Measurement of physical fitness is conducted to monitor training progress and to estimate the athlete's potential to perform at the next higher level of competition in all team sports. Although the particular tests that are administered vary across settings, the comprehensive evaluation typically encompasses assessments of speed, strength, agility, aerobic and anaerobic energy systems, and body composition.
Research directed at determining the effectiveness of physical fitness measures for predicting hockey performance has yielded inconsistent findings (1-4). One reason that may account for the inconsistency between studies is the failure to take into account the covariation between the tests. Although a single test score may not be a sufficiently robust predictor of future performance, an aggregate index based on correlated scores obtained from a constellation of tests may have stronger prognostic utility.
In addition to estimating an athlete's prospects of transitioning to the next higher level of competition, fitness professionals are integrally responsible for designing and monitoring the training and conditioning program. Currently, the athlete's progress is tracked by recording changes in scores on specific tests. Aggregating test scores into conceptually coherent and empirically validated measurement domains provides the opportunity to quantify the components of fitness that require intervention.
The present investigation had three objectives: a) derive a composite index of physical fitness that takes into account the shared variances among diverse measures; b) determine the accuracy of the composite index for identifying athletes prior to the entry draft who will transition to the NHL; and, c) validate indexes in four physical fitness domains (upper body strength, lower body power, energy systems, body composition) based on their shared variance between different tests.
Experimental Approach to the Problem
This project had a twofold focus. First, while many testing programs and systems yield scores that rank athletes according to physical and physiological attributes, the information obtained is entirely sample-dependent and not predictive of future performance. For example, athletes scoring in the 99th percentile in two different cohorts are not necessarily equivalent since the rank is directly the characteristics of the particular group. Accordingly, a rank or a percentile score is not informative for making predictions about an athlete's prospects. Conducting receiver operating curve analysis, however, surmounts the limitations of relying on rank scores and enables prediction of individual outcome.
Second, from the conceptual perspective, this project expanded the evaluation focus of physical fitness from measures/scores on tests to also include quantification of the underlying processes. Because of the strong intercorrelations between different tests, extracting the variance that is shared among the different components of physical fitness provides the opportunity to quantitatively assess distinct processes. In effect, aggregating variables based on their statistical properties enables development of composite indexes that could not only simplify interpretation of test results, but importantly also document physical fitness in a more robust and precise manner.
The sample consisted of 345 players invited over a 4-year period to the annual NHL scouting combine administered by Central Scouting Services, an entity of NHL Operations. These athletes do not represent a random sample but rather constitute a select group based on their hockey performance record and scout reports. In accord with York University policy, this investigation was approved by the office of human research ethics, and written informed consent was obtained from participants, as described by the ACSM guidelines, with the understanding that data would be used only in a summarized form and that partipant names would remain confidential.
The test variables obtained in the combine were organized into four categories: a) upper body strength, b) lower body power, c) body composition, and d) energy systems. Exploratory factor analysis was performed on the scores in each category. This analysis revealed that 18 variables, shown in Table 1, have a factor loading of 0.4 or higher. As the strongest indicators of the particular process, the factor score was used to index the particular process.
The four factors were next submitted to a second-order factor analysis. This analysis reduced the 18 variables to one global index, termed the composite physical fitness index (CPFI). The score on this index was evaluated to determine its accuracy for identifying athletes who successfully transitioned to the NHL, defined herein as playing 5 or more games within a 4-year period after the draft.
The measures listed in Table 1 were obtained in a uniform manner using standard procedures and equipment. The tests were administered by fitness professionals under supervision of two of the authors (NG and VJ).
Five measurements were utilized to derive the upper body strength index:
While extending the hand downward, a dynamometer is squeezed as forcefully as possible with the dominant hand followed by the nondominant hand.
The athlete performs a 150-pound bench press repeated at the rate of 25 per minute in time with a metronome. While performing this task, the athlete lies on his back on a standard padded bench while buttocks are in contact with the bench and his feet are upon the floor. The barbell is gripped such that the thumbs are aligned with the shoulders. In the starting position, the bar is touching the chest at approximately the axilary line from which the athlete pushes the bar to full arm extension in synchrony with the metronome. The number of properly executed consecutive repetitions is scored.
Three trials are administered using an upper body strength meter. While standing, the athlete pushes the handles away from his body in an even (nonjerking) motion.
While lying on his stomach, legs together, and hands positioned under the shoulders with fingers pointing forward, the athlete performs push-ups at the rate of 25 per minute in time with a metronome. A successful trial requires the athlete to fully straighten the elbows and then lower the upper body to the point where the elbows are at a 90o angle provided that the stomach or thigh are not touching the mat. The score is the number of properly executed consecutive repetitions before losing synchrony with the metronome.
