Off-field athlete testing is often used to gather information about functional capacities and skill level to evaluate and identify elite performers (4,6,9,18,24). Typically, anthropometry is used to describe the ‘ideal’ body size and composition for individual sports and positions (17), whereas aerobic and anaerobic performance may be monitored to identify changes during the season or in response to training programs (7,10,13). A wide variety of both laboratory and field tests can be used to assess on- and off-field physiological characteristics (e.g., maximal oxygen consumption, vertical jump) that may be related to performance during competition. Besides physical traits, psychological attributes, nutrition, and team cohesion are also factors that may impact the eventual success of a team or individual athlete (26,27).
The sophistication and complexity of tests range from stepping on a scale to obtain body weight to intricate agility and physical fitness evaluations. However, the most difficult aspect of these tests often lies not in their administration but in their interpretation. For example, what is a “good” result? How do characteristics such as speed, skill, body composition, and aerobic capacity predict game performance? Examining elite performers and describing these characteristics and their relation to game performance can help elicit answers to these questions.
Individual sports, especially those where the objective is relatively straightforward (e.g., crosscountry runners must run faster than their competitors to win), present fewer facets of performance to be examined. In contrast, prediction of success in team sports is more complicated because there is a need to integrate the skills of many individuals and the tactical strategies of the coaching staff to create the most successful team possible. For example, hockey players must not only be fast skaters but also explosive, agile, competent puck handlers and passers that work well with their teammates and within an offensive and defensive system compatible with their style of play. Several articles have described the typical physical characteristics of elite hockey players (1,3-5,9,11,12,16,19,21,22,25,29-31) but how a player combines these physical traits with on-ice physical skills such as passing and shooting to be successful must also be examined. Preseason testing of physical characteristics is common in collegiate athletics, but for this testing to be most valuable to a coach, the relationship between the parameters being tested and overall game performance should be established. This is an important challenge presented to today's sports scientists, coaches, and players.
The purpose of this article was twofold: (a) to identify relationships between results of preseason testing and hockey game performance over the course of a season and (b) to describe physical fitness and skill levels of players grouped by a coach's subjective evaluation.
Experimental Approach to the Problem
We used data from the 2007 National Collegiate Athletic Association Division I men's hockey championship team in an attempt to identify variables related to a general measure of game performance (the plus-minus [+/−] system). Anthropometric and physiologic data (height, weight, body composition, aerobic fitness, strength, speed, and agility) were compared with +/− score and coach rankings to determine if any relationships existed or if any of these factors predicted performance. All tests are part of the typical preseason testing battery performed by the team each fall and were selected by the coaching staff. Many of the tests included are either part of the National Hockey League (NHL) Draft Entry Camp or routinely used in hockey-specific testing protocols and are similar to what other National Collegiate Athletic Association (NCAA) Division I hockey teams use for on- and off-ice testing (personal communication, Michael Vorkapich, December 18, 2010).
All forwards (F) and defensemen (D) who played on the 2007 National Collegiate Athletic Association Division I men's ice hockey championship team were included in this study (N = 24). Goaltenders were not included in our analysis because they have a distinctly different role in the game and therefore require a different skill set to be successful. Values for O2max, height, weight, body composition, strength, speed, and agility were collected as part of annual testing conducted in the fall of 2006 by members of the Human Energy Research Laboratory and the team strength and conditioning staff. Because data were collected previously and personal player identifiers were removed, this project was deemed exempt by the University Institutional Review Board.
Standing height was measured to the nearest 0.1 cm using a wall-mounted, calibrated stadiometer (Holtain Limited, Crymtch, Dyfed, United Kingdom), and body mass was measured to the nearest 0.1 kg on a calibrated electronic scale (BOD POD). Both height and body mass were measured using standard procedures (17). Body volume was assessed via air displacement plethysmography using BOD POD version 1.69 (Life Measurement Inc. Concord, CA, USA). The system was calibrated with a cylinder of a known-volume before testing, and thoracic gas volume was measured in each participant using standard procedures. Each athlete's body density was calculated and converted to % fat using a modified Siri equation (28). As part of standard procedures, participants wore spandex clothing and Lycra cap and were requested not to eat or exercise for 3 hours before body composition testing.
