- Email: [email protected]
Match-to-match variation in physical activity and technical skill measures in professional Australian Football
G Model JSAMS-981; No. of Pages 5
ARTICLE IN PRESS Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Journal of Science and Medicine in Sport journal homepage: www.elsevier.com/locate/jsams
Original research
Match-to-match variation in physical activity and technical skill measures in professional Australian Football Thomas Kempton a , Courtney Sullivan a,b , Johann C. Bilsborough b , Justin Cordy a , Aaron J. Coutts a,b,∗ a b
Sport & Exercise Discipline Group, Faculty of Health, University of Technology Sydney (UTS), Australia Carlton Football Club, Australia
a r t i c l e
i n f o
Article history: Received 19 June 2013 Received in revised form 22 November 2013 Accepted 20 December 2013 Available online xxx Keywords: GPS Time-motion analysis Performance analysis Reliability Football
a b s t r a c t Objectives: To determine the match-to-match variability in physical activity and technical performance measures in Australian Football, and examine the influence of playing position, time of season, and different seasons on these measures of variability. Design: Longitudinal observational study. Methods: Global positioning system, accelerometer and technical performance measures (total kicks, handballs, possessions and Champion Data rank) were collected from 33 players competing in the Australian Football League over 31 matches during 2011–2012 (N = 511 observations). The global positioning system data were categorised into total distance, mean speed (m min−1 ), high-speed running (>14.4 km h−1 ), very high-speed running (>19.9 km h−1 ), and sprint (>23.0 km h−1 ) distance while player load was collected from the accelerometer. The data were log transformed to provide coefficient of variation and the between subject standard deviation (expressed as percentages). Results: Match-to-match variability was increased for higher speed activities (high-speed running, very high-speed running, sprint distance, coefficient of variation %: 13.3–28.6%) compared to global measures (speed, total distance, player load, coefficient of variation %: 5.3–9.2%). The between-match variability was relativity stable for all measures between and within AFL seasons, with only few differences between positions. Higher speed activities (high-speed running, very high-speed running, sprint distance), but excluding mean speed, total distance and player load, were all higher in the final third phase of the season compared to the start of the season. Conclusions: While global measures of physical performance are relatively stable, higher-speed activities and technical measures exhibit a large degree of between-match variability in Australian Football. However, these measures remain relatively stable between positions, and within and between Australian Football League seasons. © 2013 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.
1. Introduction In the Australian Football League (AFL), most teams now routinely collect time motion data during match-play using global positioning system (GPS) devices. In addition, information on a range of technical involvements is also available from commercial statistical providers. Both physical and technical data obtained during competition are commonly used in research to provide an understanding of the factors that affect Australian Football performance. For example, recent research in Australian Football has examined differences in physical and technical performance profiles of individual players depending on match outcome,1 match
∗ Corresponding author. E-mail address: [email protected] (A.J. Coutts).
location and time of day,2 coaches’ perception of performance,3 and fatigue status.4 From a practical perspective, sport scientists and coaches also use the physical activity and technical performance information to gain a greater insight into an individual player’s performance during a match and to assess changes during a season or playing career. However, despite the wide applications of these data, there is presently a poor understanding of the typical variability in these measures. A better understanding of the typical variation in these measures can be useful for designing applied research studies, selecting reliable performance measures for applied studies and to assist interpreting worthwhile changes in performance. Australian Football match-play is dynamic with a complex interplay between individuals on both the same and opposing teams. As such, common measures of physical performance—such as variables in higher speed zones—are not stable and can show large
1440-2440/$ – see front matter © 2013 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jsams.2013.12.006
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
G Model JSAMS-981; No. of Pages 5
ARTICLE IN PRESS T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
2
variation between matches. The sources of variation of physical performance within individual players over successive matches may be related to internal factors (e.g. fitness status, motivation) and external factors (e.g. opposition, tactics, measurement system and environment).5–7 Previous research has reported that measures such as total distance and mean speed (m min−1 ) are relatively stable, while higher-speed activities show higher variability between matches in both soccer7,8 and rugby league.9 Others have reported that the magnitude of the variability between matches may also be influenced by playing position.8 Finally, while Gregson et al.8 found no clear pattern of change in variability of physical performance measures between seasons, others have reported changes in physical performance profiles within professional soccer competition seasons.5,7 While the match-to-match variation in physical activity measures has been reported for soccer7,8 and rugby league,9 there have been no studies which have investigated the variability of physical performance during the AFL competition. It is important to examine the variability of these measures specific to AFL competition due to differences related to field size and player number between Australian Football and other sports—such as soccer and rugby league—where previous investigations have determined between match-variation. Specifically, the greater pitch area available per player may explain the increased higher-speed activities in Australian Football compared to soccer and rugby league.10,11 As high-speed activities display large variation during team sport match-play, the increased amount of high-speed activity in Australian Football may predicate greater variability in physical performance measures compared to previous studies in soccer and rugby league.7–9 Additionally, there have been no studies in any football code that have examined the variability of key individual technical performance measurements. This is somewhat surprising considering that recent research in both soccer12 and Australian Football1 has highlighted the importance of technical performance to overall match outcome. Furthermore, research has shown that certain technical actions can differentiate between higher and lower calibre AFL players.3 Indeed, it is likely that key technical performance metrics vary between matches and this may be related to both individual player characteristics (e.g. experience, technical ability) and fluctuations in game-related factors (e.g. match outcome, strength of opposition). Further research is therefore warranted to investigate the between-match variation of common technical performance metrics in AFL match-play. Accordingly, the aims of this study were to: (1) determine the match-to-match variability for measures of physical and technical performance during successive seasons of Australian Football; (2) examine the influence of playing position on these variables; and (3) assess changes in physical performance within competition seasons.