Upper Body Development
The degree of upper body development was rated by a physician using one of four categories: below average, average, above average, extensive.
Four measurements were administered to derive the lower body power index:
With fingers outstretched, the athlete reaches as high as possible while standing flat footed. Next, three trials are conducted that require the athlete to jump as high as possible without a prestep and with fingers outstretched in an attempt to touch the highest marker on the Vertec apparatus. Leg power is calculated using the Sayers equation: Peak leg power (Watts) = [60.7 × jump height (cm)] + [45.3 × body mass (kg)] - 2055.
Standing Long Jump
With feet slightly apart, the athlete leaps as far as possible using an arm swing to maximize distance. The best score in three trials is recorded consisting of the distance between the start line and the heel mark on the landing spot.
While supine, the athlete is positioned such that the knees are flexed to a 90o angle and the heels are in contact with the ground. Arms are crossed and the hands rest on the opposite shoulder during performance of the task in which the athlete sits up and curls over to contact the elbows and thighs while his heels remain stationary on the ground. Upon return to the supine position, the shoulder blades contact the mat before beginning the next trial. The curl-ups are performed in time to a metronome that is set to a rate of 25 per minute. The task continues up to a maximum of 100 repetitions or until cadence with the metronome cannot be maintained.
Lower Body Development
The degree of upper body development was rated by a physician using one of four categories: below average, average, above average, extensive.
Three measurements were administered to derive the body composition index.
Sum of Skinfolds
A Harpenden fat caliper was used to measure skinfold at six locations on the right side of the body: a) chest area slightly lateral to the right nipple; b) triceps; c) subscapula, d) umbilicus, e) suprailiac, and f) front thigh. The sum of these measurements was recorded.
The above data are used to compute percent body fat using the following formula: 0.097 (∑ 6 skinfolds) + 3.64 = X% fat.
The athlete was weighed using a standard scale.
The following variables were used to derive the energy systems index:
This was measured using a computerized version of the Wingate protocol. While sitting on a cycle ergometer with the feet secured in stirrups, the pedals were adjusted so that when a foot was in the down position, the leg was only slightly flexed. The athlete began the task by pedaling for 2 minutes with the resistance set to a low level. The athlete then pedaled at progressively faster speed until maximum workload, at which point pedaling continued for 30 seconds. Workload was calculated by the formula: body weight in kilograms × 0.90. The number of pedal revolutions was recorded in each 5-second interval. Mean power output (watts/kg) and peak power output (watts/kg) were calculated for the highest 5-second interval and for the total 30-second task period.
This was also measured on a cycle ergometer. During warm-up, the athlete was familiarized with the task by pedaling with a mouthpiece. Next, with a metronome set at 70 rpms, the athlete pedaled with resistance set at 2, 3, and 4 kg for respectively 0-2, 2-4, and 4-6 minutes of the task. This was followed by four more segments in which resistance was set at 4.5, 5.0, 5.5, and 6.0 kg for the next 6-7, 7-8, 8-9, and 9-10 minutes. The task was discontinued when the athlete either stopped performing the task or he could no longer maintain the rpms despite exhortation by the test administrators A cool-down period was then implemented. Aerobic power is reported as a relative score (ml/kg/min) together with aerobic final workload (watts) and aerobic test duration (seconds).
Applying receiver operating curve analysis, it was found that defensemen scoring at the 80th percentile on the CPFI have almost 70% probability of playing 5+ games in the NHL within 4 years after the draft. Forwards scoring in the 80th percentile have a 50% chance of playing in the NHL. When they have scores on the CPFI at the 90th percentile, defense and forwards respectively have 72% and 61% probability of playing in the NHL.
Although these results indicate that fitness measurements aggregated into a composite index have predictive utility, administering a comprehensive protocol is time, labor, and technology intensive. Accordingly, additional analyses were conducted to assess whether a brief portable protocol could be derived using selected tests (see Table 1), which have good factor loadings. This simplified assessment protocol consisted of right- and left-hand grip strength, push-ups, long-jump, body composition, and body weight. Factor analysis of these variables revealed that they are indicators of a major factor which accounts for 35% of overall variance.
The utility of this simplified protocol was explored by determining whether it can detect athletes who have the potential to transition to the NHL. Toward this goal, the sample was thus divided into three groups-low, middle, and high-based on their percentile rank on this brief protocol. The results indicated that there is a 5% progressive increase in the probability of playing in the NHL across low, middle, and high tertiles. At the 90th percentile, where physical fitness is truly superior, the likelihood of transitioning to the NHL for defensemen and forwards is respectively 51% and 46%. Thus, as a quick screening assessment, the brevity of the simplified protocol (taking approximately 10-15 minutes to administer) can identify superior athletes who have a reasonable chance of playing in the NHL. This protocol also enables easy quantitative charting of the athlete's progress during chronological development, as well as monitoring improvement concomitant to training and conditioning.