Maximal aerobic fitness was assessed via indirect calorimetry using a discontinuous treadmill protocol performed to volitional exhaustion. Expired respiratory gases were measured continuously with a SensorMedics 2900 metabolic cart (Yorba Linda, CA, USA). Calibration of the metabolic cart was performed in accordance with the manufacturer's specifications. The protocol consisted of 3-minute stages with 90 seconds of rest between each stage. Stage 1 consisted of running at 161 m·min−1 on a level treadmill. Stage 2 consisted of running at 161 m·min−1 at 5% elevation. After stage 2, treadmill speed and grade increased 27 m·min−1 and 1%, respectively, for each successive stage. The protocol was terminated when a player achieved volitional exhaustion and could no longer continue, at which point time to exhaustion was recorded. O2max was considered to be the highest 1-minute O2 value achieved during the test.
Blood lactate samples were obtained between stages via finger prick (Yellow Springs Instruments, Yellow Springs, OH, USA). Heart rate (HR) was measured continuously using a 12 lead electrocardiography system (Max-1 Stress System, Milwaukee, WI, USA). Body mass for metabolic testing was obtained (with clothing and electrocardiograph pack) with a calibrated beam balance.
Strength tests were conducted by the team strength and conditioning staff at the university strength and conditioning center. All strength tests were performed either with body weight or using a predetermined weight, and participants performed as many repetitions as possible until volitional fatigue.
After a light warm-up set, participants performed a leg press test on a Hammer Strength Bilateral Leg Press using 400 lb (182 kg). The athlete performed as many repetitions as possible until volitional exhaustion was achieved. The number of repetitions performed was recorded. The athlete was required to lower the bar until the thigh was perpindicular to the floor and was allowed no more than 2seconds at the top of each repetition. One participant with exceptional lower body strength performed the test using 1,000 lb (455 kg) to avoid the possible injuries that may have occurred during the high amount of repetitions he could have performed before fatigue. Because of this discrepancy in methodology, his results were not included in analysis of this exercise.
After a light warm-up set, participants performed a bench press with a bar containing 155 lb (70 kg) until they could no longer continue with correct form. Each participant was required to lower the bar to his chest and then return to full extension. Four participants with pre-existing shoulder injuries were allowed to stop at 90° of elbow flexion. The number of repetitions performed was recorded.
Each athlete performed continuous reverse grip chin-ups until fatigue. The athlete was required to reach full extension at the elbows and pull up so that his chin cleared the top of the bar. The number of repetitions performed was recorded.
Maximum consecutive push-ups were performed on the toes until fatigue, and the number of repetitions performed was recorded. The athlete was required to lower until his chest made contact with a teammate's fist, which was located on the ground under the sternum and lift to full extension at the elbows.
Bench press, chin-ups, and push-ups were evaluated by 2 teammates, while a strength and conditioning coach observed the testing area. Leg press was directly observed and spotted by the strength coach and an assistant.
Athletes performed 12 consecutive 110-m land sprints. Players were led through a dynamic hip warm-up including 50-yd strides before testing. Testing was performed on an indoor turf field. The entire team was tested in the same session but split into 2 groups so that assistant coaches could record the time for each sprint. Subsequent sprint trials were started every 45 seconds so that players who completed the sprint the fastest had more rest time before beginning the subsequent sprint. Time to completion was recorded for each sprint and then averaged to obtain one score for each participant.
Athletes were allowed free-skate time including puck drills before on-ice testing but were not led through a structured warm-up. Beginning with their front foot on the face-off circle at one end of the ice, athletes skated to the face-off circle at the other end (Figure 1). Each participant performed the test twice with full gear and stick while time was recorded with a stopwatch. The fastest time was recorded and used for analysis.
Athletes skated from the near blue line, to the red line, near blue line, far blue line, red line, and finally through the far blue line (Figure 2). Each participant was allowed 2 trials, which were timed and observed by a strength coach to ensure that at least 1 foot touched each line. If a skater failed to touch a line before turning, the test was terminated and he was given another trial. He was also allowed another trial if he fell during the test. The fastest time for completion of the test was recorded.
Athletes began with the front foot on the center line and proceeded to skate around both goals and back through the center line (Figure 3). Participants were given the choice of skating clockwise or counterclockwise at their own discretion. There was no restriction to keep players in certain lanes along the straightaways, as they naturally drift toward the outside boards. Only 1 trial was performed, and time to completion was recorded.
