Resistance training (RT) is a well-established method of enhancing athletic capabilities and improving muscle size, strength, power, and overall physical fitness.1,2 The physiological adaptations depend largely on the correct manipulation of training variables, such as intensity (ie, load lifted), and level of effort (ie, number of repetitions completed with respect to the maximum number of repetitions that can be performed to failure).3,4 Consequently, repetition maximum (RM) targets are one of the most popular RT prescription methods, as it provides valuable information to regulate both exercise intensity and level of effort.5 This methodology requires assigning the load that enables a certain number of repetitions to be completed before reaching muscular failure (eg, 10RM is the load that enables one to complete a maximum of 10 repetitions), and then athletes are instructed to perform a specific number of repetitions based on the desired proximity to failure.5 Some of the primary limitations of this method for updating the RM loads are: (1) conventional methods are based on frequent execution of sets to failure and (2) the use of subjective measures of repetitions in reserve is needed. To overcome these limitations, some studies have examined the possibility of using lifting velocity to predict the maximum number of repetitions that can be performed to failure (RTF).6–10
Recent research has demonstrated that the fastest mean velocity (MVfastest) and fastest peak velocity (PVfastest) within a set may be able to predict different RTFs.6–10 When modeling data involving multiple loads, regardless of considering the data from a single subject (individualized RTF–velocity relationship) or across different subjects (general RTF–velocity relationship), a positive association has been found between RTF and the fastest set velocity for upper- and lower-body exercises.6–10 These relationships can be easily constructed by applying a linear regression model to RTF and the fastest repetition velocity within a set (MVfastest or PVfastest) recorded under at least 2 distinct loads.6–10 Once this relationship is constructed, coaches can prescribe in subsequent training sessions the MVfastest or PVfastest corresponding to the desired RTF. Although this approach allows athletes and coaches to objectively predict the RTF using a noninvasive method (ie, recording lifting velocity), no previous systematic review has analyzed the different methodological factors that could influence its accuracy under field conditions. Therefore, the specific objectives of the present systematic review are 2-fold: (1) to determine the basic properties of the RTF–velocity relationship, such as goodness-of-fit, reliability, and accuracy (ie, error in RTF prediction); and (2) to offer guidance on implementing various methodological factors that can impact the accuracy of RTF prediction, including the magnitude of loads lifted, the number of loads, and the specific lifting velocity variable considered.
Methods
Search Strategy
The systematic research was performed by 2 authors (Miras-Moreno and García-Ramos) based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.11 The research was searched on the following academic databases: PubMed, SPORTDiscus and Scopus, gathering data to the last updating of March 27, 2024. The following terms were included into the search: (“resistance train*” OR “resistance exercise*” OR “strength train*” OR “strength exercise*” OR “weight train*” OR “weight lift*”) AND (“monitor* velocity” OR “lift* velocity” OR “record* velocity” OR “barbell velocity”) AND (“repetition* to failure” OR “maxim* repetition*” OR “musc* failure”). The studies were identified by searching “titles, abstracts, and key words,” and the results were then imported into a reference manager (www.rayyan.ai).12 The systematic review title, objectives, search strategy, eligibility criteria, and risk of bias assessment was registered on January 30, 2024, on the Open Science Framework (osf.io/trkb4; Figure 1).
—PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram detailing the inclusion and exclusion of the records screened.
Citation: International Journal of Sports Physiology and Performance 20, 3; 10.1123/ijspp.2024-0337
Eligibility and Data Extraction
Only original articles were included when the following criteria were met: (1) full-text published in English, (2) published in the selected academic databases, (3) human data, (4) at least 2 sets at different loads were performed to failure within the same testing session, (5) the maximum number of repetitions and the repetitions with the fastest velocity of the set were monitored and reported, (6) the procedures were performed without a current experimental condition (except for under controlled fatigue states) that may enhance or compromise the performance (eg, nutritional supplements or blood flow restriction), and (7) only multijoint weight-lifting exercises (eg, traditional fixed-load RT exercises). Cross-sectional and exercise training studies were eligible, but only preintervention tests were used for training studies. Additionally, a backward screening (ie, reference lists from eligible full-texts) and forward screening (ie, citations from eligible full-texts) were performed manually. In the case of the identification of relevant papers, the same abovementioned inclusion criteria were performed. Disagreements were solve establishing a consensus between the reviewers or when necessary, through an additional reviewer (Pérez-Castilla).
