The National Football League (NFL) annually hosts a National Scouting Combine (NSC) of approximately 300 elite Division 1 college football players (16). During the week-long event held in Indianapolis, IN, draft prospects participate in a variety of psychological and physical fitness tests that include anthropometric measurements, the Wonderlic Personnel Test, the 40-yd dash, standing long jump (SLJ), countermovement vertical jump (CMVJ), 3-cone drill, 20-yd shuttle run, 225-lb bench press (quarterbacks [QBs] excluded), and position-specific skills (16). Players invited to the combine represent the athletes most likely to be drafted into the NFL (16). Coaches, executives, scouts, and medical personnel from the 32 NFL teams are present to conduct interviews and observe field testing (7).
A significant amount of revenue is generated from the combine each year (7). In addition to the 6.51 million viewers who watched the NSC on the NFL Network in 2013, proceeds were also made through sponsorship and advertising (7). In addition, prospects may invest up to $25,000 on precombine strength and conditioning training (7). Given the public interest, player obligations, coaching staff commitments, and business aspects of the combine, several questions have arose in recent years concerning the applicability of combine scores to eventual draft status and NFL performance.
The performance measurements taken at the NFL scouting combine possess face validity (12). An American football team is comprised several positions, each requiring different athletic abilities and skill sets. For instance, wide receivers (WRs) and defensive backs routinely run more than 30 yd per play, whereas lineman rarely runs more than 10–12 yd at a time (17). The game of football is primarily an anaerobic event consisting of high-intensity work periods of approximately 5 seconds followed by an average rest period of 36 seconds when not including for extended rest periods due to injury or timeout (9). The body relies heavily on the creatine phosphate system to provide immediate energy for short duration (<10 seconds), explosive movements (8,11,21). During sport competition, players routinely sprint, change direction, move laterally, tackle, dive, or jump (10). The player who executes the prescribed set of movements most quickly often wins their positional battle. For example, a WR that outruns a cornerback on a particular pass pattern has a greater chance making an uncontested catch. This is also true of defensive positions. For instance, a defensive end's (DE) ability to explode out of his stance and quickly cover the distance between himself and the QB may mean the difference between a sack and a completed pass. Therefore, power is a necessary attribute for successful performance in the NFL (21).
Speed is an essential component of horizontal anaerobic power. Sprint power has been proposed as an expression of force (mass) × velocity (d/t) (5). Consequently, converting sprint raw scores (time) to absolute and relative power measurements may yield significant relationships. Scores for vertical jump height and the 40-yd sprint time have been shown to differentiate between division of play (I, II, or III) and player ability on the collegiate level (3). The CMVJ is also considered a valid tool for measuring anaerobic power (21). Therefore, it is possible that CMVJ power may be indicative of success on the professional level. Due to the complex interaction between vertical and horizontal displacement necessary to perform the SLJ, no equation for power has been validated at this time. However, the SLJ has been identified as one of the best predictors of agility and speed in NFL players (2).
The 20-yd shuffle represents an agility test known to be valid in football and previously used in research on American football players (20). Numerous studies have investigated the correlations between scores rendered from field tests administered in the NSC and draft status (6,14,15,18,22), or collegiate level football playing ability (14,20). A study in 2012 by Robbins and Young examined the relationship between various sprint and jump abilities by football position and stated >20 distinct player positions exist between offense, defense, and special teams (17), with each position requiring different skills and abilities. Therefore, it is possible that certain combine tests are more relevant to specific player positions than others. Furthermore, success at one position is often affected by the skills and abilities of another player position all together. For instance, running back (RB) success is impacted by the blocking efficacy of the offensive line.
Only 2 studies to date have sought to investigate the relationship between scores on NSC events and future NFL performance (12,23). Combine sprint times were shown to predict RB success in the NFL (12). However, sprint time alone did not significantly correlate with performance variables for the WR or QB position (12). Defensive positions were not included in either study (12). Furthermore, to the author's knowledge, no previous studies have specifically focused on measures of power.