included for analysis. To assess the influence of the stage of the season on variation, the regular season was divided into three, 8week blocks representing the start (n = 11 games), middle (n = 11 games) and end (n = 9 games). Informed consent and institutional ethics approval were obtained. Physical activity profiles were measured using portable GPS units (Team Sport 2.5, Firmware 6.54, Catapult Innovations, 10 Hz Melbourne, Australia). The unit was fitted within a custom made pouch in the playing jumper of each player prior to the match. Following each match, data were downloaded using proprietary software (Catapult Sprint v5.0.6). All match files were trimmed so that only data obtained during each quarter was retained for further analysis with the proprietary software. The data was categorised into total distance, relative total distance (m min−1 ), high-speed (>14.4 km h−1 ) and very-high speed (>19.9 km h−1 ) running distance, and sprint (>23.0 km h−1 ) distance. Finally, an inbuilt algorithm was used to provide a measure of “Player Load” during each match which has been reported elsewhere.13 The reliability of physical activity measures used in the present study had been previously reported.13–15 The total kicks (number of times a player disposes of the ball by foot), handballs (number of times a player disposes of the ball by hand), and possessions (number of times a player receives the ball) and a player rank score were obtained from a commercial statistics provider (Champion Data© , Victoria, Australia.1 The Champion Data (CD) player rank is an objective measure provided by an algorithm (Champion Data© , Victoria, Australia) which is weighted towards effective ball use and winning disputed possessions and also considers the contribution—both positive and negative—of a range of individual game-specific actions to the overall match outcome. The between match variation in physical and technical performance measures were analysed using a customised Microsoft Excel spreadsheet (Microsoft, Redmond, USA).16 The data were log transformed and the coefficient of variation (%CV) was calculated (i.e. the typical error expressed as a percentage of the mean score).17 The smallest worthwhile change (SWC) in performance for each variable was also obtained by multiplying the between subject standard deviation by 0.2.18 For within-season changes in performance and differences in on field playing time, a magnitude based approach was used to assess the chances of true differences (i.e. greater than the SWC). Quantitative chances of real differences in variables were assessed qualitatively as: <1%, almost certainly not; 1–5%, very unlikely; 5–25%, probably not; 25–75%, possibly; 75–97.5%, likely; 97.5–99%, very likely; >99%, almost certain.18 Standardised effect sizes (ES) were interpreted as <0.2, trivial; 0.2–0-6, small; 0.6–1.2, moderate; 1.2–2.0, large; >2.0, very large.18 Data are presented with 95% confidence intervals (CI).
3. Results 2. Methods Physical and technical performance data were collected from 33 players (age: 23.2 ± 1.8 y; mass: 87.3 ± 7.6 kg; stature: 188 ± 7 cm) from the same club over two successive AFL seasons. The players were classified into three positional groups: backs, midfields or forwards. We excluded ‘ruck players’ from our analysis due to insufficient sample size arising from the highly specialised role that they perform. A total of 511 individual match files were obtained from 31 official regular season competition matches played at outdoor stadiums during the observation period. Due to the unlimited interchange of players during AFL competition match play, it is uncommon for an individual player to complete an entire match. To reduce the effect of playing time on variation, only data from players who were not injured or substituted during the match were
The %CV and SWC of physical and technical performance measurements for positional groups are reported in Table 1. The data show that the variability of physical performance parameters increases as movement classification speed also increases. Lower variability was observed for mean speed compared to total distance for all positions. There was generally high match-to-match variability for all technical performance variables. There were very likely and almost certainly small differences in time on field between midfields and forwards (−3.2; −4.8 to −1.5 min), and midfields and backs (−4.0; −5.4 to −2.5 min), respectively. The variation in physical performance parameters was similar over the 2011 and 2012 competition seasons (Table 2). There was a possibly trivial difference in time on field between season 2011 and season 2012 (1.4; 0.1–2.7 min).
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
ARTICLE IN PRESS
G Model JSAMS-981; No. of Pages 5
T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
3
Table 1 Between match variation of physical and technical measures for positional groups (%CV and 95% CI; %SWC). Backs
Midfields
Forwards
Overall
N = 175
N = 216
N = 120
N = 511
%CV
%SWC
%CV
%SWC
%CV
%SWC
%CV
%SWC
Physical Total distance (m) Speed (m min−1 ) HSR (m) VHSR distance (m) Sprint distance (m) Sprints (n) Body Load (AU)
7.0 (6.1–8.3) 6.4 (5.5–7.5) 13.8 (11.9–16.4) 20.0 (17.3–24.0) 30.4 (26.1–36.6) 24.0 (20.7–28.8) 7.2 (6.3–8.6)
1.9 2.0 4.9 7.3 10.8 8.5 3.3
5.6 (5.0–6.4) 4.5 (4.0–5.2) 11.9 (10.5–13.7) 20.9 (18.5–24.2) 29.3 (26.3–33.3) 23.8 (21.4–27.0) 10.5 (9.4–12.1)
1.4 1.5 3.7 6.4 10.4 8.5 3.8
4.9 (4.2–6.1) 4.8 (4.0–5.9) 11.7 (9.9–14.5) 15.1 (12.8–18.8) 23.9 (20.1–30.1) 25.3 (21.3–31.9) 9.5 (8.1–11.8)
2.5 2.2 6.1 9.9 14.7 12.7 4.6
6.1 (5.7–6.6) 5.3 (4.9–5.7) 13.2 (12.2–14.3) 19.6 (18.1–21.4) 28.6 (26.3–31.3) 24.4 (22.5–26.6) 9.2 (8.5–10.0)
2.0 2.1 5.4 7.5 11.5 9.5 3.9
Technical Possessions (n) Kicks (n) Handballs (n) CD Rank (AU)
32.7 (28.1–39.5) 51.6 (43.9–63.1) 63.3 (53.2–78.6) 37.1 (31.7–44.9)
7.4 11.1 15.4 7.8
23.8 (21.0–27.6) 33.7 (29.6–39.4) 44.2 (38.7–51.9) 26.4 (23.3–30.7)
5.6 8.6 12.3 6.4
31.5 (26.3–39.9) 42.9 (35.6–54.9) 62.4 (50.9–81.1) 31.7 (26.5–40.1)
7.3 10.2 15.2 7.7
28.3 (26.1–31.0) 42.7 (39.2–47.0) 55.3 (50.6–61.2) 31.5 (29.0–34.5)
10.1 11.3 19.4 9.1
HSR: high-speed running, VHSR: very high-speed running, CD Rank: champion data rank.