The following descriptive data were extracted from eligible studies: title, participants characteristics (ie, sex, 1RM strength and relative to body mass), RT exercise, number of sets, loads, resting time between sets, lifting velocity variables (ie, MVfastest and PVfastest), type of relationships (ie, generalized and individualized), modeling procedure used (ie, multiple- or 2-point methods), and type of muscle failure reached (ie, volitional or momentary; Table 1). Regarding the basic properties of the RTF–velocity relationships, the following data were extracted: goodness-of-fit, reliability values of the velocity values associated to each RTF, absolute number of repetition error, and sensitivity to fatigue protocol (ie, number of sets, loads and resting time; Table 2). In the case of unreported data, one author (Miras-Moreno) emailed the manuscript’s corresponding author to request the raw or mean values (RTF and fastest velocity within a set). Extracted data were assessed for accuracy by a third reviewer (Pérez-Castilla).
Summary of Descriptive Values Derived From the Studies Included Exploring the RTF–Velocity in This Systematic Review
| Study | Participants | Exercise(s), load(s), and resting time | Lifting velocity variables | Type of relationships | Modeling procedures | Type of muscular failure |
|---|---|---|---|---|---|---|
| García-Ramos et al9 | 21 males—1RM: 84.0 (17.7) kg; 1.15 (0.21) kg/kg BM | SM BP 60–70–80–90% 1RM (random) 10 min between sets | MV | Generalized and individualized | Multiple-point | Momentary failure |
| Miras-Moreno et al8 | 20 males—1RM: 78.8 (12.0) kg; 1.02 (0.15) kg/kg BM 3 females—1RM: 44.4 (4.0) kg; 0.71 (0.06) kg/kg BM | SM PBP 60–70–80–90% 1RM (random) 10 min between sets | MV PV | Generalized and individualized | Multiple-point | Momentary failure |
| Jukic et al6 | 31 males—1RM: 149.3 (23.6) kg; BM: 85.05 (13.71 kg) 15 females—1RM: 83.2 (19.9) kg; BM: 67.45 (8.25) kg | FW BS 90–80–70% 1RM (fixed) 10 min between sets | MV | Generalized and individualized | Multiple-point | Momentary failure |
| Miras-Moreno et al7 | 23 males—1RM: 84.8 (12.9) kg; 1.06 (0.17) kg/kg BM | FW PBP 60–90–70–80% 1RM (fixed) 5 min between sets | MV | Generalized and individualized | Multiple-point 2-point | Momentary failure |
| Miras-Moreno et al10 | 20 males—1RM: 88.9 (11.0) kg; 1.10 (0.17) kg/kg BM | SM PBP with straps 60–80–70% 1RM (fixed) 5 min between sets | MV PV | Generalized and individualized | Multiple-point 2-point | Momentary failure |
| Janicijevic et al15 | 30 males high-level wrestlers—1RM: 164.8 (25.5) kg; BM: 79.90 (7.90) kg | FW BS 90–80–70% 1RM (fixed) 5 min between sets | MV | Generalized and individualized | Multiple-point | Momentary failure |
Abbreviations: BM, body mass; BP, bench press; BS, back squat; FW, free-weight; MV, mean velocity; PBP, prone bench pull; PV, peak velocity; 1RM, 1-repetition maximum; RTF, repetitions to failure; SM, Smith machine.
Summary of the Basic Properties of the RTF–Velocity Relationships From Studies Included in This Systematic Review
| Study | Goodness-of-fit | Between-sessions reliability | Accuracy (absolute repetitions errors) | Sensitivity to fatigue protocol |
|---|---|---|---|---|
| García-Ramos et al9 | Generalized (MV): r2 = .77; SEE = 3.5 repetitions Individualized (MV): r2 = .98 (.86–.99) | Within-subject CV (MV) = 6.2% (4.7–7.1) ICC (MV) = .88 (.87–.93) | ||
| Miras-Moreno et al8 | Generalized (MV): r2 = .70; SEE = 3.6 repetitions Individualized (MV): r2 = .96 (.83–1.00) SEE = 1.7 (0.3–4.7) repetitions Generalized (PV): r2 = .67; SEE = 3.7 repetitions Individualized (PV): r2 = .97 (.84–1.00) SEE = 1.4 (0.2–4.7) repetitions | Within-subject CV (MV) = 3.8% (3.5–5.4) ICC (MV) = .82 (.78–.83) Within-subject CV (PV) = 3.5% (3.0–6.7) ICC (PV) = .88 (.87–.93) | Generalized vs multiple-point Set 1 (MV): 3.4 (2.4) vs 2.3 (1.5) Set 2 (MV): 4.3 (2.3) vs 2.9 (1.8) Set 3 (MV): 4.0 (2.2) vs 2.6 (1.5) Set 4 (MV): 4.1 (2.4) vs 2.7 (1.5) Set 1 (PV): 2.8 (2.4) vs 2.1 (1.0) Set 2 (PV): 3.8 (1.7) vs 2.5 (1.5) Set 3 (PV): 3.6 (1.6) vs 2.2 (1.4) Set 4 (PV): 3.4 (1.7) vs 2.4 (1.3) | 4 sets 75% 1RM 2-min interset rest |