Testing was conducted on 5 separate days. Day 1: Speed (110 m × 12), day 2: Bench press and leg press, day 3: Chin-ups and push-ups, day 4: On-ice testing, day 5: Anthropometics and O2max. To relate testing to the training program being used by the team, a sample workout is included (Figure 4).
The principal outcome measure of hockey performance in this study was the +/− score recorded for each player over the course of the season (23) Briefly, a player is credited with a plus every time he is on the ice and the team scores an even-strength or shorthanded goal. A player receives a minus every time he is on the ice and the other team scores an even-strength or shorthanded goal. Power play goals and penalty shot goals are not used in the +/− score. A player's overall +/− score is calculated by subtracting the minuses from the pluses. In general, players with a higher total are considered to be better players. An advantage of the +/− system is that it reflects both offensive and defensive effort and is not largely impacted by games missed due to illness, injury, or position played in the way that total goals scored might be.
A team assistant coach was asked to identify the top 6 players that he would “almost always” want on the ice for power plays or penalty kill situations. He was also asked to list the bottom 6 who he would “almost never” want to play on these occasions. Although these players represented the top and bottom quartiles (in this coach's opinion) in terms of overall skill level, the number 6 was chosen arbitrarily, and there was no required distribution among these groups for forwards and defensemen.
Players were ranked to denote their status within the team for their performance for each test with 1 denoting the best performance and 24 denoting the worst. Sample means and SDs were calculated for each variable. Pearson correlations were used to examine relationships between all variables, including +/− score, and Spearman Rank correlations were used to compare relationships between variable ranks. Stepwise regression analysis was performed using any variables that emerged as significant from correlational analysis to examine variance prediction. One-way analysis of variance was used to determine differences between top and bottom 6 player groups based on anthropometric and physiological variables. All analyses were performed using Predictive Analytics Software (PASW) Statistics 18 software with the alpha level set at p ≤ 0.05.
Anthropometric and physiological characteristics of participants are shown in Table 1. All variables of interest were normally distributed within our sample. The only variable that was significantly different between positions was O2max with forwards achieving a higher value compared to defensemen. There was also a trend for forwards to be faster than defensemen in the 12 × 110-m off-ice sprint (p = 0.07).
The Pearson correlation coefficients between anthropometric and on- and off-ice variables are shown in Table 2. Four variables were significantly correlated with absolute +/− score: 12 × 110-m sprint repeat (r = −0.568, p = 0.006); chin-ups (r = 0.462, p = 0.030), leg press (r = 0.554, p = 0.009), and bench press (r = 0.499, p = 0.021). No other variables were significantly correlated with +/− score. Although not correlated with +/− score, treadmill time to exhaustion (TM time) in the O2max test was significantly related to a number of other predictor variables (% body fat, O2max, HR4/HRMax, Lac4/LacMax, push-ups, leg press, and dot-to-dot test) (r = −0.769 to 0.616). There was some variation in the Pearson correlations by position. For forwards, only chin-up performance (r = 0.728, p = 0.007) was significantly correlated with +/− score. Among defensemen, +/− score was significantly correlated with body mass (r = 0.651, p = 0.041), fat-free mass (FFM, kg) (r = 0.682, p = 0.030), and bench press performance (r = 0.720, p = 0.029).
Table 3 shows the results of Spearman Correlations for rank scores of all participants. Although 12 × 100-m sprint repeat, chin-ups, leg press, and bench press were significantly correlated with +/− score during absolute analysis, only 12 × 110-m sprint repeat rank (r = 0.455, p = 0.033) and leg press rank (r = 0.488, p = 0.021) were significantly correlated with +/− rank within the team.
When variables significantly correlated with +/− score were entered into stepwise regression, chin-ups and repeat sprints emerged as the best predictors of +/− score, accounting for 49% of the variance. When examining variables by rank, repeat sprint, and leg press combined to account for 29% of the variance in +/− rank.
Differences in the physical characteristics and +/− score between top and bottom players as selected by a coach are shown in Table 4. Of the 6 top players, 5 were forwards, and 1 was a defensemen. Three of the bottom players were forwards, and 3 were defensemen. The only variable that exhibited significant differences between groups was +/− score (top = 7.5 [5.3] vs. bottom = −2.0 [4.8] [p = 0.009]). However, there appears to be a slight trend for the top 6 to also be younger (p = 0.11), heavier (p = 0.11), faster in the off-ice 12 × 110 m (p = 0/12), and possess greater leg strength (p = 0.19).