The primary purpose of this study is to investigate the relationship between specified NSC scores and measures of performance by player position during an individual's first 3 seasons in the NFL. Quarterbacks and kickers have longer careers than other positions. Hence, the first 4 years of QB performance were included. A secondary aim was to determine whether correlated variables could predict player performance at the QB, RB, WR, DE, defensive tackle (DT), and linebacker (LB) positions.
Experimental Approach to the Problem
The study was approved as a secondary data analysis by the University of Rhode Island's Institution Review Board. A secondary data analysis of combine scores and player performance variables was conducted. Combine scores were obtained from NFL.com and a commercial web site devoted to combine data (NFLcombineresults.com). Data were cross-referenced and deemed accurate. Player performance statistics were acquired solely from NFL.com.
Subjects in this study were combine participants between the years 2005–2010 who subsequently played in the NFL. The subjects ranged from 19-25 years of age. Informed consent was not necessary as all data was publically available and no identifying information was disclosed. University of Rhode Island Institutional Review Board approved the study. These years were chosen as it allowed for analysis of the most recent combine attendees who had played for a minimum of 3 years in the NFL. This was done in an attempt to control for possible significant differences due to changes in the style of game play or player size/ability over time. The positional groups investigated were QBs (N = 44), RBs (N = 82), WRs (N = 116), LBs (N = 139), DEs (N = 59), and DTs (N = 72). All measurements were expressed as mean ± SD.
Inclusion criteria were QBs who averaged a minimum of 1 passing attempt over 4 years, RB with a minimum of 1 rushing attempt, WRs with 1 catch, and defensive players with 1 tackle averaged over the first 3 years of their respective careers. Those excluded were combine attendees who did not perform in physical fitness tests and those who attended but either did not play in the NFL, played at a different position, or on special teams only.
Combine Measurements: Anthropometrics
Each player had his height, body mass, arm length, and hand length measured at the combine. Body composition testing was conducted on RBs and linemen with the BOD POD (Life Measurement Instruments, Concord, CA, USA). For the purpose of this study, body mass (kg) was used in formulas to calculate power and relative scores.
For the purpose of this study, 40-yd (36.6 m) sprint times were used to determine horizontal power (HP). Horizontal power was estimated by the following equation (5):
Countermovement Vertical Jump
Countermovement vertical jump was investigated on the basis of raw score, power, and relative power. Power was calculated by the Lewis formula (5):
Standing Long Jump
The SLJ was investigated on the basis of jump length and relative jump length.
The shuttle run was investigated on the basis of raw score (time).
On-field performance measures indicative of success at each player position were investigated.
On-field performance measures assessed for the QB position were pass attempts (ATTP), completion % (COMP%), total yards (YDP), yards/attempt (YD/ATTP), touchdowns (TD), interceptions (INT), passer rating, rushing attempts (ATTR), rushing yards (YDR), and yards/carry (YD/Rush).
On-field performance measures assessed for the RB position were attempts (ATT), attempts/game (ATT/G), total yards (YD), yards/carry (YD/RUSH), yards/game (YD/GAME), TD, and longest rush (LONG).
On-field performance measures assessed for the WR position were receptions (REC), total yards (YD), yards/catch (YD/REC), yards/game (YD/GAME), TD, first downs (1DN), and first down % (1DN%).
On-field performance measures assessed for the LB position were combined tackles (CT), solo tackles (ST), tackle assists (AT), sacks (SK), passes defended (DEF), and interceptions (INT).
On-field performance measures assessed for the DE position were CT, ST, AT, and SK.
On-field performance measures assessed for the DT position were CT, ST, AT, and SK.
Data analysis was conducted on SPSS version 15.0 for Windows (SSPS Inc., Chicago, IL, USA). Assumptions for normality were tested for and met. Pearson correlation was used to evaluate relationships between variables (position and on-field performance). Correlations were considered low when r = <0.40, moderate r = 0.40–0.60, and large r = >0.60. Hierarchal linear regressions were performed to determine the contribution of specific combine predictors to on-field performance variables at each position. The use of hierarchal linear regression allowed for the examination of unique contributions of variables with high collinearity by testing how much variance could be accounted for by each independent variable in the model when all other predictive variables were already included in the analysis. Significance was set at (p ≤ 0.05).