Global measures of match activity—total distance, mean speed and player load—were stable throughout the season (Table 3). However, there was a possibly small increase in HSR, and a likely small increase in VHSR, sprint distance and sprint number at the end of the season compared to the start. There was a likely small increase in time on field at the end of the season compared to the start (2.8; 1.3–4.4 min). 4. Discussion This study examined the match-to-match variation of physical and technical performance parameters during professional AFL competition. The main findings were that global measures of physical performance were relatively stable; however higher speed activities and technical performance measures displayed greater variability. The variability of physical performance measures was similar over two successive competition years; although there were some within-season changes in higher-speed activities. These findings have important implications for interpreting individual player’s physical and technical performance during matches and examining temporal changes in match performance. Total distance was the most stable physical activity measure in the present investigation. Similar findings have been reported in soccer7,8 and rugby league,9 although the overall variation in the present data was somewhat higher than these investigations. The variability in total distance was reduced when data were expressed relative to time spent on field (m min−1 ). This approach may be useful for reducing variability when comparing match samples of differing duration as is common in Australian Football. In contrast, the higher speed activities were more variable (%CV 13.2–28.6) compared to both total distance and m min−1 , and there was an increase in the variability of these measures with increases in the running speed. This trend has also been reported in previous studies of soccer7,8 and rugby league9 and is likely associated with poorer reliability of measurement devices at higher velocities.14,15,19 Indeed, the variation in HSR in the present
study was similar to previous research in rugby league which also used GPS devices, although both studies have higher CVs than reported previously in soccer,7 which employed semi-automated computer based tracking technology. Differences in both the timemotion measurement systems and in the nature of soccer, rugby league and Australian Football—relating to rules, size of playing field and running demands—may explain these results.20,21 While sprint distance was the least stable of all physical activity measures, sprint number showed less variability and may therefore be a more suitable marker of sprint performance. Nonetheless, sprint number was still quite variable and in contrast to our findings, others have reported similar variability as total sprint distance in soccer.8 A further limitation is that sprint number does not account for the speed or distance of sprint efforts and as such may be less representative of sprint performance. Since good measurement reproducibility is important for assessing small but practically important changes in performance measures in individual players, caution is required when using sprint or other higher speed activities for interpreting changes in an individual players match activity profile. The match-to-match variation in most activity profile measures were similar between positional groups, although there were some small differences, with forwards showing less variability in match performance measures compared to midfields and backs. The specific roles and constraints imposed by playing position may be partially responsible for the differences in variability observed. Indeed, while all Australian Football players are able to roam around the entire field during matches, forwards tend to remain in the offensive half of the field. Furthermore, team tactics influence game structures with many teams adopting ‘running patterns’ whereby forward players perform pre-arranged leading patterns to create scoring opportunities. It may be that reducing the area of the field that players usually work within during matches and running pattern repetition reduces the variability of measurement in higher speed activities. Additionally, the diverse tactical roles of the backs and midfielders may partly explain their greater variability in activity profile during matches.
Table 2 Between match variation in physical parameters during seasons 2011 and 2012 (%CV and 95% CI).
2011 (B = 86; M = 108; F = 64) 2012 (B = 89; M = 108; F = 56)
Total distance (m)
Speed (m min−1 )
HSR distance (m)
VHSR distance (m)
Sprint distance (m)
Sprints (n)
Player load (AU)
5.6 (5.0–6.3)
4.7 (4.2–5.3)
12.7 (11.4–14.4)
20.0 (17.9–22.7)
29.2 (26.1–33.4)
24.9 (22.2–28.3)
9.4 (8.5–10.7)
6.6 (5.9–7.4)
5.9 (5.3–6.6)
13.5 (12.1–15.3)
18.6 (16.6–21.1)
27.0 (24.0–30.9)
23.5 (20.9–26.8)
8.8 (7.9–9.9)
B, backs; M, midfields; F, forwards; HSR, high-speed running, VHSR, very high-speed running.
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
ARTICLE IN PRESS
G Model JSAMS-981; No. of Pages 5
T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
1246 (1215–1276) 19.8 (18.7–21.0) 459 (427–490) 124 (122–127) 13,193 (12988–13398)
B, backs; M, midfields; F, forwards; HSR, high-speed running; VHSR, very high-speed running. a Likely small difference to end of season. b Possible small difference to end of season.
1145 (1089–1200)
1056 (1003–1109)
3492 (3347–3638)
125 (123–127) 13,199 (13001–13397)
3696 (3552–3840)
1266 (1236–1309) 18.2 (17.1–19.3) 431 (401–461)
1263 (1225–1301) 17.9 (16.9–18.9) 405 (379–431)
b
1023 (977–1070)
b
123 (121–125)
Start of season (B = 62; M = 78; F = 43) Middle of season (B = 59; M = 76; F = 42) End of season (B = 54; M = 62; F = 35)
12,948 (12775–13120)
3462 (3336–3589)
Sprint distance (m)
a a
VHSR distance (m) HSR distance (m) Speed (m min−1 ) Total distance (m)
Table 3 Seasonal variations in physical performance parameters (mean and 95% CI).
b
Sprints (n)
a
Player Load (AU)
4
The variability of match activity measures was relatively constant over the two successive seasons, with some variables increasing slightly, while others were reduced. This finding supports previous research which has failed to find a consistent pattern of change in higher-speed activities over three seasons of professional soccer competition.8 While the variability in physical performance measures did not change between seasons, we observed small changes in the profiles of higher-speed, but not global, parameters during the course of a season. Specifically, HSR, VHSR, sprint distance and sprint number were all increased at the end of the season compared with the start. These results are in agreement with previous research in soccer,5,7 and may be related to changes in team tactics, improved fitness or specific adaptations to match-play during the season.7 Indeed, previous research has shown a case study of increased physical activity profile in a team during finals matches compared to home and away games.22 It is also possible that a greater importance is placed on matches approaching finals, which may contribute to the increased activity profile towards the end of the season. Regardless of the cause, it is important to consider this trend when interpreting temporal changes in performance during an AFL season. The technical performance measures examined in this study displayed high levels of overall variability (%CV 28.3–55.3). This is a novel aspect of the study and has important implications given that recent applied studies of AFL competition have employed these variables as performance indicators.1–3 For example, Sullivan et al.1 reported that several technical performance measures including total kicks and CD rank were all higher during quarters won compared to quarters lost. Similarly, others have shown that the frequency of some technical involvements differentiates between higher and lower-calibre players.3 These studies are in agreement with previous soccer research which has shown that certain technical actions differ between players on more and less successful teams.12 Collectively, these studies demonstrate the importance of technical involvements for overall match performance during match-play. While it appears that technical variables such as CD rank and kicks are sensitive performance measures, caution is required when using these parameters to interpret a player’s performance as they display higher match-to-match variability. The very high level of variability of technical performance of individual players may be due to internal (experience, technical