Participants

Twelve (n = 12) professional male 3x3 basketball players (age: 26.6 ± 5.6 years, body height: 196.8 ± 5.5 cm, body mass: 92.0 ± 9.5 kg, training experience: 15.4 ± 3.9 years, and 3x3 basketball training experience: 4.8 ± 1.6 years) from the Lithuanian national team were recruited for this study. The selected participants were 3^rd place winners in the 3x3 Europe CUP 2022, 3^rd place winners in the 3x3 basketball competition at the 2024 Summer Olympic Games, and 3^rd place winners in the 3x3 Europe CUP 2024. All athletes were cleared by medical and coaching staff to participate in team training activities and competition. The case study received ethical approval from the Institutional Ethics Committee of the Vytautas Magnus University (approval code: SA-EK-21-04; approval date: 30 November 2021) in accordance with the Declaration of Helsinki guidelines.

Design and Procedures

An observational study was conducted during the preparation phase for the 2024 Summer Olympic Games and the 2024 European CUP. A total of two simulated 3x3 basketball games were played during the data collection period, with each player participating in one simulated game. The simulated 3x3 basketball games were played outside on standard FIBA approved 3x3 flooring designed for outdoor play (15 m width, 11 m length, 3.05 m basket height), under the sunny weather conditions (air temperature 23.5 ± 3.5^°C, wind speed 3.1 ± 0.8 m/s). Prior to each game, players completed a 20-min warm-up session, which included dynamic stretching and specific basketball-based movement exercises, under the supervision of a strength and conditioning coach. The simulated games were structured into four playing periods of varying duration (1, 2, 3, and 4 min), with 1 min of passive recovery. The official 3x3 basketball rules (https://fiba3x3.com/docs/fiba-3x3-basketball-rules-full-version.pdf) were modified by removing free-throw shots and substitutions, and by incorporating 1-min recovery periods between each playing period. These adjustments were made to replicate the actual playing conditions (substitution of players and resting on the bench) and to assess HR and SmO₂ recovery rates, following different durations of 3x3 basketball play.

Data Collection and Processing

External load variables were measured using VXSport inertial measurement devices (VXSport, Wellington, New Zealand), which sampled data at a rate of 100 Hz (Smith et al., 2024). The VXSport units were placed between the scapulae using the manufacturer’s purpose-designed vest, ensuring a secure fit prior to the simulated 3x3 basketball games. These triaxial devices, equipped with gyroscopes, accelerometers, and magnetometers capture movement changes in all directions. VXSport devices were used to measure both the volume and intensity of the external load. Total distance (TD) (m), sprints (ST) (n), and the jump count (n) were selected as variables of external load volume. For external load intensity, the following variables were assessed: a distance rate (DR m/min), average speed (speed_avg, km/h), maximum speed (speed_max, km/h), a jump rate (JR, jumps/min), and average height jump zone total (AHJZT, cm) (Smith et al., 2024).

VXSport devices were also employed to analyse continuous HR measurements, using Suunto HR sensors (Suunto Smart sensor, Suunto, Oy, Finland) (Smith et al., 2024). These sensors were integrated into a manufacturer-designed vest compatible with strapless HR monitoring. The HR devices were positioned between the scapulae, next to the VXSport inertial measurement devices. This purpose-designed vest, with integrated HR sensors, allowed for strapless HR measurements. The HR data were stored on the VXSport devices via Bluetooth connection. The collected HR data were used to calculate the average HR (HR_avg, bpm), and the maximum HR (HR_max, bpm) during play periods and the resting HR (rest HR, bpm) during recovery periods.

Following data collection, the external load and HR data were downloaded, stored, and processed using VXSport software (VXSport, VXSport release 7.1.0.4). The processed data were then exported for further analysis and calculations.

Near-infrared spectroscopy (NIRS) measurements were conducted using the Moxy monitor devices (Moxy, Fortiori Design LLC, Minnesota, USA) (Paulauskas et al., 2020). The Moxy device operates by sequentially emitting light waves (630–850 nm) from four light-emitting diodes into the underlying tissue and recording the amount of scattered light returning at two detectors positioned 12.5 and 25 mm from the light source.

The depth of light penetration corresponded to half the distance between the light source and the detector. The scattered light was processed by an algorithm that integrated a tissue light propagation model and the Beer-Lambert law to quantify the absorption of light at wavelengths associated with oxygenated and deoxygenated hemoglobin (Hb).

