Interrelationships between countermovement jump,assisted acceleration–deceleration,and isometric belt squat metrics in American football players - exploratory study

Participants

Twenty-four male American football athletes from a Polish semi-professional team volunteered to participate in the study (mean ± SD: age, 23.5 ± 4.6 years; body mass, 90.6 ± 8.3 kg; and height, 179.7 ± 4.0 cm). The participants competed at the highest level of play in Poland. All participants played skill positions (including linebackers n = 7, tight ends n = 2, wide receivers n = 4, running backs n = 1, and defensive backs n = 10). They were accustomed to structured strength and conditioning programs and had completed at least three of the same test batteries in the previous 12 months. The average strength and conditioning training experience was (6.9 ± 4.0 years), and the average American football playing experience was (6.1 ± 4.3 years). The typical weekly volume during the testing period was 4.9 (±1.1) weekly strength and conditioning units per weekly microcycle, and 0.4 (±0.7) specific field units. All participants declared no acute musculoskeletal injuries and were cleared for complete training by the team medical staff. Inclusion criteria included participation in full training during the preparatory period and the absence of pain that limited training and testing. Exclusion criteria included acute musculoskeletal symptoms, functional limitations, a history of injury within the previous 6 months, and lack of consent to participate in the study.

The tests were performed three months before the start of the league, in the preseason period, during which the training programme focused primarily on physical preparation rather than tactical and technical content. In American football, scheduling a battery of neuromuscular tests approximately 8–12 weeks before the season is a typical approach, as it establishes a baseline for athletic readiness and provides a safe, practical direction for further periodisation of strength, power, and speed. This is the phase in which players typically demonstrate a high level of general physical preparedness, which facilitates the implementation of demanding neuromuscular tests.

No confounding factors were identified on the day of the study. Participants were instructed to refrain from intense training for 24–48 h before testing, and caffeine/supplementation standards were established. The study was conducted in accordance with the Declaration of Helsinki. Before testing, all participants signed informed consent to participate in the study, and were informed that they could withdraw from the study at any time without incurring any sporting or organizational consequences. The study received a positive opinion from the Research Ethics Committee at the Jan Długosz University in Częstochowa. Resolution No. KE-U/9/2026 of 19 January 2026.

Research protocol

A cross-sectional study design was used. All tests were conducted within a single morning session (08:00–11:00) following one rest day without formal training, and players were tested in small sub-groups to reflect applied practice and to allow natural verbal encouragement from teammates and coaches. The testing battery consisted of a standardised warm-up protocol, CMJ no-arms-swing performed on dual force plates, ADA30 in assisted mode on the MRD, and IBS at approximately 120° knee flexion using a portable fixed dynamometer. The order of tests was consistent across sub-groups.

OPEN IN VIEWER

Figure 1.

The figure shows a graphical representation of the tests performed A - countermovement jump (CMJ); B - assisted 30 m acceleration-deceleration (ADA30); C - isometric belt squat (IBS).

Warm-up

Players completed a ∼15 min warm-up based on the RAMP (raise–activate–mobilise–potentiate) model, ³⁷ comprising: low-intensity running, skipping and mobility drills, lower-body and trunk bodyweight exercise, extensive low-amplitude plyometrics, and three progressive accelerations over 10 m. The same sequence was used for all sub-groups.

Countermovement jump (CMJ)

CMJ ground reaction force was sampled at 1000 Hz using a wireless dual force plate system (Hawkin Dynamics, Westbrook, Maine, USA) and transmitted in real time to the web-based Hawkin Dynamics Cloud via the Android Hawkin Capture application (version 9.10.1, Westbrook, Maine, USA). This system has demonstrated high criterion validity for assessing CMJ force–time variables compared with laboratory-grade force platforms.^38,39 Athletes stood with feet approximately shoulder-width apart and hands-on-hips. After a stable quiet phase, they performed a self-selected countermovement and jumped vertically in response to an audio-visual cue from a tablet. Each athlete performed 3 maximal CMJs, separated by approximately 5 s. The instruction provided was: “jump as high and as fast as possible”. Feedback was given after each attempt in the form of jump height, and the force plates were zeroed regularly between participants to ensure calibration stability.