Of the 116 QBs who attended the combine between 2005 and 2010, height (190.8 ± 3.94 cm) and body mass (101.5 ± 4.93 kg), 44 went on to average a minimum of 1 pass attempt over his first 4 seasons in the NFL. Of those 44 with statistics, 4 were undrafted (9%). All 44 QBs included in the final analysis completed the 40-yd dash, whereas 33 completed the CMVJ, 32 the SLJ, and 35 the 5-10-5.
Of the 139 RBs who attended the combine between 2005 and 2010, height (180.3 ± 4.76 cm) and body mass (97.3 ± 6.44 kg), 82 went on to average a minimum of 1 rushing attempt over his first 3 seasons in the NFL. Of those 82 with statistics, 16 were undrafted (19.5%). All the 82 RBs included in the final analysis completed the 40-yd dash, whereas 79 RBs completed the CMVJ, 73 the SLJ, and 60 the 5-10-5.
Of the 219 WRs, height (185.9 ± 5.68 cm) and body mass (91.8 ± 6.80 kg), who attended the combine between 2005 and 2010, 116 went on to average a minimum of 1 reception over his first 3 seasons in the NFL. Of those 116 with statistics, 11 were undrafted (9.5%). All 116 WRs included in the final analysis completed the 40-yd dash, whereas 102 WRs completed the CMVJ, 95 the SLJ, and 85 the 5-10-5 shuttle.
Of the 218 LBs who attended the combine between 2005 and 2010, height (187.6 ± 3.42 cm) and body mass (109.2 ± 3.81 kg), 139 went on to average a minimum of 1 tackle over his first 3 seasons in the NFL. Of those 139 with statistics, 26 were undrafted (18.7%). All 139 LBs included in the final analysis completed the 40-yd dash, whereas 124 LBs completed the CMVJ, 121 the SLJ, and 112 the 5-10-5.
Of the 120 DEs who attended the combine between 2005 and 2010, height (192.9 ± 3.64 cm) and body mass (121.9 ± 5.75 kg), 59 went on to average a minimum of 1 tackle over his first 3 seasons in the NFL. Of those 59 with statistics, 9 were undrafted (15.3%). All 59 DEs included in the final analysis completed the 40-yd dash, whereas 51 DEs completed the CMVJ, 50 the SLJ, and 47 the 5-10-5.
Of the 120 DTs who attended the combine between 2005 and 2010, height (190.4 ± 3.15 cm) and body mass (139.4 ± 6.36 kg), 72 went on to average a minimum of 1 tackle over his first 3 seasons in the NFL. Of those 72 with statistics, 8 were undrafted (11.1%). All 72 DTs included in the final analysis completed the 40-yd dash, whereas 65 DTs completed the CMVJ, 60 the SLJ, and 54 the 5-10-5 (see Tables 1 and 2 for descriptive statistics).
Correlations of Combine Measures With On-Field Performance for Offensive Positions
At the QB position, the number of passing attempts and total yards passing were correlated with HP only. Rushing attempts were correlated with VJP and HP. Significant correlations were found between QB total rushing yards and 40 time, HP, CMVJ, VJP, and VJRP. All RB on-field performance measures were related to 40 time (p ≤ 0.01). Rushing attempts per game and TD were correlated with HP; rushing attempts per game, total yards and yards per game with SLJ. At the WR position, yards/reception and first down percentage were related to HP (Tables 3,4,5).