Moxy devices were positioned on the participant’s dominant leg, specifically on the vastus lateralis muscle, halfway between the greater trochanter and the lateral epicondyle of the femur. Before placement, the area was trimmed with an electric razor (if necessary) and cleaned with alcohol swabs. The device was secured with a light shield and athletic tape to prevent ambient near-infrared light from interfering with the detectors. SmO₂ was continuously monitored throughout the simulated 3x3 basketball game, and the data were used to calculate average SmO₂ (avg SmO₂, %) during playing periods and resting SmO₂ (rest SmO₂, %) during recovery periods. After data collection, SmO₂ data were downloaded for further calculations.

To assess the metabolic-anaerobic demand in athletes, blood lactate (BLa) concentration (mmol/L) was measured 3 min post-completion of simulated 3x3 basketball game (Paulauskas et al., 2020). Blood lactate samples were collected from the fingertips of each participant and immediately analysed using a validate lactate analyser (Lactate Pro, Arkray, Tokyo, Japan).

The HR and SmO₂ relative (%) delta changes (Δ) during recovery phases were utilized as key metrics for evaluating recovery indices. These changes were quantified by subtracting values recorded at the 15^th, the 30^th, the 45^th, and the 60^th s as a percentage of the game stop signal, which was defined as the HR and SmO₂ measurements at the initiation of the recovery period (Dupuy et al., 2012). Recovery indices were calculated for each 1-min recovery period, following playing intervals of 1, 2, 3, and 4 min during a 3x3 basketball game.

Statistical Analysis

A descriptive analysis was conducted, providing means and standard deviations for the dataset. The normality of the data distribution was assessed using the Shapiro-Wilk test. Group comparison was performed using a mixed linear model for repeated measures, incorporating both fixed and random effects. The random factor accounted for inter-subject variability, while intraclass correlation coefficients were calculated to evaluate the reliability of repeated measures. Marginal and conditional R2 values were provided to explain variance attributable to fixed effects and the entire model, respectively. To estimate the strength of significant findings, effect sizes (ES) were determined using the Cohen’s d. Effect size values were interpreted as follows: < 0.20 represented a trivial effect, 0.20 to 0.49 was classified as a small effect, 0.50 to 0.79 corresponded to an intermediate effect, and 0.80 and higher was considered a large effect (Cohen, 1988; Cumming, 2017). All data sets were tested for each statistical technique corresponding assumptions. Statistical analyses were performed using SPSS (Version 20 for Mac; SPSS Inc., Chicago, IL, USA) and statistical significance was set at p < 0.05.

Training Load Progression

This study represents the first attempt to quantify the physical performance variables during a simulated 3×3 basketball game across distinct intervals: 1 min, 2 min, 3 min, and 4 min. The findings reveal that DR, Speed_avg, and the JR progressively decreased as play duration increased, while TD, ST, and JC volumes significantly increased. These changes had a corresponding impact on internal exercise variables such as SmO_2avg, SmO_2min, HR_avg, as well as HR_max and Bla (Table 1), underscoring the intricate relationship between intensity and volume in 3x3 basketball.

Previous research demonstrated that members of the German national 3x3 team maintained an average DR of 64 m/min during an international tournament (Willberg et al., 2023). Our study, using distinct game intervals, exceeded this benchmark. Specifically, in the 4-min interval, DR was lower than in shorter intervals, with an average of 76.67± 7.52 m/min. This difference might reflect the higher skill level of our participants or unique game dynamics, including recovery strategies. DR was also correlated with both speed_avg (km/h) and speed_max (km/h), both of them decreased as the exercise duration extended. Speed is a critical performance metric in basketball, reflecting overall physical capacity (Fitness-Requirements-of-3x3-Players.Pdf, n.d.). However, in-game velocity is often constrained by tactical roles, which vary with factors such as the playing style, the position, and game outcomes (Buchheit and Laursen, 2013), alongside limitations imposed by court dimensions and space availability. These tactical limitations can attenuate the relationship between a player’s maximal speed capacity (measured in lab or field settings) and their actual game performance (Mendez-Villanueva et al., 2011). Our comparison of in-game speed with controlled test results demonstrated that Speed_max during the 4-min interval was 30% lower than previously established speed benchmarks (Fitness-Requirements-of-3x3-Players.Pdf, n.d.). This supports the argument that in-game speed is a more reliable indicator of physical performance in 3x3 basketball than test-derived maximal velocity estimates.