From the manufacturer's web-based export function, the following 30 CMJ variables were retained for analysis and categorised into: performance outcomes (e.g., jump height, mRSI), movement strategy and timing measures (e.g., countermovement depth, phase durations, time to take-off), and kinetic/kinematic metrics from braking and propulsive phases (force, velocity, power, and impulse). All relative force and power variables were normalised to body mass (%BM, W/kg, N·s/kg) to enable comparisons between athletes. A complete list of the extracted variables, units, and their operational definitions is provided in Supplementary Table S1. For each variable, the mean of three CMJ trials was used for analysis. A graphical representation of the test is in Figure 1A.

Assisted 30 m acceleration-deceleration (ADA30)

Assisted acceleration, sprinting and deceleration were assessed using a 1080 Motion Sprint 2 device sampling at 333 Hz (1080 Motion, Lidingö, Sweden), in the assisted mode. Data were streamed live to the web-based 1080 Motion Cloud via the Windows control application (version 1.9.14.2, Lidingö, Sweden). MRDs of this type have demonstrated good criterion validity for sprint time measurements and good-to-excellent test–retest reliability for phase-specific outcome metrics in team sports athletes.^27–29 Application of resisted and assisted locomotion tasks using similar devices has been shown to acutely modify sprint performance and mechanical outputs.^30–32,34 Key settings were as follows: assisted load of 3 kg in both “concentric” and “eccentric” phases (identical for all athletes); mode “no fly weight” (NFW) with an upper limit of 14 m/s as the target maximum velocity benchmark; run start and direction change identified automatically using default system thresholds; and athletes wearing a waist belt attached at the hips level. Testing took place on indoor artificial turf. No cones or visual markers were placed in the deceleration zone beyond 30 m, in order not to predefine or constrain the athletes’ braking distance or strategy. Immediately after the 30 m sprint segment, athletes were instructed to perform a maximal-intent backpedal, allowing the system to correctly detect the change-of-direction manoeuvre.

During the ADA30 test, participants completed two maximal trials, rested for about 5 min, and received feedback on phase 1a peak sprint velocity after each trial. Participants were instructed to: “run 30 m in the shortest possible time while reaching the highest possible velocity before decelerating aggressively over the shortest possible distance to change direction and complete the test with a maximal-intent backpedal into the designated finish zone”. Only variables from phase 1a were included in the analysis, while phase 1b (backpedal distance) was excluded. For statistical analysis, the best trial for each participant was chosen based on the highest phase 1a average deceleration (peak sprint velocity/braking time) value to highlight the whole braking component. A complete list of variables, units, and operational definitions is provided in Supplementary Table S2. A graphical representation of the test is in Figure 1B.

Isometric belt squat (IBS)

Maximal isometric lower-body strength was assessed using a portable fixed dynamometer (TruStrength, Hawkin Dynamics, Westbrook, Maine, USA) sampling at 1200 Hz, with data sent in real time to the web-based Hawkin Dynamics Cloud via the Android Hawkin Capture application (version 9.10.1, Westbrook, Maine, USA). Similar fixed-dynamometer configurations with the TruStrength device have shown high reliability for lower-limb isometric strength testing, including the IBS.^40,41 Athletes wore a dedicated vest connected via a heavy chain to a metal base plate, positioning them in a half-squat stance; target knee flexion was ranged from 115–125 ° relative to full extension, verified using an analog goniometer during set-up. The dynamometer was zeroed before each effort, and athletes were required to establish a stable pre-tension level just below 300 N (system-defined onset threshold for net force), which served to standardise the starting position and achieve a more stable quiet phase prior to the maximal effort, with net values subsequently calculated relative to this threshold. Participants stood on a firm, level surface with their arms crossed over their chest or with their hands-on-hips to match CMJ posture. Each isometric trial lasted 5 s from the verbal cue, and participants performed two maximal effort trials separated by approximately 3 min of rest. They were instructed to: “generate as much force as possible in the shortest possible time”. Feedback in the form of peak force was displayed after each trial. Nine variables describing peak force and the IBS force–time characteristics were retained from the Hawkin Dynamics Cloud export. Relative variables (normalised to body mass) formed the basis of the main analysis. A complete list of metrics, units, and operational definitions is provided in Supplementary Table S3. For each variable, the mean of two trials was used in the analyses.^10,18–20 A graphical representation of the test is in Figure 1C.