This study aimed to identify physical measures that are indicators of potential success in collegiate hockey players as designated by both an objective (+/−) and a subjective (coach grouping) outcome measure of season-long performance. The main finding was that the +/− score was significantly correlated with sprint repeat (r = −0.568), chin-up (r = 0.462), bench press (r = 0.499), and leg press (r = 0.554). When evaluated by team rank, only sprint repeat (r = 0.455) and leg press (r = 0.488) were significantly correlated with +/− score. Few previous studies have used the +/− score as the outcome measure for evaluating predictive ability of physical measures. In a previous study, our group (14) used a similar system of scoring chances and found O2max to be a significant predictor of performance. In this study, O2max was not correlated with +/− score or significantly different between top and bottom players. The difference in findings may be reflected in the outcome measure because +/− score is highly dependent on goalie ability as only goals impact the score. Scoring chances, on the other hand, is not directly impacted by goalie play.
Several studies (2,4,6,8,9,18) have found vertical jump and off-ice 40-yd dash to be significant predictors of on-ice maximum skating speed and acceleration time. However, these studies did not assess performance during actual games or over the course of the season. Because vertical jump and 40-yd dash were not assessed in this study, it is not possible to compare our results to these previous studies. We did however assess 3 on-ice skating tests. None of the skating tests were significantly correlated with +/−, and none of the physical measures were significantly correlated with the Lap Sprint test except for age among forwards (r = 0.734, p < 0.01). Fernandez et al. (9) found push-ups to be a significant predictor of acceleration ability. In this study, push-up performance was not correlated with any on-ice measures or with +/− score (r = 0.098). The lack of correlation between push-ups and +/− (0.098, p NS) is somewhat surprising because of the significant correlation between bench press and +/− score (r = 0.449) and the likelihood that push-ups and bench press used similar muscle groups. Therefore, lack of significant relationship may have been because all participants bench pressed the same amount of weight, whereas the work required for push-ups was dependent on body mass, giving larger players an advantage in the bench press.
Player position groups differed by which predictor variables were significantly correlated with +/− score. Among forwards only chin-ups was significantly correlated with +/− score (r = 0.728, p < 0.01), whereas weight (r = 0.651), FFM (r = 0.682), and bench press performance (r = 0.72, p < 0.05) were significantly correlated with +/− score in defensemen. The significant correlations for weight and FFM among defensemen may highlight the importance of body size for this position.
Few differences were found between the top and bottom 6 players as designated by a coach. Although others (11,15) have found “better” players to be older and stronger (as reflected by squat and bench press tests), findings in body mass and body composition are less decisive. Geithner et al. (11) found elite female players to be lighter and leaner than their counterparts, whereas Hoff et al. (15) found more successful players to be heaver with no difference in height or body fat percentage. The lack of significant differences between the top and bottom players in this study may be because of the small number of participants in each group. Only +/− score was significantly different between these 2 groups. This suggests that, in this specific group of players, the +/− system is an accurate reflection of the coach's perception of talent and contribution to the team.
Similar to this study, Agre et al. (1) found no differences between professional forwards and defensemen in height, weight, or body fat percentage. However, others (16,25) have found defensemen to be significantly taller and heavier than forwards. Although not statistically significant, defensemen in this study were actually shorter and weighed less than forwards. However, the small sample size in this study may have precluded the finding of statistically significant differences. Previous studies have also characterized aerobic and anaerobic ability of elite hockey players (20). The mean O2max of almost 59 ml·kg−1·min−1 in this sample is higher than reported in previous studies (1,5,29). However, many studies were performed several years ago and it appears that that O2max levels among elite hockey players may be increasing (5). O2max has also been shown to be related to recovery ability (9), an important aspect in the game of hockey. Although recovery ability was not directly measured in this study, performance in the repeat sprint test was significantly correlated with O2max both in absolute (r = −0.536, p < 0.01) and rank (r = 0.593, r < 0.05) tests. Because the work-rest-repeat nature of this test, it may be viewed as an index of recovery ability.