Hierarchical Linear Regression of Significant Combine Measures to On-Field Performance for Offensive Positions
Hierarchical linear regressions were run using the combine measure to predict each performance variable by position. These models allow for the reporting of the total variance accounted for by all the combined predictor variables as well as the unique contribution (after accounting for all other predictors) of each variable to that total variance. Results at the QB position indicated unique contributions between HP and passing attempts (r2 = 0.102) and HP and total yards passing (r2 = 0.097), 10.2 and 9.7% of the variance, respectively. Horizontal power and VJP were predictive of rushing attempts (r2 = 0.370). Rushing yards were significantly correlated with 40T, CMVJ, VJP, VJRP, and HP. Variables were grouped as power VJP, VJRP, and HP (r2 = 0.427) and nonpower variables 40T and CMVJ (r2 = 0.126), accounting for 62.2% of the total variance. Total yards rushing at the QB position were predicted by 40T (r2 = 0.073), HP, and VJP (r2 = 0.201) combined accounting for 31.3% of the total variance.
Combine variables and performance variables at the RB position showed predictive values between 40T and yards per rush (r2 = 0.080) and longest rush (r2 = 0.187). 40T and SLJ combined were predictive of total rushing yards (r2 = 0.200), rushing attempts (r2 = 0.195), and yards per game (r2 = 0.197). Horizontal power and 40T accounted for 13.5% of the variance for TDs (r2 = 0.135). Horizontal power, SLJ, and 40T combined to account for 21% (r2 = 0.214) of the number of rushing attempts per game.
At the WR position, hierarchal regression showed minimal predictive values between CMVJ and yards/pass (r2 = 0.049), as well as CMVJ and first down % (r2 = 0.053).
Correlations of Combine Measures With On-Field Performance for Defensive Positions
At the LB position, sacks were significantly related to 40 times, HP, CMVJ, VJP, and 5-10-5. Solo tackles and combined tackles were correlated with 40 times. For DEs, all tackling variables were significantly correlated with HP and VJP. A significant relationship was also shown between VJP and sacks. Only the SLJ was significantly correlated with SK at the DT position (Tables 4,6,7,8).
Hierarchical Linear Regression of Significant Combine Measures to On-Field Performance for Defensive Positions
Hierarchical linear regressions were run using the combine measure to predict each performance variable by position (Tables 5–8). Significant correlations showed unique contributions of sacks at the LB position. Variables were grouped as 40T, CMVJ, and 5-10-5 (r2 = 0.056), and 40HP, VJP, and VJRP (r2 = 0.153), accounting for a total of 18.3% of the total variance. Linebacker 40T accounted for 3% of the variance for CT (r2 = 0.031).
Power scores were predictive of CT at the DE position with unique contributions of 40HP (r2 = 0.096) and VJP (r2 = 0.018), accounting for a total of 21% of the variance. VJP and SLJ accounted for 14% (r2 = 0.141) of the variance for DE sacks.
Hierarchical regression performed on the SLJ and sacks at DT, accounted for 6.7% (r2 = 0.067) of the variance.
Significant correlations were found between several NSC scores and subsequent player performance, and when examined through multiple regression, those combine variables accounted for between 4 and 62% of on-field performance measures. In accordance with previous studies, RB success was most strongly related to 40-yd dash time (6,12). This was true for all RB on-field performance measures included in this study at (p ≤ 0.01). Forty-yard dash time, combined with the SLJ and HP, accounted for 18.7–21.4% of the variance for RB attempts, attempts per game, total yards, yards per game, and longest rush. The 10-yd dash was also shown to predict yards per carry of RB in the NFL over the career of RBs attending between 2000 and 2009 (23). Therefore, although linear speed may not be integral to on-field performance in other positions, it is essential to successful rushing.
The SLJ was related to sacks for DTs and DEs. This is likely due to the combined vertical and horizontal displacement required for movements that originate from a 3-point stance. Interestingly, the SLJ was significantly correlated with performance measures for RBs, but not LBs. Although both groups generally initiate movement from the 2-point stance, the LBs are taught to perform 2–3 read steps at the snap to determine whether it is a running or passing play. In turn, RBs often take an explosive step toward the line of scrimmage at the start of a play. Forty-yard dash time was related to all tackling variables at the LB position. Sacks were related to all NSC variables with the exception of SLJ. The remaining NSC variables accounted for a total of 18.3% of the total variance in LB SK. This suggests a modest benefit of the current testing battery in predicting defensive player success on the professional level.