Data processing and sign conventions

All data were exported to .csv files and subjected to analysis within the Python programming environment, leveraging libraries including: Pandas, NumPy, and SciPy. As part of the initial data cleaning, decimal points and non-breaking spaces were converted to standard decimal point notation, before all columns, except identifiers (e.g., athlete name to ID), were converted to numeric data. No missing or incorrect data were recorded. For the CMJ: braking phase velocity, power, and stiffness data were negative, with lower values representing a greater performance magnitude. To maintain biomechanical interpretation, more negative values were assumed to indicate a higher magnitude of the parameter in the deceleration phase (i.e., potentially greater deceleration capacity or higher stiffness as computed by the manufacturer's algorithm). For the deceleration metrics in the ADA30 test (phase 1a average deceleration and phase 1a peak deceleration), values were reported as positive magnitudes, with higher values interpreted as more effective deceleration. Additionally, all time metrics in the ADA30 test (phase 1a time, phase 1a 0–5 m time, phase 1a deceleration time) were interpreted in such a way that lower values indicated better performance (faster acceleration or shorter deceleration time), while a shorter deceleration distance (phase 1a deceleration distance) was considered advantageous, indicating more efficient deceleration over a shorter distance. The sign and direction conventions used were explicitly considered when interpreting the statistical results, particularly the signs of the correlation coefficients for the inhibition and time-related variables, to avoid erroneous conclusions resulting solely from the arbitrary way of reporting values (negative vs. positive; less is better vs. more is better).

Statistical analysis

The statistical analysis was developed to identify potential proxy metrics. Therefore, a broad set of dependent variables were retained, but subsequent interpretation focused on FDR-significant and mechanistically plausible relationships to minimise the risk of data-driven over-interpretation. Given the sample size (n = 24) and the expectation of non-normal distributions, Spearman's rank-order correlation (ρ) was used to quantify monotonic relationships between:

- CMJ and ADA30 variables (30 × 8 = 240 pairs),

In total, 582 cross-test correlations were calculated.

To address multiplicity while preserving reasonable sensitivity in this discovery-oriented analysis, the Benjamini–Hochberg procedure was applied across all 582 two-sided correlation tests to control the false discovery rate (FDR) at q = 0.10. This threshold was selected because the analysis was intended to identify candidate associations in a high-dimensional setting rather than to serve as a confirmatory test of a small number of prespecified hypotheses; in such contexts, FDR control is commonly used to balance false discoveries against loss of power. Benjamini–Hochberg-adjusted p-values (hereafter reported as q-values) are provided for all tests. Associations with q < 0.10 were considered statistically significant after FDR control and interpreted as candidate rather than definitive relationships. Associations with nominal p ≤ 0.05 that did not survive FDR correction were treated as exploratory only. For consistency with Table 1 reporting, 95% confidence intervals for Spearman's ρ were calculated using a Fisher z-transformation approximation (SE = 1/√(n−3)) and then back-transformed to the correlation scale. Effect magnitudes were interpreted descriptively using conventional thresholds (|ρ| < 0.10 trivial, 0.10–0.29 small, 0.30–0.49 moderate, 0.50–0.69 large, ≥0.70 very large). ⁴² As most force- and power-related outcomes were already expressed relative to body mass, additional body-mass-adjusted partial correlations were not performed in the primary analysis to avoid redundant adjustment. Absolute metrics were therefore interpreted with appropriate caution. When “lower is better” (temporal metrics) or when device-extracted signs were unintuitive (e.g., CMJ braking variables), correlation directions were interpreted using the sign conventions defined in the data-processing section. Correlation structures were visualised as half-matrix heat maps (upper triangle masked) for internal interpretation, and selected relationships were plotted as scatterplots. Intraday reliability was assessed across three CMJ trials in 24 athletes. This analysis was performed only for the CMJ, as it was the only test represented by the mean of at least three repeated trials. For each metric, absolute and relative reliability were quantified using CV% and two-way mixed-effects absolute-agreement single-measure ICC [ICC(A,1)], respectively, with 95% confidence intervals estimated by participant-level bootstrapping. SEM and MDC95 were also calculated, and analyses were restricted to complete athlete-level triplicates. Full reliability results are provided in Supplementary Table S5.