In terms of the JR, players averaged 2.92 ± 1.98 jumps/min during 1-min intervals, closely matching the rate reported at the 2019 FIBA 3x3 World Cup rate (3.16 ± 0.9 jumps/min) (Ferioli et al., 2023). In contrast, German team players averaged 1.8 ± 0.6 jumps/min during an international 3x3 tournament (Willberg et al., 2023). In our study, the JR declined to 1.59 ± 0.90 jumps/min after 2 min of play. The significant reduction in the JR during the 4-min interval, as compared to the 1-min interval, likely resulted from neuromuscular fatigue and the critical dependence on ATP-PCr resynthesis for explosive movements.

High-intensity exercise induces rapid PCr depletion-up to 60%–80% within the first 30 s of activity, with as much as 70% depleted within the first 12 s (Spencer et al., 2005). Prolonged high-intensity exertion can lead up to 89% depletion of PCr stores (Hirvonen et al., 1992), necessitating varying recovery time post-exercise. PCr resynthesis occurs in two phases: an initial rapid phase followed by a slower one, with recovery rates depending on the type of recovery (active or passive) (Spencer et al., 2005). The half-life for PCr resynthesis ranges from 21 to 57 s, depending on exercise intensity and volume (McMahon and Jenkins, 2002). Approximately 65% of PCr can be restored within 90 s of rest, while full recovery (less than 90%) may take up to 6 min (McMahon and Jenkins, 2002). Intense exercise also leads to a sharp increase in ATP hydrolysis, which can be impeded by reductions in pH and increases in ATP, PCr, citrate, and free fatty acids. Elevated levels of ATP and PCr indicate a recovery state, reducing the reliance on glycolytic pathways during the initial stages of game intervals (Hirvonen et al., 1992).

A decrease in AHJZT during the 2-min interval may suggest the onset of rapid muscle fatigue, whereas a significant increase over the 4-min interval could reflect glycogen replenishment through the anaerobic metabolism. This conclusion is corroborated by our test results which recorded a Bla concentration of 11.71 ± 5.29 mmol/L at the conclusion of the simulation game. Glycogen is as a primary energy source for exercise performed at intensities exceeding 50%–60% of VO_2max (maximal oxygen uptake) (Mul et al., 2015). Depletion of glycogen reserves can contribute to muscle fatigue during aerobic exercise or prolonged high- intensity anaerobic activity (Abernethy et al., 1990).

The HR is frequently used as a proxy to regulate exercise intensity in sports (Achten and Jeukendrup, 2003). Elite 3x3 athletes, as studied by Montgomery and Maloney (2018), exhibited a maximum HR of 198 bpm and an average HR of 164 bpm, with consistent values across games. In our research, a similar average HR was achieved after 2 min, with a maximum HR peaking at 187 ± 9.4 bpm. However, the HR may not fully capture physical exertion during very short (30 s) or moderate-length (1–2 min) exercise intervals (da Cunha et al., 2011), largely due to the delay in HR response relative to VO₂ uptake (MacDougall et al., 1977). Muscle deoxygenation, which responds more rapidly than VO₂, is driven by the oxygen demands of active muscles and modulates O₂ delivery by the cardiovascular system (Hesford et al., 2013).

Metrics such as muscle deoxygenation and the HR are modulated by external variables such as TD, sprints, jumps, and speed. Both average and SmO₂ min suggest a consistent peripheral impact on oxygen extraction across all exercise intervals. NIRS-based assessments provide a more precise evaluation of exercise demands (Perrey and Ferrari, 2018), highlighting individualized physiological responses to players' diverse on-field activities. Consequently, individual player assessments yield deeper insights compared to evaluating the cumulative workload of a team.

Post-Exercise Recovery

This study explored post-exercise recovery dynamics by analysing HR and SmO₂ indices during specific 3x3 game intervals, as illustrated in Figures 2 and 3. The HR-derived variables demonstrated notable variations across all four game intervals, indicating that recovery rates are influenced by exercise volume.

Following a 1-min play interval, the HR was observed to be at its lowest point immediately at the onset of recovery and after one minute of rest. In contrast, after the 2-, 3-, and 4-min play intervals, HR levels were elevated immediately post-exercise and exhibited varied recovery rates during the subsequent rest periods. The relative Δ HR values also fluctuated across different interval duration, underscoring the variability in recovery (Table 4). For instance, after a 1-min play interval, the recovery Δ 45% was −21.58 ± 8.68, closely mirroring the recovery Δ 60% after a 2-min interval, at −21.20 ± 5.94. Additionally, after a 1-min play interval, the recovery Δ 15% was −3.30 ± 2.47, comparable to the recovery Δ 45% post-4-min play at −4.01 ± 1.81. This pattern is likely attributable to the metabolic demands of anaerobic glycolysis and associated energy neuromuscular strain, contributing to varied fatigue levels (Buchheit and Laursen, 2013).

Figure 2

Fluctuations in SmO₂ values during 60-s recovery periods, following 1, 2, 3, and 4 min of play in a 3x3 basketball game. Data are presented as absolute values (mean ± SD).

Figure 3

Fluctuations in HR values during 60-s recovery periods, following 1, 2, 3, and 4 min of play in a 3x3 basketball game. Data are presented as absolute values (mean ± SD).

In athletes, HR recovery is primarily governed by the autonomic nervous system, specifically through parasympathetic reactivation and sympathetic withdrawal (Perini et al., 1989). The autonomic nervous system's interaction with other physiological systems, particularly reflected in HR recovery, is considered an indicator of the body’s capacity to manage exercise-induced stress (Facioli et al., 2021). Our investigation into HR recovery dynamics in basketball players, following different playing interval duration, supports this premise. Previous research has shown that Δ 60 was significantly higher after exercise at 70–80% VO_2max compared to 60% VO_2max, implying a more rapid recovery in individuals with higher HR levels (Mann et al., 2014). However, the study by Mann et al. (2014) focused on submaximal exercises that involved relatively low anaerobic energy expenditure. Consequently, the lactate accumulation affecting HR recovery kinetics (Buchheit et al., 2007) may have been modest and less comparable to the physiological demands experienced during 3x3 basketball.

In 3x3 basketball, where only one reserve player is available, both work and rest duration is highly individualized. Recovery periods are often tailored by utilizing either a fixed value or a percentage of HR_max (Buchheit and Laursen, 2013). However, emerging evidence suggests that such an approach may not be particularly effective (Seiler and Hetlelid, 2005). During recovery, the HR is not directly linked to systemic O₂ demand or muscle energy turnover (Buchheit et al., 2012), but instead correlates more strongly with the intensity of central command and metaboreflex activation (Rowell and O’Leary, 1990). Table 3 highlights the distinct dynamics between muscle reoxygenation and HR recovery, with the rates of relative recovery Δ SmO₂ and Δ HR varying according to the work duration. In simulated play scenarios, specific training protocols facilitated quicker SmO₂ recovery to pre-exercise levels.

Notably, reoxygenation levels across all work intervals remained consistent, as reflected in the uniform Δ SmO₂ % recovery values across 1-, 2-, 3-, and 4-min intervals. Previous research has indicated that different exercise modalities do not significantly impact muscle reoxygenation (Paulauskas et al., 2020). Our findings suggest that O₂ uptake recovery may be less dependent on exercise duration or the type of activity, and more related to the athlete’s ability to resist fatigue (SmO₂% decrease) and replenish energy substrates (ATP, PCr), both of which are oxygen-dependent processes (Dupont et al., 2010).

There is strong concordance between recovery time constants derived from NIRS and PCr recovery time constants measured via magnetic resonance spectroscopy (Hamaoka et al., 1996), as well as a strong correlation with oxidative capacity assessed through muscle biopsy (Ryan et al., 2014). The similarity in muscle recovery levels may be influenced by factors such as neural drive efficiency, motor unit activation patterns, and metabolite accumulation (Glaister, 2005).

The results of this study suggest that varying training loads during 3x3 basketball do not significantly alter peripheral muscle oxygenation, and that SmO₂ can serve as a reliable predictor of hemodynamic changes. Moreover, the decline in the HR is likely linked to parasympathetic reactivation, which reflects the body’s overall response to acute stress from exercise.