ability and decision making) and external (strength of opposition, team tactics and the “flow of play”) factors. The precision of a performance indicator—either physical or technical—is an important consideration for applied research investigations.8 Indeed, HSR has been suggested as a significant physical performance parameter in soccer due to its association with “the play” during a match, ability to discriminate between playing levels and apparent sensitivity to match related fatigue.5 Similarly, recent research has shown that CD rank is a suitable measure of Australian Football performance due to its associations with match outcomes.12 While it is essential that performance variables for applied research projects are selected based on their relationships with important match characteristics, it is also critical that the measurement error of these variables is lower than the match-to-match variation. Further research is required to assess the reliability of technical measures used in the present investigations to determine their appropriateness as indicators of performance.8,18 In these cases, a large sample size may be required to overcome these limitations for detecting performance changes.23 For example, using the framework provided by Batterham and Atkinson, a variable with a CV of 30% and SWC 10%—such as overall possessions or overall CD rank—would require approximately 180 samples in order to detect a statistically significant change in performance.23
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
G Model JSAMS-981; No. of Pages 5
ARTICLE IN PRESS T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
While this study examined a large (∼500 match samples) data set obtained over two seasons, this is still smaller than previous soccer research.8 Indeed, the comparatively smaller sample size in this investigation is a limitation and future studies are required with more comprehensive data sets to provide estimates of the true variability of these measures in AFL match-play. In addition, the use of only absolute speed thresholds for determining higher-speed distances may be a limitation in the present study as some have suggested that individualised speed thresholds may complement the traditional arbitrary approach.24,25 Finally, there are many factors which may contribute to the variability of match performance, for example, opposition, match outcome, match location, time of season, physical capacity of players and environmental conditions.7,8 We have not accounted for these factors directly in the present study and future research should examine the independent effects of these influences. 5. Conclusion This study has examined the typical variability of common physical and technical performance parameters for positional groups, within a season and between two seasons of the AFL competition. The results demonstrate that match-to-match variation of the physical activity measures increased with higher speed activities. The match-to-match variability for all technical performance variables was higher than the physical activity measures. However, the variability in total distance, mean speed and player load remained relatively stable within and between seasons. In contrast, HSR, VHSR, sprint distance and sprint number were higher at the end of the season compared to the start. These findings have important implications for interpreting changes in the match activity profile and technical performance of both individual players and the team and for examining temporal changes in match performance. Practical implications • Due to the potential for high levels of variability for many common measures used in Australian Football match analysis studies, it is recommended that researchers establish and report CV’s for their data set. • The low match-to-match variation in total distance in Australian Football demonstrates that this is a relatively stable performance measure; however, the variability for higher speed activities and technical measures is much greater. • The increase in activity profile in Australian Football towards the end of the season should be taken into account when planning individual and team training loads. • Statistical approaches which use the CV may assist in designing applied research studies and interpreting the magnitude of change. Acknowledgements We would like to thank Ermanno Rampinini for his feedback on the manuscript. No external sources of funding were provided
5
for this study. The authors have no relevant conflicts of interest to declare. References 1. Sullivan C, Bilsborough J, Hocking J et al. Match score affects activity profile and skill performance in professional Australian Football players. J Sci Med Sport 2013. http://dx.doi.org/10.1016/j.jsams.2013.05.001. 2. Hiscock D, Dawson B, Heasman J et al. Game movements and player performance in the Australian Football League. Int J Perfe Analysis Sport 2012; 12(3):531–545. 3. Johnston R, Watsford M, Pine M et al. Movement demands and match performance in professional australian football. Int J Sports Med 2012; 33(2):89. 4. Mooney M, Cormack S, O’Brien B et al. Impact of neuromuscular fatigue on match exercise intensity and performance in elite Australian football. J Strength Cond Res 2013; 27(1):166–173. 5. Mohr M, Krustup P, Bangsbo J. Match performance of high-standard soccer players with special reference to development of fatigue. J Sports Sci 2003; 21:519–528. 6. Mohr M, Mujika I, Santisteban J et al. Examination of fatigue development in elite soccer in a hot environment: a multi-experimental approach. Scand J Med Sci Sports 2010; 20(3):125–132. 7. Rampinini E, Coutts A, Castagna C et al. Variation in top level soccer match performance. Int J Sports Med 2007; 28(12):1018–1024. 8. Gregson W, Drust B, Atkinson G et al. Match-to-match variability of highspeed activities in Premier League soccer. Int J Sports Med 2010; 31: 237–242. 9. Kempton T, Sirotic A, Coutts A. Between match variation in professional rugby league competition. J Sci Med Sport 2013. http://dx.doi.org/10.1016/j.jsams.2013.05.006. 10. Kennett D, Kempton T, Coutts A. Factors affecting exercise intensity in rugby-specific small-sided games. J Strength Cond Res 2012; 26(8): 2037–2042. 11. Varley M, Gabbett T, Aughey R. Activity profiles of professional soccer, rugby league and Australian football match play. J Sports Sci 2013: 1–9. 12. Rampinini E, Impellizzeri F, Castagna C et al. Technical performance during soccer matches of the Italian Serie A league: effect of fatigue and competitive level. J Sci Med Sport 2009; 12(1):227–233. 13. Boyd LJ, Ball K, Aughey RJ. The reliability of MinimaxX accelerometers for measuring physical activity in Australian football. Int J Sports Physiol Perf 2011; 6(3):311–321. 14. Varley MC, Fairweather IH, Aughey RJ. Validity and reliability of GPS for measuring instantaneous velocity during acceleration, deceleration, and constant motion. J Sports Sci 2012; 30(2):121–127. 15. Jennings D, Cormack S, Coutts AJ et al. The validity and reliability of GPS units for measuring distance in team sport specific running patterns. Int J Sports Physiol Perform 2010; 5(3):328–341. 16. Hopkins WG. Precision of measurement, In: A new view of statistics., 2011, newstats.org/precision.html. 17. Hopkins WG. Measures of reliability in sports medicine and science. Sports Med 2000; 30(1):1–15. 18. Hopkins W, Marshall S, Batterham A et al. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc 2009; 41(1):3–12. 19. Coutts AJ, Duffield R. Validity and reliability of GPS units for measuring movement demands of team sports. J Sci Med Sport 2010; 13(1):133–135. 20. Harley JA, Lovell RJ, Barnes CA et al. The interchangeability of global positioning system and semiautomated video-based performance data during elite soccer match play. J Strength Cond Res 2011; 25(8):2334–2336. 21. Randers MB, Mujika I, Hewitt A et al. Application of four different football match analysis systems: a comparative study. J Sports Sci 2010; 28(2): 171–182. 22. Aughey RJ. Increased high-intensity activity in elite Australian football finals matches. Int J Sports Physiol Perf 2011; 6(3):367–379. 23. Batterham A, Atkinson G. How big does my sample need to be? A primer on the murky world of sample size estimation. Phys Ther Sport 2005; 6:153–163. 24. Abt G, Lovell R. The use of individualised speed and intensity thresholds for determining the distance run at high-intensity in professional soccer. J Sports Sci 2009; 29(9):893–898. 25. Lovell R, Abt G. Individualization of time-motion analysis: a case–cohort example. Int J Sports Physiol Perform 2013; 8(4):456–458.