OPEN IN VIEWER

Table 1.

FDR-significant cross-test Spearman rank correlations (n = 24) between countermovement jump (CMJ), assisted 30 m acceleration–deceleration (ADA30; phase 1a metrics), and isometric belt squat (IBS) variables. Values are Spearman's ρ with 95% confidence intervals (95% CI), two-sided unadjusted p-values, Benjamini–Hochberg false discovery rate–adjusted q-values (q < 0.10), and qualitative magnitude descriptors.

Test 1	Variable 1	Test 2	Variable 2	Spearman ρ (95% CI)	Magnitude	p-value	q (FDR)
CMJ	Relative Propulsive Average Power	ADA30	Phase 1a Peak Sprint Velocity	0.807 [0.60; 0.91]	very large	< 0.0001	0.001
CMJ	Peak Relative Propulsive Power	ADA30	Phase 1a Peak Sprint Velocity	0.771 [0.53; 0.90]	very large	< 0.0001	0.003
CMJ	mRSI	ADA30	Phase 1a Peak Sprint Velocity	0.697 [0.41; 0.86]	large	0.0002	0.026
CMJ	Average Relative Propulsive Force	ADA30	Phase 1a Peak Sprint Velocity	0.692 [0.40; 0.86]	large	0.0002	0.026
CMJ	Peak Relative Braking Force	IBS	Relative Peak Force	0.663 [0.35; 0.84]	large	0.0004	0.042
CMJ	Braking Rate of Force Development	IBS	Relative Peak Force	0.662 [0.35; 0.84]	large	0.0004	0.042
CMJ	Relative Force at Minimal Displacement	IBS	Relative Peak Force	0.653 [0.34; 0.84]	large	0.0005	0.045
CMJ	Braking Rate of Force Development	IBS	Peak Force	0.636 [0.31; 0.83]	large	0.0008	0.060
CMJ	Average Relative Braking Force	IBS	Relative Peak Force	0.626 [0.30; 0.82]	large	0.0011	0.060
CMJ	Jump Height	ADA30	Phase 1a Peak Sprint Velocity	0.619 [0.29; 0.82]	large	0.0013	0.060
CMJ	Relative Propulsive Net Impulse	ADA30	Phase 1a Peak Sprint Velocity	0.616 [0.28; 0.82]	large	0.0014	0.060
CMJ	Take-off Velocity	ADA30	Phase 1a Peak Sprint Velocity	0.616 [0.28; 0.82]	large	0.0014	0.060
CMJ	Braking Phase Duration	IBS	Relative Peak Force	−0.613 [−0.81; −0.28]	large	0.0014	0.060
CMJ	Peak Velocity	ADA30	Phase 1a Peak Sprint Velocity	0.612 [0.28; 0.81]	large	0.0015	0.060
CMJ	Average Propulsive Velocity	ADA30	Phase 1a Peak Sprint Velocity	0.610 [0.27; 0.81]	large	0.0015	0.060
CMJ	Peak Relative Propulsive Force	IBS	Relative Peak Force	0.603 [0.26; 0.81]	large	0.0018	0.065
ADA30	Phase 1a 0–5 m Time	IBS	Peak Rate of Force Development	0.582 [0.23; 0.80]	large	0.0029	0.095
CMJ	Braking Phase Duration	ADA30	Phase 1a Peak Deceleration	−0.580 [−0.80; −0.23]	large	0.0029	0.095
CMJ	Peak Relative Braking Force	IBS	Peak Force	0.577 [0.23; 0.80]	large	0.0032	0.098
CMJ	Stiffness	ADA30	Phase 1a Peak Deceleration	−0.572 [−0.79; −0.22]	large	0.0035	0.100
ADA30	Phase 1a Peak Acceleration	IBS	Peak Rate of Force Development	−0.570 [−0.79; −0.22]	large	0.0036	0.100