| Jukic et al6 | Generalized (MV): r2 = .49; SEE = 3.6 repetitions Individualized (MV): r2 = .98 (.50–1.00) SEE = 1.1 (0.1–4.2) repetitions | Within-subject CV (MV) = 7.2% (6.1–10.4)a ICC (MV) = .75 (.46–.78)a | Generalized vs multiple-point Set 1 90% 1RM (MV)a: 2.1 (1.6) vs 1.7 (1.9) Set 2 80% 1RM (MV)a: 2.5 (2.1) vs 1.8 (1.5) Set 3 70% 1RM (MV)a: 3.8 (2.9) vs 2.3 (1.7) | 3 sets 90–80–70% 1RM 10-min interset rest |
| Miras-Moreno et al7 | Generalized (MV): r2 = .67; SEE = 6.6 repetitions Individualized (MV): r2 = .94 (.79–1.00) SEE = 3.0 (0.5–9.5) repetitions | Generalized vs multiple-point vs 2-point Set 1 65% 1RM (MV): 5.1 (4.0) vs 5.7 (5.7) vs 6.4 (6.1) Set 2 65% 1RM (MV): 4.6 (3.4) vs 5.0 (4.1) vs 6.7 (4.9) Set 3 85% 1RM (MV): 2.8 (2.3) vs 2.4 (2.1) vs 3.0 (2.5) Set 4 85% 1RM (MV): 3.3 (2.2) vs 3.0 (1.8) vs 3.7 (2.6) | 4 sets random sequence 65–85% 1RM 2-min interset rest | |
| Miras-Moreno et al10 | Generalized (MV): r2 = .57; SEE = 7.5 repetitions Individualized (MV): r2 = .95 (.87–.99) SEE = 2.7 (0.1–9.6) repetitions Generalized (PV): r2 = .66; SEE = 6.7 repetitions Individualized (PV): r2 = .97 (.88–1.00) SEE = 2.3 (0.1–10.3) repetitions | |||
| Janicijevic et al15 | Generalized (MV): r2 = .84; SEE = 1.9 repetitionsa Individualized (MV): r2 = .97 (.88–1.00) SEE = 1.3 (0.1–2.8) repetitionsa | Within-subject CV (MV) = 4.6% (4.0–8.0) ICC (MV) = .48 (.46–.53) | Generalized vs multiple-point Set 1 (MV): 1.4 (1.0) vs 1.3 (1.0) Set 2 (MV): 1.1 (0.8) vs 1.3 (1.0) Set 3 (MV): 1.6 (1.0) vs 1.7 (1.7) Set 4 (MV): 1.3 (1.3) vs 1.8 (1.7) | 4 sets 75% 1RM 2-min interset rest |
Abbreviations: CV, coefficient of variation; FW, free-weight; ICC, intraclass correlation coefficient model 3.1; MV, average velocity from the first positive velocity until the velocity was 0 m·s−1; PV, maximum velocity value recorded during the concentric lifting phase; r2, Pearson multivariate coefficient of determination; 1RM, 1-repetition maximum; RTF, repetitions to failure; SEE, standard error of the estimate; SM, Smith machine. Note: Generalized, pooling together the data from all subjects; Multiple-point, more than 2 experimental points included into the modeling procedure; 2-point, only 2 distant experimental points are included into the modeling procedure.
aA posteriori analyses performed not related to the original study’s aims.
Assessment of Reporting Quality
Two authors (Miras-Moreno and García-Ramos) graded the quality studies based on the modified version of the Downs and Black checklist (Supplementary Material S1 [available online]).13 Nonetheless, not all the assessment criteria were applicable to the studies from this review (only 17 of the 27 binary items [“0”: unable to determine and “1”: yes] were used). Differences were solved by reaching a consensus between the authors or when required, through the involvement of an additional author (Pérez-Castilla). Of note, this modified checklist has been previously used in systematic reviews from sports science.14
Results
Identification of Studies
The systematic search yielded 1661 studies, with no manuscripts identified by reviewing reference lists. From the manuscripts identified, 349 were removed as duplicates. During the time that involved the study analyses, 2 additional eligible studies that emerged via PubMed alerts were included. The first screening involved the analyses of the titles and abstract of the remaining 1312 studies with only 12 studies sought for full-text screening. During the full-text review, 8 studies were deemed to meet the inclusion criteria but, only 6 studies specifically analyzed the RTF–velocity relationships.6–10,15 The remaining studies focused on alternative methods to prescribe the proximity-to-failure through estimating the RTF and repetitions in reserve using the velocity loss16,17 (Figure 1).