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
ARTICLE IN PRESS Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Journal of Science and Medicine in Sport journal homepage: www.elsevier.com/locate/jsams
Original research
Match-to-match variation in physical activity and technical skill measures in professional Australian Football Thomas Kempton a , Courtney Sullivan a,b , Johann C. Bilsborough b , Justin Cordy a , Aaron J. Coutts a,b,∗ a b
Sport & Exercise Discipline Group, Faculty of Health, University of Technology Sydney (UTS), Australia Carlton Football Club, Australia
a r t i c l e
i n f o
Article history: Received 19 June 2013 Received in revised form 22 November 2013 Accepted 20 December 2013 Available online xxx Keywords: GPS Time-motion analysis Performance analysis Reliability Football
a b s t r a c t Objectives: To determine the match-to-match variability in physical activity and technical performance measures in Australian Football, and examine the influence of playing position, time of season, and different seasons on these measures of variability. Design: Longitudinal observational study. Methods: Global positioning system, accelerometer and technical performance measures (total kicks, handballs, possessions and Champion Data rank) were collected from 33 players competing in the Australian Football League over 31 matches during 2011–2012 (N = 511 observations). The global positioning system data were categorised into total distance, mean speed (m min−1 ), high-speed running (>14.4 km h−1 ), very high-speed running (>19.9 km h−1 ), and sprint (>23.0 km h−1 ) distance while player load was collected from the accelerometer. The data were log transformed to provide coefficient of variation and the between subject standard deviation (expressed as percentages). Results: Match-to-match variability was increased for higher speed activities (high-speed running, very high-speed running, sprint distance, coefficient of variation %: 13.3–28.6%) compared to global measures (speed, total distance, player load, coefficient of variation %: 5.3–9.2%). The between-match variability was relativity stable for all measures between and within AFL seasons, with only few differences between positions. Higher speed activities (high-speed running, very high-speed running, sprint distance), but excluding mean speed, total distance and player load, were all higher in the final third phase of the season compared to the start of the season. Conclusions: While global measures of physical performance are relatively stable, higher-speed activities and technical measures exhibit a large degree of between-match variability in Australian Football. However, these measures remain relatively stable between positions, and within and between Australian Football League seasons. © 2013 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.
1. Introduction In the Australian Football League (AFL), most teams now routinely collect time motion data during match-play using global positioning system (GPS) devices. In addition, information on a range of technical involvements is also available from commercial statistical providers. Both physical and technical data obtained during competition are commonly used in research to provide an understanding of the factors that affect Australian Football performance. For example, recent research in Australian Football has examined differences in physical and technical performance profiles of individual players depending on match outcome,1 match
∗ Corresponding author. E-mail address: [email protected] (A.J. Coutts).
location and time of day,2 coaches’ perception of performance,3 and fatigue status.4 From a practical perspective, sport scientists and coaches also use the physical activity and technical performance information to gain a greater insight into an individual player’s performance during a match and to assess changes during a season or playing career. However, despite the wide applications of these data, there is presently a poor understanding of the typical variability in these measures. A better understanding of the typical variation in these measures can be useful for designing applied research studies, selecting reliable performance measures for applied studies and to assist interpreting worthwhile changes in performance. Australian Football match-play is dynamic with a complex interplay between individuals on both the same and opposing teams. As such, common measures of physical performance—such as variables in higher speed zones—are not stable and can show large
1440-2440/$ – see front matter © 2013 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jsams.2013.12.006
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
G Model JSAMS-981; No. of Pages 5
ARTICLE IN PRESS T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
2
variation between matches. The sources of variation of physical performance within individual players over successive matches may be related to internal factors (e.g. fitness status, motivation) and external factors (e.g. opposition, tactics, measurement system and environment).5–7 Previous research has reported that measures such as total distance and mean speed (m min−1 ) are relatively stable, while higher-speed activities show higher variability between matches in both soccer7,8 and rugby league.9 Others have reported that the magnitude of the variability between matches may also be influenced by playing position.8 Finally, while Gregson et al.8 found no clear pattern of change in variability of physical performance measures between seasons, others have reported changes in physical performance profiles within professional soccer competition seasons.5,7 While the match-to-match variation in physical activity measures has been reported for soccer7,8 and rugby league,9 there have been no studies which have investigated the variability of physical performance during the AFL competition. It is important to examine the variability of these measures specific to AFL competition due to differences related to field size and player number between Australian Football and other sports—such as soccer and rugby league—where previous investigations have determined between match-variation. Specifically, the greater pitch area available per player may explain the increased higher-speed activities in Australian Football compared to soccer and rugby league.10,11 As high-speed activities display large variation during team sport match-play, the increased amount of high-speed activity in Australian Football may predicate greater variability in physical performance measures compared to previous studies in soccer and rugby league.7–9 Additionally, there have been no studies in any football code that have examined the variability of key individual technical performance measurements. This is somewhat surprising considering that recent research in both soccer12 and Australian Football1 has highlighted the importance of technical performance to overall match outcome. Furthermore, research has shown that certain technical actions can differentiate between higher and lower calibre AFL players.3 Indeed, it is likely that key technical performance metrics vary between matches and this may be related to both individual player characteristics (e.g. experience, technical ability) and fluctuations in game-related factors (e.g. match outcome, strength of opposition). Further research is therefore warranted to investigate the between-match variation of common technical performance metrics in AFL match-play. Accordingly, the aims of this study were to: (1) determine the match-to-match variability for measures of physical and technical performance during successive seasons of Australian Football; (2) examine the influence of playing position on these variables; and (3) assess changes in physical performance within competition seasons.