There are some limitations and methodological considerations in this study. The first lies in the main outcome measure used here-the +/− score. There is debate in the hockey community over the efficacy of the +/− system because of the influence that a goalie can have on a player's score. If a team has a “hot” goalie, it is very beneficial to the +/− score as the opposing team is less likely to score goals. Conversely, if a team is playing against a very good goalie, the team is less likely to score and accumulate plusses. The scoring chances system used by Green et al. (14) does not tally goals, but instead considers situations that may lead to potential goals, and therefore is less influenced by the goalie and may better reflect the talent of individual players and the team as a whole. Unfortunately, the scoring chance statistic could not be used in this analysis because it was not recorded for this specific season. The findings are somewhat limited by the small sample size, which was inherently limited by the fact that we examined only 1 team for 1 year. However, the unique aspect of this sample is that they were the 2007 NCAA national champions. Future studies should examine physical characteristics and the +/− or scoring chances for more players over the course of many seasons.
Future studies should include larger samples and when possible, more sensitive outcome measures. Compiling data from many seasons within the same program would allow for a larger sample while maintaining the general characteristics of this study. Summation of multiple seasons may also allow for an opportunity to compare the +/− system to other means of evaluating performance across a season, such as scoring chances. Although data for scoring chances were not available during the season in this study, it is often measured within this program and comparing results via the 2 systems would provide an insight into the appropriateness of the +/− system. Finally, inclusion of tests for 40-yd dash and vertical jump would allow for the ability to compare future findings with results of past studies as these are common measurements in studies of hockey players.
Results of this study suggest that game performance as indicated by the +/− score can be predicted by measures of strength and repeat sprint ability and that the +/− system has the ability to discriminate elite hockey players from their subelite teammates. Based on these findings, measures of strength and sprint ability appear to be significant predictors of hockey performance. Aerobic fitness and body composition do not appear to be significant predictors of player performance as measured by the +/− system or coach evaluation. To maximize the efficiency of preseason testing, coaches may rely on strength (chin-ups, leg press, and bench press) and repeat sprint tests while decreasing the number of aerobic capacity and body composition analyses to minimize player burden; however, these measures have other utility for the coaching and sports medicine staff. There does appear to be an increased importance of body size among defensemen suggesting that this may also be an important measurement among this group of players. Results from this study may be beneficial to coaches, trainers, and athletes in developing screening criteria to identify and monitor potentially elite hockey players. By knowing which tests are most predictive of season performance, coaches can choose to use these tests while excluding others, leaving more time for conditioning and on-ice skill and tactical development.
1. Agre, JC, Casal, DC, Leon, AS, McNallay, C, Baxter, TL, and Serfass, RC. Professional ice hockey players: Physiologic, anthropometric, and musculoskeletal characteristics. Arch Phys Med Rehabil
69: 188-192, 1998.
2. Behm, DG, Wahl, MJ, Button, DC, Power, KE, and Anderson, KG. Relationship between hockey skating speed and selected performance measures
. J Strength Cond Res
19: 326-331, 2005.
3. Bracko, MR, Fellingham, GW, Hall, LT, Fisher, AG, and Cryer, W. Performance skating characteristics of professional ice hockey forwards. Res Sports Med
8: 251-263, 1998.
4. Bracko, MR and George, JD. Prediction of ice skating performance with off-ice testing in women's ice hockey players. J Strength Cond Res
15: 116-122, 2001.
5. Cox, MH, Miles, DS, Verde, TJ, and Rhodes, EC. Applied physiology of ice hockey. Sports Med
19: 184-201, 1995.
6. Diakoumis, K and Bracko, MR. Prediction of skating performance with off-ice testing in deaf hockey players. Med Sci Sports Exerc
30: 272, 1998.
7. Douglas, FGV and Becklake, MR. Effect of seasonal training on maximal cardiac output. J Appl Physiol
25: 600-605, 1968.
8. Farlinger, CM, Kruisselbrink, D, and Fowles, JR. Relationships to skating performance in competitive hockey players. J Strength Cond Res
21: 915-922, 2007.
9. Fernandez, DH, Geithner, CA, Haia, CLK, and Bracko, MR. Prediction of skating performance using anthropometry and off-ice fitness tests in elite women's ice hockey. Med Sci Sports Exerc
40: S164, 2008.