Quality of Research Reporting
The research quality of the reported studies investigating the RTF–velocity relationships was found to be nearly perfect 15.0 (0) (mean [SD]) based on the Downs and Black checklist (Supplementary Material S2 [available online]). However, all the studies consistently did not include the item 17 related to the calculation of statistical power and, item 18 related to the statistical analyses used, since the assumption of independence was violated during the construction of generalized RTF–velocity relationships (ie, multiple data points from same participants are included in the equations and r2 can be inflated).
Study Characteristics
From all the participants involved in the included 6 studies (n = 145), females represented only 12.4% (n = 18).6–10,15 The most investigated RT exercise was the prone bench pull (PBP),7,8,10 followed by back squat (BS)6,15 and bench press (BP).9 The RTF–velocity relationships were constructed in the Smith machine8,10 and free-weight exercises.6,15
All the studies used the MVfastest to construct the generalized (ie, pooling together the data from all the participants) and individualized (ie, data obtained for each participant) RTF–velocity relationships6–10,15 and only 2 studies used the PVfastest.8,10 The multiple-point modeling procedure (ie, from 3 to 4 sets to failure ranging from 60% 1RM to 90% 1RM)6–10,15 was used in all these studies, whereas the 2-point method (ie, performing sets to failure against only 2 loads) was applied in 2 studies7,10 (Table 1).
Statistical Analyses Included From Selected Studies
The goodness-of-fit from RTF–velocity relationships were assessed through the median value and range from Pearson multivariate coefficient of determination (r2) and the standard error of estimate.6–10,15 The between-session reliability of the fastest velocity within a set associated to each RTF (from 1 to 15 RTFs) was assessed using the within-subjects coefficient of variation (CV; standard error of measurement/subjects’ mean score × 100) and the intraclass correlation coefficient (model 3.1).8,9,15 A high and acceptable reliability was considered when the CV was lower than 5% and 10%, respectively.8,9,15 The prediction accuracy was computed as the absolute and raw errors obtained when estimating the RTF and was calculated as follows: RTF estimation error = RTF performed – RTF predicted.6–8,15
Basic Properties of the RTF–Velocity Relationships
Goodness of Fit
All the included studies investigated the goodness-of-fit of RTF–velocity relationships through the construction of generalized (ie, pooling together the data from all subjects) and individualized (ie, separately for each subject) equations. Generalized RTF–velocity relationships always revealed a lower goodness-of-fit (r2 = .49–.84) compared with individualized RTF–velocity relationships (r2 = .94–.98) for the included RT exercises (Table 2). Moreover, the goodness-of-fit was comparable for PVfastest and MVfastest during the construction of the individualized RTF–velocity relationships for Smith and free-weight PBP exercises.8,10 The use of different interset rest time (from 5 to 10 min) did not show differences in the goodness-of-fit of RTF–velocity relationships for PBP nor BS exercises.6–8,15 The goodness-of-fit of the generalized RTF–velocity relationships in the BS exercise was superior for high-level male wrestlers (r2 = .84)15 compared with recreationally trained males and females (Figure 2).6
—Relationship between the maximum number of repetitions performed before reaching muscular failure and the fastest mean velocity of the set during the bench press, prone bench-pull exercise, and back-squat exercise obtained from all the data of the included studies. N indicates numbers of trials included in the regression analysis; r2, Pearson multivariate coefficient of determination; SEE, standard error of the estimate.
Citation: International Journal of Sports Physiology and Performance 20, 3; 10.1123/ijspp.2024-0337
Between-Sessions Reliability
The velocity associated with each RTF showed an acceptable absolute and relative between-sessions reliability from 1 to 15 RTFs during free-weight BP and BS, whereas for the Smith machine PBP exercise was found to be high (Table 2).6,8,9,15 The MVfastest and PVfastest values revealed comparable absolute and relative between-sessions reliability during the PBP exercise.8 The resting time between sets (ranging from 5 to 10 min) did not impact the between-sessions reliability of the velocity values associated for a given RTF (eg, CV = 4.6% vs 7.2% during the BS exercise, respectively).6,15
Accuracy in the RTF Prediction
The prediction accuracy was higher for individualized compared with generalized RTF–velocity relationships for all the RT exercises analyzed6–8 with the only exception of the study from Janicijevic et al,15 who found a comparable prediction accuracy for both equations in high-level wrestlers (Table 2).