included for analysis. To assess the influence of the stage of the season on variation, the regular season was divided into three, 8week blocks representing the start (n = 11 games), middle (n = 11 games) and end (n = 9 games). Informed consent and institutional ethics approval were obtained. Physical activity profiles were measured using portable GPS units (Team Sport 2.5, Firmware 6.54, Catapult Innovations, 10 Hz Melbourne, Australia). The unit was fitted within a custom made pouch in the playing jumper of each player prior to the match. Following each match, data were downloaded using proprietary software (Catapult Sprint v5.0.6). All match files were trimmed so that only data obtained during each quarter was retained for further analysis with the proprietary software. The data was categorised into total distance, relative total distance (m min−1 ), high-speed (>14.4 km h−1 ) and very-high speed (>19.9 km h−1 ) running distance, and sprint (>23.0 km h−1 ) distance. Finally, an inbuilt algorithm was used to provide a measure of “Player Load” during each match which has been reported elsewhere.13 The reliability of physical activity measures used in the present study had been previously reported.13–15 The total kicks (number of times a player disposes of the ball by foot), handballs (number of times a player disposes of the ball by hand), and possessions (number of times a player receives the ball) and a player rank score were obtained from a commercial statistics provider (Champion Data© , Victoria, Australia.1 The Champion Data (CD) player rank is an objective measure provided by an algorithm (Champion Data© , Victoria, Australia) which is weighted towards effective ball use and winning disputed possessions and also considers the contribution—both positive and negative—of a range of individual game-specific actions to the overall match outcome. The between match variation in physical and technical performance measures were analysed using a customised Microsoft Excel spreadsheet (Microsoft, Redmond, USA).16 The data were log transformed and the coefficient of variation (%CV) was calculated (i.e. the typical error expressed as a percentage of the mean score).17 The smallest worthwhile change (SWC) in performance for each variable was also obtained by multiplying the between subject standard deviation by 0.2.18 For within-season changes in performance and differences in on field playing time, a magnitude based approach was used to assess the chances of true differences (i.e. greater than the SWC). Quantitative chances of real differences in variables were assessed qualitatively as: <1%, almost certainly not; 1–5%, very unlikely; 5–25%, probably not; 25–75%, possibly; 75–97.5%, likely; 97.5–99%, very likely; >99%, almost certain.18 Standardised effect sizes (ES) were interpreted as <0.2, trivial; 0.2–0-6, small; 0.6–1.2, moderate; 1.2–2.0, large; >2.0, very large.18 Data are presented with 95% confidence intervals (CI).
3. Results 2. Methods Physical and technical performance data were collected from 33 players (age: 23.2 ± 1.8 y; mass: 87.3 ± 7.6 kg; stature: 188 ± 7 cm) from the same club over two successive AFL seasons. The players were classified into three positional groups: backs, midfields or forwards. We excluded ‘ruck players’ from our analysis due to insufficient sample size arising from the highly specialised role that they perform. A total of 511 individual match files were obtained from 31 official regular season competition matches played at outdoor stadiums during the observation period. Due to the unlimited interchange of players during AFL competition match play, it is uncommon for an individual player to complete an entire match. To reduce the effect of playing time on variation, only data from players who were not injured or substituted during the match were
The %CV and SWC of physical and technical performance measurements for positional groups are reported in Table 1. The data show that the variability of physical performance parameters increases as movement classification speed also increases. Lower variability was observed for mean speed compared to total distance for all positions. There was generally high match-to-match variability for all technical performance variables. There were very likely and almost certainly small differences in time on field between midfields and forwards (−3.2; −4.8 to −1.5 min), and midfields and backs (−4.0; −5.4 to −2.5 min), respectively. The variation in physical performance parameters was similar over the 2011 and 2012 competition seasons (Table 2). There was a possibly trivial difference in time on field between season 2011 and season 2012 (1.4; 0.1–2.7 min).
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
ARTICLE IN PRESS
G Model JSAMS-981; No. of Pages 5
T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
3
Table 1 Between match variation of physical and technical measures for positional groups (%CV and 95% CI; %SWC). Backs
Midfields
Forwards
Overall
N = 175
N = 216
N = 120
N = 511
%CV
%SWC
%CV
%SWC
%CV
%SWC
%CV
%SWC
Physical Total distance (m) Speed (m min−1 ) HSR (m) VHSR distance (m) Sprint distance (m) Sprints (n) Body Load (AU)
7.0 (6.1–8.3) 6.4 (5.5–7.5) 13.8 (11.9–16.4) 20.0 (17.3–24.0) 30.4 (26.1–36.6) 24.0 (20.7–28.8) 7.2 (6.3–8.6)
1.9 2.0 4.9 7.3 10.8 8.5 3.3
5.6 (5.0–6.4) 4.5 (4.0–5.2) 11.9 (10.5–13.7) 20.9 (18.5–24.2) 29.3 (26.3–33.3) 23.8 (21.4–27.0) 10.5 (9.4–12.1)
1.4 1.5 3.7 6.4 10.4 8.5 3.8
4.9 (4.2–6.1) 4.8 (4.0–5.9) 11.7 (9.9–14.5) 15.1 (12.8–18.8) 23.9 (20.1–30.1) 25.3 (21.3–31.9) 9.5 (8.1–11.8)
2.5 2.2 6.1 9.9 14.7 12.7 4.6
6.1 (5.7–6.6) 5.3 (4.9–5.7) 13.2 (12.2–14.3) 19.6 (18.1–21.4) 28.6 (26.3–31.3) 24.4 (22.5–26.6) 9.2 (8.5–10.0)
2.0 2.1 5.4 7.5 11.5 9.5 3.9
Technical Possessions (n) Kicks (n) Handballs (n) CD Rank (AU)
32.7 (28.1–39.5) 51.6 (43.9–63.1) 63.3 (53.2–78.6) 37.1 (31.7–44.9)
7.4 11.1 15.4 7.8
23.8 (21.0–27.6) 33.7 (29.6–39.4) 44.2 (38.7–51.9) 26.4 (23.3–30.7)
5.6 8.6 12.3 6.4
31.5 (26.3–39.9) 42.9 (35.6–54.9) 62.4 (50.9–81.1) 31.7 (26.5–40.1)
7.3 10.2 15.2 7.7
28.3 (26.1–31.0) 42.7 (39.2–47.0) 55.3 (50.6–61.2) 31.5 (29.0–34.5)
10.1 11.3 19.4 9.1
HSR: high-speed running, VHSR: very high-speed running, CD Rank: champion data rank.