In addition to prediction, it is noteworthy that the methods used to aggregate the test variables into conceptually coherent and empirically validated domains provides fitness professionals with the opportunity to expeditiously profile the athlete's strengths and weaknesses. By way of illustration, Figure 1 depicts the results of 15 randomly selected athletes who participated in the NHL combine prior to the draft. As can be seen, the indexes in the four categories comprise a straightforward qualitative and quantitative profile that captures the variability between athletes as well as variability across measurement domains within each athlete.
To succinctly recapitulate, the results of this study demonstrated that fitness professionals have the opportunity to obtain quantitative data to augment their expert judgment regarding an athlete's potential to play hockey at the elite level. As shown herein, test data can also be analyzed, arrayed, and interpreted in a fashion that informs athlete's prognosis. Where there is a large population to evaluate or there is a need to conduct repeated evaluations, the simplified protocol provides a good screening of the athlete's potential and current fitness status.
Physical fitness is only one ingredient required for success at the elite competitive level. Including additional information (e.g., scout reports, game statistics, mental efficiency) would most likely increase prediction accuracy. The diverse expertise required to evaluate the multifaceted components of athlete talent notwithstanding, the results of this study demonstrate that fitness professionals have unique and important contributions to make to the decision-making process pertaining to athlete selection. As shown herein, the CPFI has predictive validity; therefore, this index offers fitness professionals the opportunity to gauge the prospects of playing hockey at the NHL level.
Further research remains, however, to be conducted to refine and improve the CPFI so that prediction accuracy is maximized. Toward this goal, it is noteworthy that measures of agility and equilibrium were added in the 2007 NHL combine. Quantifying the athlete's agility and ability to maintain balance would appear to have face validity for improving prediction of on-ice performance.
The main finding of this study indicates that it is practical and feasible to estimate an athlete's potential for success as a NHL player using physical fitness data obtained in the combine prior to the entry draft. In broader terms, the measurement and analytic procedures described herein are useful for also predicting to increasingly higher levels of competition prior to the NHL. In other words, while the results demonstrate the utility of comprehensive physical evaluation for predicting transition to the elite level, the same framework is appropriate for application at earlier stages in the athlete's career. Accordingly, it can be concluded that the evaluation protocol and methods used for prediction can be utilized to track the athlete's progression beginning during hockey competition as a youth to systematically predict likelihood of success at increasingly more competitive levels during career development.
In addition, the brief portable protocol, which also was shown to have predictive utility, offers several practical applications. It can be used, for example, to quickly gauge an athlete's potential during evaluations by scouts. In this regard, comprehensive accurate appraisal of the athlete requires the collaborative interaction of fitness professionals and scouts. Furthermore, the brief protocol is an efficient procedure to quantify fitness during the recovery process after injury, declines consequent to aging, or change after a long layoff from training.
Lastly, it should be noted that emerging findings in the empirical literature demonstrates that low physical fitness impedes mental efficiency. Accordingly, suboptimum physical fitness not only promotes fatigue during competition but may also precipitate a mental lapse. This inseparability of mind and body points to the value of fitness professionals to extend their competencies beyond measurement and enhancement of physical fitness to also include basic knowledge and skills required for evaluation of mental efficiency. Notably, measurements of mental efficiency have recently been added to the NHL combine. Tests that measure mental speed in the order of milliseconds complemented by evaluation of concentration, working memory, and spatial awareness are administered to more fully characterize fitness in both the physical and mental spheres.
Accordingly, a “promising practice” model of athlete evaluation, shown in Figure 2, integrates physical fitness data, scout and coach reports, measures of mental efficiency, game performance statistics, and psychological traits pertinent to success at the elite level. Repeated assessment using this model enables graphical charting of the athlete's trajectory to the elite level. In this fashion, the model is useful for informing fitness professionals and coaches about interventions that could be implemented to further improve the athlete's prospects of success.
In conclusion, the present study demonstrated that planned data reduction and aggregation procedures substantially improve the utility of physical fitness evaluation for prediction as well as for profiling the athlete's current strengths and weaknesses. The results, although confined to hockey players in this study, also point to the utility of this framework for identifying talented athletes in other team sports. For example, a recent study has shown that athletes drafted by the National Football League score higher on key facets of physical fitness compared to nondrafted athletes (5). These findings are complemented by the results obtained herein demonstrating that these measures can also be used as predictors of success. Indeed, a fertile area of future research involves joining protocols that capture the features of physical fitness pertinent to success that generalize across sports in conjunction with tests that measure processes integral to success in each specific sport.