The increment of the load was accompanied by a higher prediction accuracy during PBP exercise (absolute errors: set 1 at 65% 1RM = 5.7 [5.7] repetitions; set 1 at 85% 1RM = 3.0 [1.8] repetitions) and BS exercise (absolute errors: 70% 1RM = 2.3 [1.7] repetitions; 80% 1RM = 1.8 [1.5] repetitions; 90% 1RM = 1.7 [1.9] repetitions).6,7
The increment of fatigue (ie, from 10 to 2 min of interset rest) was accompanied by a lower prediction accuracy for PBP exercise (absolute errors: set 1 at 75% 1RM = 2.3 [1.5] repetitions; set 4 at 75% 1RM = 2.7 [1.5] repetitions) and for BS exercise (absolute errors: set 1 at 75% 1RM = 1.3 [1.0] repetitions; set 4 at 75% 1RM = 1.8 [1.7] repetitions).8,15 Additionally, the number of loads used for modeling the RTF–velocity relationships during the PBP exercise could affect the prediction accuracy with the multiple-point method (absolute errors: set 1 at 85% 1RM = 2.4 [2.1] repetitions; set 2 at 85%1 RM = 1.8 [3.7] repetitions) revealing a greater accuracy than the 2-point method (absolute errors: set 1 at 85% 1RM = 3.0 [2.5] repetitions; set 2 at 85% 1RM = 3.7 [2.6] repetitions).7
Discussion
The main findings of this systematic review are that regardless of the equipment used (ie, Smith vs free-weight exercises), lifting velocity variables (ie, MVfastest vs PVfastest), magnitude of the loads (ie, from 60% 1RM to 90% 1RM), number of the sets (ie, from 3 to 4 sets) and rest time provided (ie, from 5 to 10 min), the construction of the individualized RTF–velocity relationships present: (1) an excellent goodness-of-fit, (2) acceptable to high between-session reliability for the velocity values were associated fora given RTF (from 1 to 15 RTFs), (3) an acceptable prediction accuracy (∼2 repetitions errors) during fatigue-free sets (ie, 10 min of resting time), but (4) unacceptable estimation errors under fatigue conditions for inexperienced subjects. These results suggest that individualized RTF–velocity relationships are a valuable tool for practitioners using RM zones for training prescription and that the RTF–velocity relationship should be constructed under fatigue conditions (ie, interset rest) similar to those that will be experienced by the subject during actual RT sessions. Finally, further studies should explore how different levels of fatigue affect the prediction accuracy of RTF when: (1) different modeling procedures are used (ie, multiple- vs 2-point), (2) different loads (eg, 60–90% 1RM) are intended to predict (ie, the increment of load allows for lower RTF and consequently, it cannot be used the same absolute errors criterion accuracy for different loads), and (3) athletes with different RT experience are involved.
During the construction of the load–velocity relationships, it is advisable to use linear relationships rather than second-order polynomial regression models due to its greater simplicity and reliability.18,19 Furthermore, research indicates that there is less within-subject than between-subjects variability for the velocity corresponding to specific submaximal %1RMs.18,19 This finding advocates for the construction of individual rather than general load–velocity relationships.18,19 In agreement with previous velocity-based training literature, our systematic review confirms that individualized RTF–velocity relationships are linear (from 1 to 30 RTFs) and also stronger (ie, high r2 and low standard error of estimate) when compared to generalized regression models during the BP, PBP, and BS exercises6–10,15 Essentially, this indicates that there is a lower within-subject variability for the velocity corresponding to different RTFs compared to the between-subjects variability. Moreover, considering that RTF–velocity relationships vary with different variants of the same RT exercise (eg, differences between free-weight and Smith machine PBP exercise were found), it is imperative for coaches to construct the RTF–velocity relationships tailored to each specific RT exercise.7,8,10
In all the included studies, the individualized RTF–velocity relationships were constructed when multiple loads were lifted to failure (eg, 60% 1RM, 70% 1RM, 80% 1RM, and 90% 1RM referred to as “multiple-point method”).6–10,15 Nevertheless, the multiple-point method can be quite time-intensive and may lead to substantial fatigue, potentially affecting training adaptations or impairing performance in subsequent sessions. Based on the high linearity of individualized RTF–velocity relationships, Miras-Moreno et al7,10 recently proposed to simplify the RTF–velocity construction by lifting only 2 distant loads (eg, 60% 1RM and 90% 1RM, referred to as the “2-point method”). However, the multiple-point method revealed a lower RTF for a given lifting velocity compared with the 2-point method (eg, multiple-point: 1 RTF = 0.57 [0.06], while 2-point: 3 RTF = 0.57 [0.05], respectively) during the PBP exercise.7,10 These differences may be attributed to the less intensive procedure of the 2-point method allowing more RTF for a given velocity (ie, since fatigue affects more the number of RTF than velocity, a lower slope was found for the multiple-point method).7,10 However, these results should be taken with caution, since the 2-point method has been only explored during the PBP exercise and the optimal experimental points for its construction are still unknown.20,21
A critical methodological consideration for modeling the load–velocity relationship is choosing the best lifting velocity variable.19,22 In fact, the selection of the most common lifting velocity variables (ie, MV, average velocity from the beginning of the concentric phase to the point where the bar reaches its highest elevation; mean propulsive velocity [MPV average velocity from the beginning of the concentric phase until the barbell’s acceleration falls below gravity [−9.81 m·s2]; and PV attained at any point during the concentric phase]) directly impact the linearity and goodness-of-fit of the load–velocity relationships.19,22 From these variables, the MV is generally preferred due to greater linearity of the load–velocity relationships and between-session reliability for the velocities associated for a given %1RM.19,22 In contrast, 3 studies reported a comparable goodness-of-fit and between-session reliability of the RTF–velocity relationships using both MVfastest and PVfastest during the PBP exercise7,8,10 Nonetheless, unlike the load–velocity relationships, this issue has been less explored with RTF–velocity relationships and further studies should consider incorporating other lifting variables, such as MPV, during other RT exercises as this will help establish the best velocity output to monitor.