Global measures of match activity—total distance, mean speed and player load—were stable throughout the season (Table 3). However, there was a possibly small increase in HSR, and a likely small increase in VHSR, sprint distance and sprint number at the end of the season compared to the start. There was a likely small increase in time on field at the end of the season compared to the start (2.8; 1.3–4.4 min). 4. Discussion This study examined the match-to-match variation of physical and technical performance parameters during professional AFL competition. The main findings were that global measures of physical performance were relatively stable; however higher speed activities and technical performance measures displayed greater variability. The variability of physical performance measures was similar over two successive competition years; although there were some within-season changes in higher-speed activities. These findings have important implications for interpreting individual player’s physical and technical performance during matches and examining temporal changes in match performance. Total distance was the most stable physical activity measure in the present investigation. Similar findings have been reported in soccer7,8 and rugby league,9 although the overall variation in the present data was somewhat higher than these investigations. The variability in total distance was reduced when data were expressed relative to time spent on field (m min−1 ). This approach may be useful for reducing variability when comparing match samples of differing duration as is common in Australian Football. In contrast, the higher speed activities were more variable (%CV 13.2–28.6) compared to both total distance and m min−1 , and there was an increase in the variability of these measures with increases in the running speed. This trend has also been reported in previous studies of soccer7,8 and rugby league9 and is likely associated with poorer reliability of measurement devices at higher velocities.14,15,19 Indeed, the variation in HSR in the present
study was similar to previous research in rugby league which also used GPS devices, although both studies have higher CVs than reported previously in soccer,7 which employed semi-automated computer based tracking technology. Differences in both the timemotion measurement systems and in the nature of soccer, rugby league and Australian Football—relating to rules, size of playing field and running demands—may explain these results.20,21 While sprint distance was the least stable of all physical activity measures, sprint number showed less variability and may therefore be a more suitable marker of sprint performance. Nonetheless, sprint number was still quite variable and in contrast to our findings, others have reported similar variability as total sprint distance in soccer.8 A further limitation is that sprint number does not account for the speed or distance of sprint efforts and as such may be less representative of sprint performance. Since good measurement reproducibility is important for assessing small but practically important changes in performance measures in individual players, caution is required when using sprint or other higher speed activities for interpreting changes in an individual players match activity profile. The match-to-match variation in most activity profile measures were similar between positional groups, although there were some small differences, with forwards showing less variability in match performance measures compared to midfields and backs. The specific roles and constraints imposed by playing position may be partially responsible for the differences in variability observed. Indeed, while all Australian Football players are able to roam around the entire field during matches, forwards tend to remain in the offensive half of the field. Furthermore, team tactics influence game structures with many teams adopting ‘running patterns’ whereby forward players perform pre-arranged leading patterns to create scoring opportunities. It may be that reducing the area of the field that players usually work within during matches and running pattern repetition reduces the variability of measurement in higher speed activities. Additionally, the diverse tactical roles of the backs and midfielders may partly explain their greater variability in activity profile during matches.
Table 2 Between match variation in physical parameters during seasons 2011 and 2012 (%CV and 95% CI).
2011 (B = 86; M = 108; F = 64) 2012 (B = 89; M = 108; F = 56)
Total distance (m)
Speed (m min−1 )
HSR distance (m)
VHSR distance (m)
Sprint distance (m)
Sprints (n)
Player load (AU)
5.6 (5.0–6.3)
4.7 (4.2–5.3)
12.7 (11.4–14.4)
20.0 (17.9–22.7)
29.2 (26.1–33.4)
24.9 (22.2–28.3)
9.4 (8.5–10.7)
6.6 (5.9–7.4)
5.9 (5.3–6.6)
13.5 (12.1–15.3)
18.6 (16.6–21.1)
27.0 (24.0–30.9)
23.5 (20.9–26.8)
8.8 (7.9–9.9)
B, backs; M, midfields; F, forwards; HSR, high-speed running, VHSR, very high-speed running.
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
ARTICLE IN PRESS
G Model JSAMS-981; No. of Pages 5
T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
1246 (1215–1276) 19.8 (18.7–21.0) 459 (427–490) 124 (122–127) 13,193 (12988–13398)
B, backs; M, midfields; F, forwards; HSR, high-speed running; VHSR, very high-speed running. a Likely small difference to end of season. b Possible small difference to end of season.
1145 (1089–1200)
1056 (1003–1109)
3492 (3347–3638)
125 (123–127) 13,199 (13001–13397)
3696 (3552–3840)
1266 (1236–1309) 18.2 (17.1–19.3) 431 (401–461)
1263 (1225–1301) 17.9 (16.9–18.9) 405 (379–431)
b
1023 (977–1070)
b
123 (121–125)
Start of season (B = 62; M = 78; F = 43) Middle of season (B = 59; M = 76; F = 42) End of season (B = 54; M = 62; F = 35)
12,948 (12775–13120)
3462 (3336–3589)
Sprint distance (m)
a a
VHSR distance (m) HSR distance (m) Speed (m min−1 ) Total distance (m)
Table 3 Seasonal variations in physical performance parameters (mean and 95% CI).
b
Sprints (n)
a
Player Load (AU)
4
The variability of match activity measures was relatively constant over the two successive seasons, with some variables increasing slightly, while others were reduced. This finding supports previous research which has failed to find a consistent pattern of change in higher-speed activities over three seasons of professional soccer competition.8 While the variability in physical performance measures did not change between seasons, we observed small changes in the profiles of higher-speed, but not global, parameters during the course of a season. Specifically, HSR, VHSR, sprint distance and sprint number were all increased at the end of the season compared with the start. These results are in agreement with previous research in soccer,5,7 and may be related to changes in team tactics, improved fitness or specific adaptations to match-play during the season.7 Indeed, previous research has shown a case study of increased physical activity profile in a team during finals matches compared to home and away games.22 It is also possible that a greater importance is placed on matches approaching finals, which may contribute to the increased activity profile towards the end of the season. Regardless of the cause, it is important to consider this trend when interpreting temporal changes in performance during an AFL season. The technical performance measures examined in this study displayed high levels of overall variability (%CV 28.3–55.3). This is a novel aspect of the study and has important implications given that recent applied studies of AFL competition have employed these variables as performance indicators.1–3 For example, Sullivan et al.1 reported that several technical performance measures including total kicks and CD rank were all higher during quarters won compared to quarters lost. Similarly, others have shown that the frequency of some technical involvements differentiates between higher and lower-calibre players.3 These studies are in agreement with previous soccer research which has shown that certain technical actions differ between players on more and less successful teams.12 Collectively, these studies demonstrate the importance of technical involvements for overall match performance during match-play. While it appears that technical variables such as CD rank and kicks are sensitive performance measures, caution is required when using these parameters to interpret a player’s performance as they display higher match-to-match variability. The very high level of variability of technical performance of individual players may be due to internal (experience, technical ability and decision making) and external (strength of opposition, team tactics and the “flow of play”) factors. The precision of a