and as well the stimulus provided by competitive matches, could lead to a further optimization of players performance, more robustly concerning the start of competition phase (e.g., BCP to MCP). However, within these periods there are by average observed changes of trivial magnitude. Specifically, substantial alterations where evident only for IE (decreased) and sub-maximal exercise performance (improved). From MCP to ECP all the examined parameters tend to decrease with a trivial magnitude and substantial negative alterations been observed for VO_2max and IE. This undesirable dynamic in certain physiological determinants and endurance-related performance measures could be explained by the tight in-season schedule, with most of the time dedicate to recover from the previous match and prepare the strategy for the next opponent. In this regard, if a “window of opportunity” occurs (e.g., player ban as result of a red card and players not selected for national team breaks) further in-season improvements in aerobic and anaerobic qualities determinant for the running capacity and sub-maximal and maximal soccer-running performances than can be achieved through normal training routines may be obtained by incorporation of short duration HIT blocks (Christensen et al., 2011; Wahl, Guldner & Mester, 2014). As an example, although positive adaptations in running economy have mainly been reported and investigated during the pre-season, there are recent reports of increased running economy (75% of MAS) in players after performing 2 weeks of intense HIT executed just after the competitive season ended (Christensen et al., 2011); this result suggests that players still have significant physiological and performance adaptation potential to be explored. Nevertheless, caution is needed when extrapolating these findings for professional players as these experimental studies were performed by amateur and semi-professional players (Christensen et al., 2011; Wahl, Guldner & Mester, 2014). Nonetheless, it seems that special attention should be given to neuromuscular involvement during HIT (Bogdanis et al., 2009; Bogdanis et al., 2011) and to the concurrent effect of HIT (McGawley & Andersson, 2013), as it may be a determinant of the gains in running capacity during a short in-season intervention period. Nevertheless, being a very sensible process, these intervention periods require individualized management of the training/match load (Silva & Rebelo, 2019). In fact, within the same team a player may “underperform” as result of an over exposure while other player could be “underperforming” because of a detraining-related condition (Silva & Rebelo, 2019). Finally, the observations of a long-term persistent trend towards faster players and increased game speed (shorter and more “explosive” sprints and higher maximal running speeds) as well specific technical variables (e.g., passing rates) should be reflected not only on players selection but also in the training organization (e.g., physical conditioning). Regarding the latter, training should “feed” the players ability to perform maximal neuromuscular efforts and to repeat them over time; with the level of perceptual-cognitive demands varying according to each individual player needs.

Monitoring

Notwithstanding some techniques applied in research settings provide valuable information (valid, reliable), its utilization in routine operations within the club setting is limited (Silva & Rebelo, 2019). The imposed physical demands (e.g., maximal tests) and the invasive nature may explain at least in part the scarce applicability of several techniques in the real-world scenario (Carling et al., 2018). How motivated a player is for performing an end of season maximal testing session? This has obviously implications in the analyzed performance measures derived from testing sessions and match analysis, more influential through ECP assessments. Training may represent the perfect ecological setup to use as a ‘lab’ and shed some light as to the training status of the player (Silva & Rebelo, 2019). To this aim, a more action-oriented approach is needed; information derived from training sessions with tools that allow the simultaneous, instantaneous and non-invasive capture of multiple sources of information (Carling et al., 2018; Morin et al., 2021). As example, there are specific periods such as the warm-up and/or the main part of the training session (during gym or field sessions) that can be used to collect more precise information (neuromuscular and cardiorespiratory) regarding the players training status (Silva & Rebelo, 2019). As example, Morin et al. (2021) recently investigate an in-situ approach to directly assess individual acceleration-speed profile. Moreover, standardized drills with planned (e.g., sub-maximal running drill and or passing drills) (Buchheit et al., 2013) and unplanned (e.g., small sided-games) external load (precise and imprecise behaviors, respectively) can be applied within those parts of training practice to gain insight on players training status (Brink et al., 2012; Morin et al., 2021; Rago et al., 2017; Rago et al., 2018; Rowell et al., 2018b). As example, given the clear disconnection between running economy assessment methodology and soccer-specific activity during training and matches, there is some evidence that soccer-specific work economy may somewhat improve during the season; the relevant gains may not be detectable by conventional treadmill testing (Helgerud et al., 2001; McMillan et al., 2005). Nevertheless, although this monitoring strategy is not applicable without the coach’s approval, it builds an avenue for increased player “buy in” (Silva & Rebelo, 2019). A wide range of information can be collected during a standardized warm-up (Buchheit et al., 2013) that can inform on the training status of the athlete (Halson, 2014). As an example, the examination of physiological (HR) and perceptual (RPE) indicators of load in sub-maximal running exercise can provide valuable information on players cardiorespiratory fitness and fatigue level (Halson, 2014). Additionally, information on neuromuscular training status can be collected within this training stage and other periods (e.g., standardized small-sided game) by means of GPS and accelerometer-derived metrics (e.g., load per minute, load triaxial contributions) (Cormack et al., 2013; Morin et al., 2021; Rowell et al., 2018b). This latter monitoring strategy when applied during specific moments of the microcycle offers a great ecological and valid option for monitoring training status (Rago et al., 2017; Rago et al., 2018; Rowell et al., 2018b). In fact, this has been recently investigated in order to overcome the limitations (e.g., time for testing, isolated tests) of assessing elite players when using more traditional moments and tools (Rago et al., 2017; Rago et al., 2018; Rowell et al., 2018a). Nevertheless, when using standardized small-sided games, well-known factors that affect players exercise intensity need to be regarded (e.g., space, duration and team structure), but alsoteam constitution should be maintained stable (if possible the same players in each team) (Silva & Rebelo, 2019).