10. Game, AB and Bell, GJ. The effect of a competitive season and environmental factors on pulmonary function and aerobic power in varsity hockey players. Appl Physiol Nutr Metab
31: 95-100, 2006.
11. Geithner, CA, Fernandez, DH, Haia, CLK, and Bracko, MR. Profiling of nationally ranked female ice hockey players by multidimensional performance and fitness characteristics. Med Sci Sports Exerc
40: S381, 2008.
12. Green, H, Bishop, P, Houston, M, McKillop, R, Norman, R, and Stothart, P. Time-motion and physiological assessments of ice hockey performance. J Appl Physiol
40: 159-163, 1976.
13. Green, HJ and Houston, ME. Effect of a season of ice hockey on energy capacity and associated functions. Med Sci Sports Exerc
7: 299-303, 1975.
14. Green, MR, Pivarnik, JM, Carrier, DP, and Womack, CJ. Relationship between physiological profiles and on-ice performance of a national collegiate athletic association division I hockey team. J Strength Cond Res
20: 43-46, 2006.
15. Hoff, J, Kemi, OJ, and Helgerud, J. Strength and endurance differences between elite and junior elite ice hockey players. The importance of allometric scaling. Int J Sports Med
26: 537-541, 2005.
16. Houston, ME and Green, HJ. Physiological and anthropometric characteristics of elite Canadian ice hockey players. J Sports Med Phys Fitness
16: 123-128, 1976.
17. Malina, R. Anthropometry: Physiological Assessment of Human Fitness
. C. Foster. and P.J. Maud, ed. Champaign, IL: Human Kinetics, 1995.
18. Mascaro, T, Seaver, BL, and Swanson, L. Prediction of skating speed with off-ice testing in professional hockey players. J Orthop Sports Phys Ther
15: 92-98, 1992.
19. Minkoff, J. Evaluating parameters of a professional hockey team. Am J Sports Med
10: 285-292, 1982.
20. Montgomery, DL. Physiology of ice hockey. Sports Med
5: 99-126, 1988.
21. Montgomery, DL. Physiological profile of professional hockey players-A longitudinal comparison. Appl Phsyiol Nutr Metab
31: 181-185, 2006.
22. Montgomery, DL, Nobes, K, Pearsall, DJ, and Turcotte, RA. Task analysis (hitting, shooting, passing and skating) of professional hockey players. In: Safety in Ice Hockey: Fourth Volume, ASTM STP 14446
. D.J. Pearsall and A.B. Ashare, eds. West Conshohocken, PA: ASTM International, 2004.
23. Plusminus. Plusminus.edu. A European informational website. Available at: http://plusminus.eu
. Retrieved April 8, 2010.
24. Potteiger, JA, Smith, SL, Maier, M, and Foster, T. Relationship between laboratory testing and on-ice skating performance in division I men's hockey athletes [Abstract]. Med Sci Sports Exerc
41: 145-146, 2009.
25. Quinney, HA, Dewart, R, Game, A, Snydmiller, G, Warburton, D, and Bell, G. A 26 year physiological description of a National Hockey League team. Appl Physiol Nutr Metab
33: 753-760, 2008.
26. Rodrigues, NR, Di-Marco, NM, and Langley, S. American College of Sports Medicine position stand. Nutrition and athletic performance. Med Sci Sport Exerc
41: 709-731, 2009.
27. Rovio, E, Eskola, J, Kozub, SA, Duda, JL, and Lintunen, T. Can high group cohesion be harmful? A case study of a junior ice-hockey team. Small Group Res
40: 421-435, 2009.
28. Siri, WE. Body composition from fluid spaces and density: Analysis of methods. In: Techniques for Measuring Body Composition
. J. Brozek and A. Henschel, eds. Washington, DC: National Academy of Science, 1961. pp. 223-224.
29. Smith, DJ, Quinney, HA, Steadward, RD, Wenger, HA, and Sexsmith, JR. Physiological profiles of the Canadian Olympic Hockey Team (1980). Can J Appl Sport Sci
7: 142-146, 1982.
30. Vescovi, JD, Murray, TM, and Venheest, JL. Positional performance profiling of elite ice hockey players. Int J Sports Physiol Perform
1: 84-94, 2006.
31. Voyer, D and Wright, EF. Predictors of performance in the national hockey league. J Sport Behav
21: 456-473, 1998.