performance indicator—either physical or technical—is an important consideration for applied research investigations.8 Indeed, HSR has been suggested as a significant physical performance parameter in soccer due to its association with “the play” during a match, ability to discriminate between playing levels and apparent sensitivity to match related fatigue.5 Similarly, recent research has shown that CD rank is a suitable measure of Australian Football performance due to its associations with match outcomes.12 While it is essential that performance variables for applied research projects are selected based on their relationships with important match characteristics, it is also critical that the measurement error of these variables is lower than the match-to-match variation. Further research is required to assess the reliability of technical measures used in the present investigations to determine their appropriateness as indicators of performance.8,18 In these cases, a large sample size may be required to overcome these limitations for detecting performance changes.23 For example, using the framework provided by Batterham and Atkinson, a variable with a CV of 30% and SWC 10%—such as overall possessions or overall CD rank—would require approximately 180 samples in order to detect a statistically significant change in performance.23
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006
G Model JSAMS-981; No. of Pages 5
ARTICLE IN PRESS T. Kempton et al. / Journal of Science and Medicine in Sport xxx (2014) xxx–xxx
While this study examined a large (∼500 match samples) data set obtained over two seasons, this is still smaller than previous soccer research.8 Indeed, the comparatively smaller sample size in this investigation is a limitation and future studies are required with more comprehensive data sets to provide estimates of the true variability of these measures in AFL match-play. In addition, the use of only absolute speed thresholds for determining higher-speed distances may be a limitation in the present study as some have suggested that individualised speed thresholds may complement the traditional arbitrary approach.24,25 Finally, there are many factors which may contribute to the variability of match performance, for example, opposition, match outcome, match location, time of season, physical capacity of players and environmental conditions.7,8 We have not accounted for these factors directly in the present study and future research should examine the independent effects of these influences. 5. Conclusion This study has examined the typical variability of common physical and technical performance parameters for positional groups, within a season and between two seasons of the AFL competition. The results demonstrate that match-to-match variation of the physical activity measures increased with higher speed activities. The match-to-match variability for all technical performance variables was higher than the physical activity measures. However, the variability in total distance, mean speed and player load remained relatively stable within and between seasons. In contrast, HSR, VHSR, sprint distance and sprint number were higher at the end of the season compared to the start. These findings have important implications for interpreting changes in the match activity profile and technical performance of both individual players and the team and for examining temporal changes in match performance. Practical implications • Due to the potential for high levels of variability for many common measures used in Australian Football match analysis studies, it is recommended that researchers establish and report CV’s for their data set. • The low match-to-match variation in total distance in Australian Football demonstrates that this is a relatively stable performance measure; however, the variability for higher speed activities and technical measures is much greater. • The increase in activity profile in Australian Football towards the end of the season should be taken into account when planning individual and team training loads. • Statistical approaches which use the CV may assist in designing applied research studies and interpreting the magnitude of change. Acknowledgements We would like to thank Ermanno Rampinini for his feedback on the manuscript. No external sources of funding were provided
5
for this study. The authors have no relevant conflicts of interest to declare. References 1. Sullivan C, Bilsborough J, Hocking J et al. Match score affects activity profile and skill performance in professional Australian Football players. J Sci Med Sport 2013. http://dx.doi.org/10.1016/j.jsams.2013.05.001. 2. Hiscock D, Dawson B, Heasman J et al. Game movements and player performance in the Australian Football League. Int J Perfe Analysis Sport 2012; 12(3):531–545. 3. Johnston R, Watsford M, Pine M et al. Movement demands and match performance in professional australian football. Int J Sports Med 2012; 33(2):89. 4. Mooney M, Cormack S, O’Brien B et al. Impact of neuromuscular fatigue on match exercise intensity and performance in elite Australian football. J Strength Cond Res 2013; 27(1):166–173. 5. Mohr M, Krustup P, Bangsbo J. Match performance of high-standard soccer players with special reference to development of fatigue. J Sports Sci 2003; 21:519–528. 6. Mohr M, Mujika I, Santisteban J et al. Examination of fatigue development in elite soccer in a hot environment: a multi-experimental approach. Scand J Med Sci Sports 2010; 20(3):125–132. 7. Rampinini E, Coutts A, Castagna C et al. Variation in top level soccer match performance. Int J Sports Med 2007; 28(12):1018–1024. 8. Gregson W, Drust B, Atkinson G et al. Match-to-match variability of highspeed activities in Premier League soccer. Int J Sports Med 2010; 31: 237–242. 9. Kempton T, Sirotic A, Coutts A. Between match variation in professional rugby league competition. J Sci Med Sport 2013. http://dx.doi.org/10.1016/j.jsams.2013.05.006. 10. Kennett D, Kempton T, Coutts A. Factors affecting exercise intensity in rugby-specific small-sided games. J Strength Cond Res 2012; 26(8): 2037–2042. 11. Varley M, Gabbett T, Aughey R. Activity profiles of professional soccer, rugby league and Australian football match play. J Sports Sci 2013: 1–9. 12. Rampinini E, Impellizzeri F, Castagna C et al. Technical performance during soccer matches of the Italian Serie A league: effect of fatigue and competitive level. J Sci Med Sport 2009; 12(1):227–233. 13. Boyd LJ, Ball K, Aughey RJ. The reliability of MinimaxX accelerometers for measuring physical activity in Australian football. Int J Sports Physiol Perf 2011; 6(3):311–321. 14. Varley MC, Fairweather IH, Aughey RJ. Validity and reliability of GPS for measuring instantaneous velocity during acceleration, deceleration, and constant motion. J Sports Sci 2012; 30(2):121–127. 15. Jennings D, Cormack S, Coutts AJ et al. The validity and reliability of GPS units for measuring distance in team sport specific running patterns. Int J Sports Physiol Perform 2010; 5(3):328–341. 16. Hopkins WG. Precision of measurement, In: A new view of statistics., 2011, newstats.org/precision.html. 17. Hopkins WG. Measures of reliability in sports medicine and science. Sports Med 2000; 30(1):1–15. 18. Hopkins W, Marshall S, Batterham A et al. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc 2009; 41(1):3–12. 19. Coutts AJ, Duffield R. Validity and reliability of GPS units for measuring movement demands of team sports. J Sci Med Sport 2010; 13(1):133–135. 20. Harley JA, Lovell RJ, Barnes CA et al. The interchangeability of global positioning system and semiautomated video-based performance data during elite soccer match play. J Strength Cond Res 2011; 25(8):2334–2336. 21. Randers MB, Mujika I, Hewitt A et al. Application of four different football match analysis systems: a comparative study. J Sports Sci 2010; 28(2): 171–182. 22. Aughey RJ. Increased high-intensity activity in elite Australian football finals matches. Int J Sports Physiol Perf 2011; 6(3):367–379. 23. Batterham A, Atkinson G. How big does my sample need to be? A primer on the murky world of sample size estimation. Phys Ther Sport 2005; 6:153–163. 24. Abt G, Lovell R. The use of individualised speed and intensity thresholds for determining the distance run at high-intensity in professional soccer. J Sports Sci 2009; 29(9):893–898. 25. Lovell R, Abt G. Individualization of time-motion analysis: a case–cohort example. Int J Sports Physiol Perform 2013; 8(4):456–458.
Please cite this article in press as: Kempton T, et al. Match-to-match variation in physical activity and technical skill measures in professional Australian Football. J Sci Med Sport (2014), http://dx.doi.org/10.1016/j.jsams.2013.12.006