Legend: Very large correlations (|ρ| ≥ 0.70) are shown in bold.

Use of AI tools

An AI-based writing assistant, Grammarly (web-based version, Pro license, at the time of analysis; Grammarly Inc., USA) was used to support linguistic revision, including grammar, spelling, stylistic, and consistency checks. Additionally, an AI-based large language model, ChatGPT (OpenAI, USA; version 5.5, Pro license), was employed to generate Python code for data processing. The generated scripts were executed within the same ChatGPT computational environment, and the resulting outputs were exported as comma-separated values (.csv) files for subsequent statistical analysis and figure preparation conducted by the authors.

CMJ propulsive parameters and assisted peak sprint velocity

A distinct cluster of large-to-very-large positive relationships was identified between CMJ propulsive phase variables and ADA30 phase 1a peak sprint velocity. Phase 1a peak sprint velocity showed very large positive correlations with CMJ average relative propulsive power (ρ = 0.807; q = 0.001) and CMJ peak relative propulsive power (ρ = 0.771; q = 0.003). Several other CMJ variables—mRSI, jump height, average relative propulsive force, take-off velocity, peak velocity, average propulsive velocity, and relative propulsive net impulse—also showed large positive correlations with phase 1a peak sprint velocity (ρ ≈ 0.61–0.70; q < 0.06; Table 1). A representative examples of this trend are presented in Figures 3 and 4.

OPEN IN VIEWER

Figure 3.

Scatterplot showing the relationship between CMJ average relative propulsive power and ADA30 phase 1a peak sprint velocity. A very large positive correlation is present (ρ = 0.807, q = 0.001), indicating that greater propulsive output in the CMJ is associated with higher assisted sprint velocity.

CMJ braking parameters, horizontal braking, and maximum force in the IBS

Significant relationships were observed between CMJ braking phase metrics and IBS isometric strength expression, as well as braking-related variables in the ADA30 task. Peak/average relative braking force, braking RFD, relative force at minimum displacement exhibited large positive correlations with IBS relative peak force (ρ ≈ 0.63–0.66; q < 0.06; Table 1). Braking phase duration was negatively correlated with both IBS relative peak force (ρ = –0.613; q = 0.060) and phase 1a maximal deceleration (ρ = –0.580; q = 0.095). CMJ stiffness was also negatively correlated with phase 1a maximal deceleration (ρ = –0.572; q = 0.100).

Isometric RFD and initial acceleration (0–5 m)

IBS peak RFD was associated with early phase 1a acceleration metrics. Specifically, peak RFD was positively correlated with phase 1a 0–5 m time (ρ = 0.582; q = 0.095, large) and negatively correlated with phase 1a peak acceleration (ρ = –0.570; q = 0.100, large; Table 1). Figure 5 illustrates these relationships.

All significant inter-test correlations with at least a large effect size (ρ ≥ 0.5) are additionally illustrated in Figure 5, providing a graphical complement to the data in Table 1.

OPEN IN VIEWER

Figure 4.

Scatterplot showing the relationship between CMJ peak relative propulsive power and ADA30 phase 1a peak sprint velocity. A very large positive correlation is present (ρ = 0.771, q = 0.003), indicating that greater propulsive output in the CMJ is associated with higher assisted sprint velocity.

OPEN IN VIEWER

Figure 5.

Scatter plots of all remaining large cross-test correlations between CMJ, ADA30 (phase 1a), and IBS variables.