Only one other study to date used combine measures to predict on-field player performance (23). Only RBs and WRs who attended the combine between 2000 and 2009 were included in previous research (23). Therefore, this study allowed for a more comprehensive investigation through the addition of QBs, LBs, and defensive linemen. Furthermore, to the best of the author's knowledge, measures of horizontal and vertical power were not included in any previous research regarding the NSC. As football is a sport that relies on explosive movements, it was hypothesized that power would be a better predictor of performance than raw scores for speed (40T) or jumping ability (CMVJ).
Performance variables at the QB, DE, and WR positions were significantly correlated with power scores opposed to 40-yd dash speed or CMVJ height. Horizontal and vertical power, rather than raw scores for speed and CMVJ, were predictive of tackling performance at the DE position. This suggests the ability to engage in explosive movements is relevant to success of NFL DEs, where speed or jumping ability alone is not. For QBs, HP and VJP accounted 37% of the variance in the number of rushing attempts per season, 42.7% of the variance in total rushing yards, and 20% of yards gained per rushing attempt. Hence, more powerful QBs are not only more likely to run but also to gain more yards per rush than their less powerful counterparts.
National Scouting Combine measures were least related to performance at the WR and DT positions. Previous research has shown body mass and 10-yd dash time to be the best predictors of DTs being drafted into the NFL (15). Therefore, assessing DTs for speed and HP at a 10-yd split may lead to further significant correlations. In this study, VJ was significantly correlated with yards per reception for WRs, whereas a prior investigation showed player height was the best predictor of future performance (23). Therefore, it is likely that catching radius, or the combination of height, arm length, and jumping ability combined, is likely a primary component of WR success.
There are several limitations in the current study. First, measurable performance variables at the cornerback, safety, tight end, center, and offensive line positions are difficult to discern. Therefore, several positional groups were not included in the analysis. For the player positions that were included, it is possible that some better players accrued less statistics due to the opposing team's game strategy. For example, overpass/underpass coverage of a WR or using a tight end as an additional blocker could lead to more success for players not targeted by the double team. Another potentially confounding variable is the interrelation between player positions. For instance, if a QB has a weak offensive line, they are more likely to be sacked or throw incomplete passes. Conversely, a RB with a strong offensive line is more likely to gain yards and score TD. Those whose performance was impacted by injury were also not accounted for. In addition, performance measures cannot be standardized, as teams may favor the run or pass on offense, leading to fewer opportunities for an entire position or positions. Likewise, not all teams play the same caliber of opponents or under the same environmental conditions. Finally, players who did not attend the combine or those who played at a different position were excluded from the study.
The business aspects, public interest, preparation, and player obligations, along with commitments from coaching and medical staff have brought the validity of the NSC to predict subsequent performance in NFL under scrutiny. Although agility is a necessary skill in American football, the 20-yd shuttle drill was not significantly correlated with any performance variable with the exception of LB SK. This suggests position-specific agility drills, in addition to the 20-yd shuttle drill, be added to future NSC tests. Research by Robbins and Goodale (19) support this claim and recommend reaction drills based on position-specific demands.
The SLJ was significantly related to variables at RB, DE, and DT. It is suggested strength and conditioning coaches working with these player groups specifically target exercises that maximize simultaneous vertical and horizontal displacement. Horizontal and vertical power was predictive of performance in QBs and DEs. Therefore, calculating power may be a useful addition to the current combine measures. This could be done through the mathematical conversions of raw scores or performing the CMVJ on a force plate in the case of vertical jump power (15). Power was strongly predicted to QB rushing variables. Here, coaches and coordinators should consider power when drafting a QB for a triple option or wildcat-type offense.
On-field performance in college is likely the strongest predictor of success in the NFL (13). However, the current study suggests combine tests are modest predictors of future performance. Should the NFL change the current NSC testing battery, position-specific tests are recommended. These include a 10-yd dash for linemen and change-of-direction drills that similar to those needed to execute successful pass patterns for WRs.