Obtaining metrics for monitoring an athlete’s training status with a high degree of reliability is crucial.23 Between-session reliability provides information on the consistency of individual scores (ie, absolute reliability, typically measured as within-subject CV) and the stability of an individual’s position within a group (ie, relative reliability, typically measured as intraclass correlation coefficient).23,24 The between-session reliability for the velocity values associated with each RTF was acceptable for BP and BS exercises but, high for the PBP exercise when the multiple-point method was used.6,8,9,15 However, unlike the load–velocity relationships, no previous study has explored whether the 2-point method could be a more reliable modeling procedure than the multiple-point method.21,25,26 Additionally, the long-term variability of the RTF–velocity relationships remains uncertain, which could potentially influence coaching decisions (eg, when an athlete’s RTF–velocity equation needs to be reassessed). Even more importantly, it is still unknown how different RT programs (eg, high volume [repetitions] or high power [maximal concentric velocity] training) may influence the RTF–velocity relationship. For example, previous research has demonstrated that a 4-week power training program led to a greater MV increase across the full load–velocity spectrum compared to maximal strength-oriented RT.27
The most significant factors that impact the accuracy of 1RM predictions derived from load–velocity relationships which may also affect the RTF–velocity relationships, include19: (1) RT exercise selection, (2) which velocity variable is selected, (3) the regression model applied, and (4) number of experimental points used. However, since the RTF are dependent of the load lifted, it is difficult to suggest when the prediction accuracy would be acceptable (eg, 2 repetitions errors are less problematic for 60% 1RM than 80% 1RM). Nonetheless, it is reasonable to suggest that as the load–velocity relationships, a lower degree of freedom of movement would allow a higher RTF prediction accuracy.19 However, this systematic review found comparable absolute errors for PBP and BS exercises (2.3 [1.5] vs 2.3 [1.7], respectively) against approximately the same relative load (∼70% 1RM) and fatigue experienced (accuracy obtained from the first set).6,8 In fact, it seems that a subject’s RT experience plays a more crucial role than the degrees of freedom of the movement, not only during low-fatigue conditions, but also during high levels of fatigue (eg, similar absolute repetitions errors comparing the 1 set: 1.3 [1.0] and 4 set: 1.8 [1.7] when only 2 minutes of interset rest was implemented during BS exercise for high-level wrestlers).15
Regarding the use of different lifting velocity variables, both MVfastest and PVfastest would allow similar accuracy predictions during the PBP but, it is plausible to suggest that in another RT exercise, where different velocity–time pattern occur (eg, the PVfastest is obtained earlier during the PBP compared to BP exercise28), results may differ. As expected, incorporating a differing number of experimental points into the modeling procedure (ie, multiple-point vs 2-point methods) would affect the RTF prediction accuracy. This issue can be attributed to the fact that increasing the number of sets into the modeling procedure affects the RTF–velocity relationships explained by the fact that fatigue impacts the number of RTF more than MVfastest.7,8 The unique study that directly compared both methods found that the 2-point method exhibited higher prediction errors compared to the multiple-point method.7 However, these results should be taken with caution and further studies should specifically explore the accuracy of different methods together with different levels of fatigue.
Although this is the first systematic review to outline the fundamental aspects of the RTF–velocity relationships, it is important to recognize several limitations and suggest directions for future research. First, due to the small number of studies that have investigated the RTF–velocity relationships these results can be only extrapolated to BP, PBP, and BS exercise. Second, the short-term reliability has been assessed within the same week whereas the long-term reliability and, how different RT programs can manipulate the RTF–velocity relationship, is unknown. Third, it remains uncertain under which fatigue conditions the use of either the multiple- or 2-point method is advisable, as previous studies have not investigated this matter. Fourth, it is not clear which lifting variable provides the highest goodness-of-fit, reliability, and prediction accuracy during different RT exercises. Fifth, other factors, such interset rest time (eg, which is the most optimal interset resting time for the construction of the relationship29), and feedback (eg, usually a high frequency of feedback is recommended to increase the performance on each repetition30) may influence the RTF–velocity relationships.