CMJ propulsive metrics and assisted peak sprint velocity

The strongest and most coherent cluster of associations linked CMJ propulsive metrics to phase 1a peak sprint velocity under assisted conditions. Very large correlations were observed for average and peak relative propulsive power, and large correlations for mRSI, average relative propulsive force, jump height, take-off velocity, relative propulsive net impulse, peak velocity, and average propulsive velocity. Collectively, these relationships suggest that higher force-generating capacity in the vertical propulsive phase of the CMJ is associated with achieving the higher phase 1a peak sprint velocity under assisted conditions. Given that CMJ jump height is calculated from take-off velocity, and take-off velocity is determined by net vertical impulse, variables such as relative propulsive net impulse and take-off velocity largely represent different expressions of the same impulse–momentum construct. Therefore, reporting the entire continuum of impulse–momentum-related variables may offer limited additional interpretive value beyond jump height itself.

These effect sizes are comparable to, or slightly higher than, those reported between CMJ force–time metrics and maximal or near-maximal sprint speeds in footballers, sprinters, and rowing athletes.^7–10,43 Mechanistically, CMJ propulsive force and power may relate to sprint velocity because both tasks require rapid production of large vertical support forces; while horizontal force orientation remains critical during acceleration, the vertical component dominates the ground reaction force profile during upright high-speed sprinting and has been linked to maximal running speed. ⁴⁴

It is important to note that sprinting in this study was assisted, which alters absolute mechanical demands and segmental kinematics relative to unassisted sprinting.^30,32,34 Nevertheless, the strong CMJ–peak sprint velocity relationships suggest that CMJ propulsive metrics may be used as practical proxies for high-speed running potential in similar populations, especially where MRDs are unavailable or frequent maximal sprint testing is constrained by scheduling or injury risk considerations. However, these between-athlete correlations should not be taken to imply that CMJ and sprint measures are interchangeable or that CMJ can replace direct sprint testing; longitudinal data are still needed to establish whether within-athlete changes in CMJ meaningfully track changes in sprint performance over time. These findings should not be interpreted as direct evidence that CMJ propulsive metrics predict unassisted maximal sprint velocity, but they are consistent with existing literature and warrant further corroboration.

Vertical braking behaviour, horizontal deceleration, and isometric strength

A second important cluster of findings involved CMJ braking phase metrics, ADA30 horizontal deceleration, and IBS force expression. Braking phase duration and stiffness both showed large negative correlations with phase 1a peak deceleration. Given that higher peak deceleration represents greater instantaneous braking magnitude, and shorter braking phase duration and higher stiffness magnitude are generally interpreted as favourable, these results imply that athletes who decelerate more intensely in the horizontal plane tend to present a short, stiff braking phase during CMJ. This may represent a novel contribution of the present study, as it raises the possibility that braking capacities are at least partly unified across tasks despite differences in the predominant direction of force application.

This interpretation aligns with the conceptual “Braking Performance Framework”, which emphasises the ability to tolerate and generate high forces within short time windows during deceleration tasks.^11–13,33 Although authors did not directly measure ground reaction forces during horizontal deceleration, the relationships observed between CMJ braking characteristics and MRD-derived assisted deceleration outputs lend support to the idea that vertical jump testing can provide meaningful information about multi-faceted neuromuscular capacity.

From the authors’ perspective, this pattern contrasts with earlier work in mixed-sports university athletes, where CMJ propulsive variables were the primary discriminators between groups above versus below the median for horizontal deceleration ability, with jump-derived braking phase metrics playing a more limited role. These earlier findings may be, at least in part, population- or context-specific, reflecting the characteristics of a heterogeneous university sample with varied and often relatively limited exposure to systematic strength and conditioning. In many applied environments, enhanced propulsive capacities tend to differentiate athletes with substantial, long-term engagement in structured physical preparation from those with minimal training history, particularly in developmental cohorts drawn from multiple sports. ¹³ We suspect that higher training status—particularly within a professional strength and conditioning regime emphasising comprehensive eccentric training—may be associated with a more specific and intuitive coupling between vertical and horizontal braking capacities.