1. Barker M, Wyatt T, Johnson R, Stone M, O'Bryant C, Kent M. Performance factors, psychological assessment, physical characteristics, and football playing ability. J Strength Cond Res 7: 224–233, 1993.
2. Fairchild B, Amonette W, Spiering B. Prediction models of speed and agility in NFL
combine attendees [Abstract]. J Strength Cond Res 25: 96, 2011.
3. Fry A, Kraemer W. Physical performance characteristics of American collegiate football players. J Strength Cond Res 5: 126–138, 1991.
4. Garstecki M. Comparison of selected physical fitness and performance variables between NCAA Division I and II football players. J Strength Cond Res 18: 292–297, 2004.
5. Haff GG, Dumke C. Laboratory Manual for Exercise Physiology. Champaign, IL: Human Kinetics, 2012. p. 324.
6. Hartman M. Competitive performance compared to combine performance as a valid predictor of NFL
draft status. J Strength Cond Res 25: 105–106, 2011.
7. Heitner D. “The Economics Of The NFL
Combine.” Forbes, Forbes Magazine, 21 Feb. 2013, www.forbes.com/sites/darrenheitner/2013/02/21/the-economics-of-the-nfl-combine/#23714265385e
. Accessed April 21, 2014.
8. Hoffman J. The applied physiology of American football
. Int J Sport Physiol Perform 3: 387–392, 2008.
9. Iosia M, Bishop P. Analysis of exercise-to-rest ratios during Division IA televised football competition. J Strength Cond Res 22: 332–340, 2008.
10. Kalinski M, Norkowski H, Kerner M, Tkaczuk W. Anaerobic power
characteristics of elite athletes in national level team-sport games. Eur J Sport Sci 2: 1–21, 2002.
11. Kraemer W, Fleck S, Deschenes M. Exercise Physiology Integrating Theory and Application. Philadelphia, PA: Wolters Kluwer, 2012. pp. 32–34.
12. Kuzmits F, Adams A. The NFL
combine: Does it predict performance in the National Football League? J Strength Cond Res 22: 1721–1727, 2008.
13. Lyons BD, Hoffman BJ, Michel JW, Williams KJ. On the predictive efficiency of past performance and physical ability: The case of the National Football League. Hum Perform 24: 158–172, 2011.
14. McGee KJ, Burkett LN. The National Football League combine: A reliable predictor of draft status? J Strength Cond Res 17: 6–11, 2003.
15. Miller J, Eisenman PA, Waller MA, Vanhaitsma TA, Williams DP. Identification of position-specific combine test score thresholds for drafted and non-drafted defensive players entering the National Football League [abstract]. J Strength Cond Res 24: 1, 2010.
16. National Football League. NFL
.com [Internet]. Enterprises LLC, www.nfl.com/stats/player
. Accessed April 24, 2014.
17. Robbins D, Young W. Positional relationships between various sprint and jump abilities in elite American football
players. J Strength Cond Res 26: 388–397, 2012.
18. Robbins D. The National Football League (NFL
) combine: Does normalized data better predict performance in the NFL
draft? J Strength Cond Res 24: 2888–2899, 2010.
19. Robbins DW, Goodale T. Evaluation of the physical test battery implemented at the National Football League Combine. Strength Cond J 34: 1–10, 2012.
20. Sawyer DT, Ostarello JZ, Suess EA, Dempsey M. Relationship between football playing ability and selected performance measures. J Strength Cond Res 16: 611–616, 2002.
21. Seiler S, Taylor M, Diana R. Assessing anaerobic power
in collegiate football players. J Strength Cond Res 4: 9–15, 1990.
22. Sierer SP, Battaglini CL, Mihalik JP, Shields EW, Tomasini NT. The National Football League Combine: Performance differences between drafted and nondrafted players entering the 2004 and 2005 drafts. J Strength Cond Res 22: 6–12, 2008.
23. Teramoto M, Cross CL, Willick SE. Predictive value of NFL
scouting combine on future performance of running backs and wide receivers. J Strength Cond Res 30: 1379–1390, 2016.