Conclusions
The findings of this systematic review indicate that individualized RTF–velocity relationships demonstrate a higher goodness of fit and more accurate RTF predictions compared to generalized models. These individualized relationships also show a range from acceptable to high between-sessions reliability for velocity values associated with specific RTFs (from 1 to 15 RTFs). Although the accuracy of RTF–velocity relationships under fatigue-free conditions (eg, first set prediction or apply >10 min of interset rest) is generally acceptable, it is significantly compromised by varying levels of fatigue during the training sessions aimed to predict (probably explained by those studies that found a greater reduction in RTF as fatigue accumulated, while the MVfastest values showed a smaller decline).8,10 However, it is important to note that prediction errors due to fatigue may be minimized when assessing athletes with extensive RT experience. Additionally, the basic properties of the RTF–velocity relationships seem to be unaffected using different equipment (Smith vs free-weight), lifting velocity variables (MVfastest vs PVfastest), magnitude of the loads analyzed (from 60% 1RM to 90% 1RM), number of sets (from 3 to 4 sets), and resting time (from 5 to 10 min) used for the equation’s construction. Finally, given that fatigue can impact the accuracy of RTF–velocity predictions, it is recommended to select a modeling procedure that best aligns with the specific fatigue conditions intended to be predicted.
Practical Applications
The construction of the individualized RTF–velocity relationships can be efficiently determined using a simple linear regression model by executing sets to failure with varying loads (from 2 to 3 sets). This approach requires the monitoring of 2 variables for the modeling: (1) RTF for each set and (2) MVfastest or PVfastest within each set. Once established, coaches simply need to measure the MVfastest or PVfastest against a given load (typically occurring in the first 1–3 repetitions). Then, this velocity can be inserted into the individualized equation for obtaining the RTF prediction in real-time (see Figures 3 and 4).
—Illustration of an Excel spreadsheet that can be used to estimate the maximum RTF and velocities associated with different RTF through 2 simple steps: (1) monitoring the RTF and maximum velocity within a set against at least 2 different external loads, and (2) once the individual RTF–velocity is constructed, we have to indicate the velocity obtained to predict different RTF for a given absolute (Supplementary Material S3 [available online]). RTF indicates repetitions to failure.
Citation: International Journal of Sports Physiology and Performance 20, 3; 10.1123/ijspp.2024-0337
—Basic properties of the RTF–velocity relationships and 4 steps for predicting different RTFs. RTF indicates repetitions to failure.
Citation: International Journal of Sports Physiology and Performance 20, 3; 10.1123/ijspp.2024-0337
Acknowledgments
This study is part of a PhD thesis conducted in Biomedicine Doctoral Studies of the University of Granada, Spain. This study was supported by the Spanish Ministry of University under a predoctoral grant (FPU19/01137) awarded to Miras-Moreno.
References
- 1.↑
Crewther B, Cronin J, Keogh J. Possible stimuli for strength and power adaptation: acute mechanical responses. Sports Med. 2005;35(11):967–989. doi:
- 2.↑
Schoenfeld BJ, Peterson MD, Ogborn D, Contreras B, Sonmez GT. Effects of low- vs. High-load resistance training on muscle strength and hypertrophy in well-trained men. J Strength Cond Res. 2015;29(10):2954–2963. doi:
- 3.↑
Kraemer WJ, Ratamess NA. Fundamentals of resistance training: progression and exercise prescription. Med Sci Sports Exerc. 2004;36(4):674–688. doi:
- 4.↑
Bird SP, Tarpenning KM, Marino FE. Designing resistance training programmes to enhance muscular fitness a review of the acute programme variables. Sports Med. 2005;35(10):841–851. doi:
- 5.↑
Thompson SW, Rogerson D, Ruddock A, Barnes A. The effectiveness of two methods of prescribing load on maximal strength development: a systematic review. Sports Med. 2020;50(5):919–938. doi:
- 6.↑
Jukic I, Helms ER, McGuigan MR. The fastest repetition in a set predicts the number of repetitions completed to failure during resistance training: the impact of individual characteristics. Physiol Behav. 2023;265:114158. doi:
- 7.↑
Miras-Moreno S, Pérez-Castilla A, Rojas-Ruiz FJ, García-Ramos A. Lifting velocity predicts the maximum number of repetitions to failure with comparable accuracy during the Smith machine and free-weight prone bench pull exercises. Heliyon. 2023;9(9):e19628. doi:
- 8.↑
Miras-Moreno S, Pérez-Castilla A, García-Ramos A. Lifting velocity as a predictor of the maximum number of repetitions that can be performed to failure during the prone bench pull exercise. Int J Sports Physiol Perform. 2022;17(8):1213–1221. doi:
- 9.↑
García-Ramos A, Torrejón A, Feriche B, et al. Prediction of the maximum number of repetitions and repetitions in reserve from barbell velocity. Int J Sports Physiol Perform. 2018;13(3):353–359. doi:
- 10.↑
Miras-Moreno S, García-Ramos A, Rojas-Ruiz FJ, Pérez-Castilla A. Impact of lifting straps on the relationship between maximum repetitions to failure and lifting velocity during the prone bench pull exercise. Sports Health. 2024;10:163. doi:
- 11.↑
Page MJ, Mckenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi:
- 12.↑
Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. Rayyan—a web and mobile app for systematic reviews. Syst Rev. 2016;5(1):384. doi:
- 13.↑
Downs SH, Black N. The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. J Epidemiol Community Health. 1998;52(6):377–384. doi:
- 14.↑
Weakley J, Cowley N, Schoenfeld BJ, et al. The effect of feedback on resistance training performance and adaptations: a systematic review and meta-analysis key points. Sports Med. 2023;53(9):1789–1803. doi:
- 15.↑
Janicijevic D, Şentürk D, Akyildiz Z, Weakley J, García-Ramos A. Using velocity recordings to predict squat repetitions to failure in high-level wrestlers. Eur J Sport Sci. 2024;24(3):364–371. doi:
- 16.↑
Pérez-Castilla A, Miras-Moreno S, Janicijevic D, García-Ramos A. Velocity loss is not an accurate predictor of the percentage of completed repetitions during the prone bench pull exercise. J Strength Cond Res. 2023;37:1001–1008. doi:
- 17.↑
Pérez-Castilla A, Miras-Moreno S, Weakley J, García-Ramos A. Relationship between the number of repetitions in reserve and lifting velocity during the prone bench pull exercise: an alternative approach to control proximity-to-failure. J Strength Cond Res. 2023;37(8):1551–1558. doi:
- 18.↑
Pestaña-Melero FL, Haff GG, Rojas FJ, Pérez-Castilla A, García-Ramos A. Reliability of the load-velocity relationship obtained through linear and polynomial regression models to predict the 1-repetition maximum load. J Appl Biomech. 2018;34(3):184–190. doi:
- 19.↑
García Ramos A. Resistance training intensity prescription methods based on lifting velocity monitoring. Int J Sports Med. 2024;45(4):257–266. doi:
- 20.↑
García-Ramos A. The 2-point method: theoretical basis, methodological considerations, experimental support, and its application under field conditions. Int J Sports Physiol Perform. 2023;18(10):1092–1100. doi:
- 21.↑
García-Ramos A, Jaric S. Two-point method: a quick and fatigue-free procedure for assessment of muscle mechanical capacities and the 1 repetition maximum. Strength Cond J. 2018;40:54–66.
- 22.↑
García-Ramos A, Pestaña-Melero FL, Pérez-Castilla A, Rojas FJ, Gregory Haff G. Mean velocity vs. mean propulsive velocity vs. peak velocity: which variable determines bench press relative load with higher reliability? J Strength Cond Res. 2018;32(5):1273–1279. doi:
- 23.↑
Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Med. 1998;26(4):217–238. doi:
- 24.↑
Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res. 2005;19:231–240. doi:
- 25.↑
García-Ramos A. The 2-point method: theoretical basis, methodological considerations, experimental support, and its application under field conditions. Int J Sports Physiol Perform. 2023;18(10):1092–1100. doi:
- 26.↑
Miras-Moreno S, García-Ramos A, Jukic I, Pérez-Castilla A. Two-point method applied in field conditions: a feasible approach to assess the load-velocity relationship variables during the bench pull exercise. J Strength Cond Res. 2023;37:1367–1374. doi:
- 27.↑
Pérez-Castilla A, García-Ramos A. Changes in the load-velocity profile following power- and strength-oriented resistance-training programs. Int J Sports Physiol Perform. 2020;15(10):1460–1466. doi:
- 28.↑
Sánchez-Medina L, González-Badillo JJ, Pérez CE, Pallarés JG. Velocity- and power-load relationships of the bench pull vs. bench press exercises. Int J Sports Med. 2014;35:209–216. doi:
- 29.↑
Janićijević D, Miras-Moreno S, Morenas-Aguilar MD, Moraga-Maureira E, Weakley J, García-Ramos A. Optimizing mechanical performance in the bench press: the combined influence of inter-set rest periods and proximity to failure. J Sports Sci. 2023;41:2193–2200. doi:
- 30.↑
Weakley J, Cowley N, Schoenfeld BJ, et al. The effect of feedback on resistance training performance and adaptations: a systematic review and meta-analysis. Sports Med. 2023;53(9):1789–1803. doi:
