The experimental is scheduled for the off-season period from March to April 2023. Prior to the experimental intervention, both the experimental and control groups participated in a 2-week familiarization training program consisting of four sessions. Testing was conducted 48 h before the first experimental intervention and 48 h after the last intervention. The assessments included the Countermovement Jump (CMJ), Reactive Strength Index (RSI), Running Vertical Jump, Eccentric Utilization Ratio, and a one-repetition maximal (1RM) barbell back squat test. All tests and experimental interventions were carried out at the same venue under the direct supervision of the same experimenter. The Flywheel Resistance Squat Training (FRST) group and the Traditional Resistance Squat training (TRST) group participated in a 6-week training program. Additionally, Coratella et al. found that following flywheel training, isometric peak torque and muscle soreness returned to values comparable to baseline after 48 and 72 h, respectively (Coratella et al., 2016). Therefore, our experiment will implement interventions on Mondays and Thursdays, occurring twice weekly. The total number of training sessions and the duration of technical and tactical training were consistent for both groups; therefore, we assume that the loading is equivalent between FRST and TRST.

To ensure participants’ familiarity with the content of the program, the following steps were taken in this study: ① to determine the barbell height and belt size for training, participants’ height and waist circumference were measured through interviews and anthropometric methods; ② Participants were provided with a detailed introduction to the standardized operating protocols and technical aspects of flywheel resistance training and barbell squat exercises. To standardize the experimental procedures and ensure consistency across trials.; ③ Subjects were briefed on potential risks, safety procedures, and ethical aspects by signing an informed consent form; ④ During the last familiarization session, tests related to explosive power and maximal strength were conducted, and the load for an 80% 1RM barbell squat was estimated accordingly.

For trainees inexperienced with flywheel resistance training (FRST), research suggests that a minimum of two training sessions is necessary to achieve stability in force production (Sabido et al., 2018). To ensure participants become familiar with the training equipment and master the experimental procedures as well as the correct techniques for applying force during squat training on the flywheel trainer, a total of four familiarization sessions were scheduled.

Maximal Force Test: Maximal lower-body strength was assessed using the one-repetition maximum (1RM) barbell back squat, in accordance with the protocols outlined in the NSCA’s Guide to Tests and Assessments. Before the formal test, an estimated 1RM was calculated based on the participant’s body weight and training experience. After a general warm-up, participants began with a weight they could comfortably lift for 5 to 10 repetitions. They rested for 1 min before increasing the weight by 10%–20% of their body weight, estimated to allow for 3-5 repetitions. After a 2-minute rest, the weight was increased by another 10%–20%, allow for 2-3 repetitions. Following a 2–4 min rest, the weight was increased by an additional 10%–20%, and participants were asked to perform a 1RM attempt. If successful, the weight was increased by 5%–10% of their body weight after a 2–4 min rest; if unsuccessful, the weight was decreased by 5%–10% and retested. This process continued until a successful 1RM squat was achieved (Miller, 2012).

Explosive Power Test: The assessment comprised CMJ, RSI, Eccentric Utilization Ratio and Running Vertical Jump. CMJ and RSI were measured using the Smart Jump mat from Australia, while the Running Vertical Jump was assessed with a touch-and-go device. For CMJ, participants stood with hands on hips, performed a rapid knee flexion squat to their best angle, and then jumped as high as possible. The Eccentric Utilization Ratio was derived from the ratio of CMJ to SJ (McGuigan et al., 2006), with SJ requiring participants to jump vertically with maximal effort from a knee angle close to 90°, avoiding significant lengthening of the lower limbs. The RSI test required athletes to stand on a 45 cm high box, with hands on hips. After a natural drop of one foot forward, they quickly squatted and jumped as high as possible. The Running Vertical Jump test involved a run-up and jump, with both feet touching the ground prior to takeoff, and the participants’ arms actively swinging. At the peak of the jump, the participant’s fingers brushed the touch-and-go device. All tests required participants to avoid knee flexion, hip extension, or other extraneous movements in the air that could affect the time spent in flight (Jiménez-Reyes et al., 2017). Each explosive force test was conducted twice with 2–3 min intervals, and the best result was used for analysis.

Prior to each experimental intervention, participants performed a standardized warm-up based on the RAMP principles outlined by the National Strength and Conditioning Association (NSCA) (Jeffreys, 2006; Jeffreys, 2017). The warm-up consisted of 2 min of foam rolling (targeting the gluteus maximus, quadriceps, hamstrings, tibialis anterior, and gastrocnemius), 3 min of dynamic stretching (including heel raises, cradle stretch, the greatest stretch, and A-skips), 3 min of neural activation (T-rolls, cross steps, and continuous vertical jumps with sprints), and 2 sets of 10 bodyweight squats.

The TRST group performed barbell squats using a free squat rack, with the training intensity set at 80% of their 1RM as recommended by NSCA for effectively improving lower-body maximal strength (Haff and Triplett, 2015). Thus, the TRST group’s training load was 80% of 1RM, comprising 4 sets of 8 repetitions with 3-minute rests between sets. Additionally, to optimally plan the induced stimuli, we presented a comprehensive overview of the training variables (Coratella, 2022) Specifically, ①Range of Movement: full back-squat (0°–140° knee flexion); ②Time Under Tension and Contraction form: During a complete action involving eccentric, isometric, and concentric phases, use a 1–0-1 s tempo; ③focus attention on the entire movement process or task.

Research (Stojanović et al., 2021; Coratella et al., 2019) indicates that an inertia greater than medium can effectively enhance lower-body strength. Additionally, a meta-analysis shows that an inertia moment of 0.075 kg ·m² can produce greater eccentric loads (Sabido et al., 2018) and higher contraction speeds (Piqueras-Sanchiz et al., 2019), positively influencing strength development. Therefore, the FRST group’s training intensity was set at an inertia moment of 0.075 kg ·m², equivalent to 40% of 1RM output power, consisting 4 sets of 8 repetitions with 3-minute rests intervals between sets. During training, participants completed 2 flywheel resistance squats before each set to accelerate the flywheel’s rotation. In each session, participants were required to train with maximum eccentric resistance during the eccentric phase until their hip joints were either lower than or level with their knee joint (Miller, 2012) The Desmotec (Italy) flywheel resistance training equipment was used, which connects via Bluetooth to a client device (Apple or Android) and employs the kMeter software system to monitor training data in real-time. Detailed training content is presented in Table 2.

Descriptive statistics are presented as mean ± standard deviation (SD). The level of statistical significance was set at p ≤ 0.05. All statistical analyses were conducted using SPSS version 23.0. Normality and homogeneity of variance for all variables were assessed using the Shapiro-Wilk test and the Levene test, respectively. A repeated measures analysis of variance (ANOVA) was employed to evaluate differences between the two groups (FRST vs. TRST), over time (pre-test and post-test, referred to ‘time’) and for the group-by-time interaction effects, with Sidak’s post hoc test applied to each outcome measurement. The test-retest reliability of specialized ability tests, such as the Running Vertical Jump was evaluated using coefficients of variation (CV) (Cormack et al., 2008) and intraclass correlation coefficients (ICC) (Koo and Li, 2016) along with a 95% confidence interval (CI), applying a one-way random effects model. Reliability was assessed using a custom spreadsheet (Hopkins, 2006). An ICC value of less than 0.5 indicates poor reliability, between 0.5 and 0.75 indicates moderate reliability, between 0.75 and 0.9 indicates good reliability, and greater than 0.90 indicates excellent reliability. A CV of less than 10% is considered reliable. The effect size of within-group paired differences was estimated using Hedges’ g, with the magnitude of within-group differences categorized as follows: trivial (Hedges’ g ≤ 0.2), small (0.20 < Hedges’ g ≤ 0.60), medium (0.60 < Hedges’ g ≤ 1.20), large (1.20 < Hedges’ g ≤ 2.00), or very large (Hedges’ g ≥ 2.0) (Cohen, 2013). The effect size of the intervention’s group differences was measured using partial eta squared ( η p 2 ), with the degree of group difference categorized as: small 0.01 ≤   η p 2 ≤ 0.06 , medium 0.06 ≤ η p 2   <   0.14 , and large η p 2 ≥ 0.14 (Lachenbruch and Cohen, 1989).

The ANOVA results show significant differences within the groups for the 1RM squat. (p < 0.001, η p 2 = 0.66 ; Figures 1–A), but not between groups (p = 0.99). Post-hoc comparisons revealed differences between pre- and post-tests for both FRST and TRST (p(FRST) = 0.01; p(TRST) = 0.01), with moderate effect sizes observed (Table 3).

The CMJ results show significant differences within groups (p < 0.001, η p 2 = 0.49 ), but not between groups (p = 0.10; Figures 1–B). Post-hoc analysis shows significant differences within the FRST group (p = 0.02), with a moderate effect size (Hedges’ g = 0.59), whereas another group showed the opposite (p = 0.23).

The eccentric utilization ratio (Figures 1–C) and the Reactive Strength Index (Figures 1–D) demonstrated no significant within-group or between-group differences (Table 3). Conversely, the Running Vertical Jump (Figures 1–E) revealed significant within-group differences (p = 0.04, η p 2 = 0.22 ), with no significant between-group differences (p = 0.30; Table 3). Reliability measures for the RSI at baseline showed an ICC of 0.60 (95% CI: 0.34–0.80) and a CV of 26.3%. For the two-foot vertical jump test, reliability was reported with an ICC of 0.45 (95% CI: 0.01–0.60) and a CV of 9.0%.

Maximal strength is a critical indicator of basketball performance capability, and the 1RM squat is a comprehensive measure of strength levels (Izquierdo et al., 2006). After 6 weeks of FRST and TRST, both groups of female basketball players showed improvement in lower limb maximal strength, with moderate effect sizes within groups and high effect sizes between groups. A recent meta-analysis on flywheel resistance training and traditional resistance strength training conducted on healthy individuals reported similar results. This analysis included ten studies, founding no significant difference between flywheel resistance training and traditional resistance training in enhancing maximal strength (Hu et al., 2024). However, the study indicated that a flywheel resistance training program lasting 12 to 18 sessions can significantly enhance gains compared to traditional resistance training and increase statistical power, as it provides sufficient recovery and over-compensation periods. Therefore, the training cycle will be one of the important factors influencing the effectiveness of flywheel resistance training on maximal strength. Additionally, Bordas et al., ’s 2018 study conducted a study on flywheel resistance training from the perspectives of root mean square electromyography and maximum voluntary contraction (Vicens-Bordas et al., 2018), which also provides physiological supports for our findings.

Several studies confirm that TRST primarily promotes muscle growth and strength increase by increasing load and repetition (Hather et al., 1991; Dudley et al., 1991). In contrast, Fa 6-week FRST study on female soccer players found that flywheel resistance training produced greater overload during the eccentric phase (Pecci et al., 2022), which is advantageous for muscle performance. Furthermore, a systematic review indicates that flywheel resistance training, when properly executed, can produce mechanical eccentric overload, generating greater force during the eccentric phase compared to concentric peak force or power (Beato et al., 2024). FRST’s eccentric overload induces prolonged and high eccentric strain (Douglas et al., 2017), resulting in greater peak tension for the same amount of work, while reducing metabolic cost, thereby maximizing muscle strength, remodeling muscle structure, enhancing motor unit synchronization (Sale, 1988), and stimulating higher electromyographic activity (Moritani and Vries, 1979).

On the other hand, skeletal muscle damage and inflammation can promote muscle protein synthesis and satellite cell differentiation (Chazaud, 2016). Annibalini et al. demonstrated that muscle damage induced by FRST can trigger early molecular responses and systemic muscle adaptations (Annibalini et al., 2019), while altering key markers in the later stages of muscle remodeling and functional adaptation. However, Hakkinen’s research suggests that traditional resistance training can also induce similar muscle adaptations (Häkkinen and Fitness, 1989). Overall, these physiological differences support our findings and highlight the effectiveness of both flywheel and traditional resistance training in enhancing strength in female basketball players.

Nevertheless, a study on amateur football players found that after 6 weeks of intervention, TRST was more effective in increasing squat maximal strength (Sagelv et al., 2020). This discrepancy might be due to differences in maximal neural activation and recovery abilities among athletes of varying skill levels (Petré et al., 2018). Additionally, this study observed that both the experimental and control groups had increasing loads (experimental group: 0.025 kg ·m⁻² to 0.1 kg ·m⁻²; control group: 70% 1RM to 85% 1RM). Research indicates that moderate to high inertial moments are most effective in improving maximal strength, which may explain the differences in results.

Specialized basketball techniques (e.g., critical rebounds, blocks, and dunks) are closely related to lower limb explosive power (Abdelkrim et al., 2010; Alemdaroğlu, 2012) This study selected CMJ, RSI, Reactive Strength Index and Eccentric Utilization Ratio as direct variables to assess the impact of FRST and TRST on explosive power. After 6 weeks of intervention, results showed no significant differences between FRST and TRST in improving CMJ scores for female basketball players. However, both methods exhibited varying degrees of improvement compared to baseline, with the FRST group showing significant within-group differences. This finding aligns with Coratella et al.'s previous research (Coratella et al., 2019), which found that both FRST and TRST improved CMJ performance in football players, although no significant between-group differences were observed. Additionally, a 6-week study on female football players (Pecci et al., 2022) found no significant effects of either training method on CMJ scores, possibly due to a lack of high-intensity jumping in their training. However, Fernandez et al.'s 6-week study (Gonzalo et al., 2014) indicated that FRST improved CMJ performance in both male and female subjects, suggesting that higher training frequency and longer training periods may be more beneficial for improving CMJ scores in female athletes. In contrast, Wang Jiaoqin et al.'s research (Wang et al., 2024) found a significant improvement in CMJ scores for female volleyball players after 8 weeks of intervention (p = 0.04). Thus, training frequency, cycle length, and the types of exercises selected may contribute to discrepancies between this study’s results and those of other studies. A meta-analysis (Hu et al., 2024) also supports this view, indicating that the choice of training program significantly impacts research outcomes.

Gonzalo et al. proposed that FRST can activate more cross-bridges, generating higher levels of force during the eccentric phase of the stretch-shortening cycle (SSC) in CMJ (Gonzalo-Skok et al., 2017). This mechanism may lead to improved SSC performance. Additionally, increased threshold excitability of the Golgi tendon organs (McNeely and Sandler, 2007) and changes in neuromuscular coordination may contribute to the improvement in CMJ performance among female basketball players. However, other studies suggest that traditional resistance training, by enhancing muscle and tendon elasticity, can achieve similar changes (Häkkinen and Fitness, 1989).

The Eccentric Utilization Ratio has been validated as a useful indicator for assessing elite athletes’ training status and their utilization of the stretch-shortening cycle (Hawkins et al., 2009). After 6 weeks of intervention, the ratio showed no statistical differences either within or between groups, but both training methods demonstrated positive effects. A value of 1.1 not only reflects excellent explosive performance but is also an ideal standard sought by coaches (Balsom, 1994). Despite the absence of significant differences in Eccentric Utilization Ratio between the groups, the FRST group was higher than another, approaching the optimal value of 1.1. Thus, FRST may be more effective in improving this metric. Additionally, a study involving 142 team athletes indicated that the Eccentric Utilization Ratio based on height before and after the season showed results consistent with those observed in this study (McGuigan et al., 2006). Another study (Martinez-Aranda and Fernandez-Gonzalo, 2017) suggests that FRST can enhance the ability to repeatedly perform high-intensity eccentric actions following a period of training, leading to improvements in the Eccentric Utilization Ratio. The Reactive Strength Index measures an athlete’s reactive jumping ability and ability to handle impact forces (Jarvis et al., 2022). A meta-analysis on flywheel resistance training showed results differing from this study (Buonsenso et al., 2023), indicating positive results for RSI in most samples with different training programs. This discrepancy might be due to the short experimental period, which may not reach the threshold required to cause changes in RSI. However, the low Reliability measures for the Reactive Strength Index and the Running Vertical Jump in this study may have introduced errors in the analysis. Therefore, this study could not explore the impact of flywheel resistance training on specialized explosive power. This might be due to the complex nature of these jump modes, which require a combination of factors such as ground force application, proper force distribution, rhythm changes, and core stability. Future research should consider these factors comprehensively.