The relationships between CMJ braking phase metrics and IBS force production were particularly coherent. Relative peak force measured in IBS showed large positive correlations with several CMJ braking variables (peak/average relative braking force, braking RFD, and relative force at minimum displacement) and a large negative correlation with braking phase duration. These results are consistent with IMTP-based literature showing that higher maximal isometric force and RFD are associated with greater CMJ braking and propulsive outputs and better locomotive performance.^{10,18–20,22,23}

From a practical standpoint, these findings suggest that IBS testing in a belt squat configuration may serve as a viable alternative to IMTP for profiling lower-limb force capacities that underpin vertical braking behaviour, and by extension, certain aspects of horizontal deceleration ability. Additionally, IBS has several applied advantages: grip is not limiting, set-up is relatively rapid, the position aligns closely with frequently prescribed squat variations, and loading is largely localised to the lower-body.

Isometric explosiveness and early assisted acceleration

The relationships between IBS peak RFD and early phase 1a acceleration metrics (0–5 m time and peak acceleration) were statistically large but directionally counterintuitive. Higher IBS peak RFD was associated with slightly longer (worse) 0–5 m time and lower measured peak acceleration under assisted conditions.

Several non-mutually exclusive explanations may be considered. First, under assisted conditions, a portion of horizontal acceleration is generated externally by the device. Athletes with greater isometric RFD may not need, or may not be able, to exploit their capacity in the same way as during unassisted or resisted sprinting.^32,34 Second, athletes may adopt more conservative movement strategies to maintain stability at high assisted speeds, prioritising control over maximal horizontal push-off in the initial steps. Finally, with only twenty-four players, correlation estimates are susceptible to sampling variability, and some associations–especially those that are counterintuitive–may be specific to this cohort or protocol, as has been noted previously in small-sample strength-performance work.^18,19

Overall, we view these RFD–acceleration findings as hypothesis-generating rather than definitive. They emphasise that isometric RFD measured in a belt squat position should not be assumed to directly reflect early horizontal acceleration capacity in an assisted sprinting context.

Practical implications for American football and related team sports

For practitioners working with American football skill/running positions, and potentially athletes in comparable roles in different team sports, these findings offer several tentative practical implications. CMJ propulsive metrics–particularly relative propulsive force, power, impulse, mRSI, jump height, and velocity–appear to provide useful but indirect proxy information about an athlete's ability to achieve high assisted running velocities. Monitoring these variables over time may therefore yield insights into changes in high-speed running potential, especially when direct sprint testing is constrained.^7–9 CMJ braking phase characteristics—including braking phase duration, relative braking force, force at minimal displacement, stiffness, and braking RFD—appear to complement IBS force expression as potential indicators of braking-oriented neuromuscular capacity. Accordingly, athletes who display shorter braking phases, stiffer countermovement, and higher relative IBS peak force may possess a neuromuscular profile better suited to the intense deceleration demands of multidirectional field sports.^11–13,33 At the same time, the complex and occasionally counterintuitive associations–particularly involving IBS peak RFD and early assisted acceleration–underline that no single measure can fully capture task-specific locomotive performance. Proxy metrics should therefore be interpreted within a broader framework that includes technical observation, positional role, fatigue state and tactical context. ¹⁵

Taken together, the present results support the concept of an integrated yet relatively lean monitoring battery combining: (1) vertical jumps, (2) assisted acceleration-deceleration tasks, and (3) simple, multi-joint isometric strength tests. These tools should be viewed as complementary sources of information rather than definitive stand-alone indicators. In this regard, proxy metrics remain simplifications of complex behaviours: their interpretation must always account for test context, technical execution and individual athlete characteristics, and should complement–rather than replace–direct assessments and broader performance monitoring. Practitioners are therefore encouraged to test their athletes across different environments and tasks to obtain the most robust, training-relevant insights. At present, the evidence base around proxy metrics is still at too early stage for coaches and practitioners to rely on them in isolation when making high-stakes programming decisions.