Journal of Science and Medicine in Sport 17 (2014) 223–228

Contents lists available at ScienceDirect

Journal of Science and Medicine in Sport

journal homepage: www.elsevier.com/locate/jsams

Original research

Evolution of World Cup soccer final games 1966–2010:

Game structure, speed and play patterns
Jarryd Luke Wallace, Kevin Ian Norton ∗
School of Health Sciences, University of South Australia, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Objectives: There are relatively few performance analysis studies on field sports investigating how they
Received 21 June 2012 evolve from a structural or tactical viewpoint. Field sports like soccer involve complex, non-linear dynam-
Received in revised form 16 January 2013 ical systems yet consistent patterns of play are recognisable over time and among different sports. This
Accepted 28 March 2013
study on soccer trends helps build a framework of potential causative mechanisms for these patterns.
Available online 2 May 2013
Design: Retrospective correlational study.
Methods: Broadcast footage of World Cup finals between 1966 and 2010 was used to assess patterns of
Keywords:
play and stop periods, type and duration of game stoppages, ball speed, player density (congestion) and
Play speed
Game speed
passing rates. This involved computer-based ball tracking and other notational analyses. These results
Player density were analysed using linear regression to track changes across time.
Sports evolution Results: Almost every variable assessed changed significantly over time. Play duration decreased while
Association football stoppage duration increased, both affecting the work: recovery ratios. Ball (game) speed increased by
15% over the 44-year period. Play structure changed towards a higher player density with a 35% greater
passing rate.
Conclusions: Increases in soccer ball speed and player density show similarities with other field sports
and suggest common evolutionary pressures may be driving play structures. The increased intensity of
play is paralleled by longer stoppage breaks which allow greater player recovery and subsequently more
intense play. Defensive strategies dominate over time as demonstrated by increased player density and
congestion. The long-term pattern formations demonstrate successful coordinated states within team
structures are predictable and may have universal causative mechanisms.
© 2013 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.

1. Introduction and decision-making as play moves between attacking and defend-

ing actions. These physical and motor skill attributes occur within
Association Football (soccer) is the most popular sport in the a tactical framework encompassing set plays, playing styles and
world, played in almost every country and by an estimated 265 team strategies, and that may need to be modified depending on
million participants. On the international stage, the best players the opposition, environmental conditions and score line.
represent their nations every four years at the FIFA World Cup, the There is a great deal of scientific enquiry into how these var-
largest single-event sporting competition (http://www.fifa.com). ious elements of the game interact and how performance can
Qualification for the 2010 World Cup tournament held in South be enhanced using tools such as notational, motion and per-
Africa, for example, involved 204 countries competing for a position formance analyses to link to game outcomes.2 As technologies
in the final 32 teams. have improved these methods have moved from being essentially
Soccer, like all field sports, is composed of a series of play periods descriptive, for example, quantifying player movement distances
that are randomly interspersed with stop periods, typically where and speed breakdowns within games, to helping coaches recall
the referee has called an infringement or the ball is out of the play- important information from games, and in optimising player con-
ing field. Soccer is non-continuous and the energy demands involve ditioning through specificity of training drills.2,3 More explanatory
a variety of intense anaerobic movements interspersed with more applications have described relationships between team pattern
low-intensity aerobic activities.1 Elite players require a large range formations and their link to match results,4 and with underlying
of motor skills, as well as the need for rapid information processing theoretical explanations of performance.4
Glazier5 argues that performance analysis is a science sub-
discipline working from the intra-individual to the inter-individual
∗ Corresponding author. level using biomechanics and notational analysis to examine emer-
E-mail address: [email protected] (K.I. Norton). gent pattern formations and their relationships with performance

1440-2440/$ – see front matter © 2013 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jsams.2013.03.016
224 J.L. Wallace, K.I. Norton / Journal of Science and Medicine in Sport 17 (2014) 223–228

outcomes. Performance analysis research typically operates over a comparing the averages of the repeat measures to the averages of
relatively small time frame when assessing interpersonal and spa- duplicate games of a second tracker.
tial interactions on the playing field, and their association with TrakPerformance software was used to measure the speed of the
success and failure. This may vary from a few seconds, such as ball, which is our measure of game speed. The method involved a
the player velocity and positioning in discrete plays to passing computer-based tracking system where the tracker visually and
sequences, and even entire games such as assessing variability mechanically tracked the movement of the ball using a scaled,
in high speed movements from match to match.4,6 We extend miniaturised playing field and tracking pen. As the ball moved
this idea to a broader level of field sport research. This incorpo- around the field the tracker mirrored this movement across a track-
rates game and seasonal comparisons of patterns of play that often ing tablet or computer screen. The movement was calibrated to
change over a longer period of time and add insight into how and represent the true displacement of the ball over the field and was
why field sports evolve.7–9 used to calculate ball speed (m/s). Mean ball speed was calculated
Virtually all sports evolve across time because of changes to using the speed of all the individual play periods within a half or
rules and game tactics, increases in professionalism, use of new entire game.
technologies, global exposure, and transformations in training Previous studies have shown the reliability of the tracking sys-
and selection pressures affecting player physiology, morphology tem to be within 5–7% of the true distances, and not different from
and motor skills.10 Understanding these evolutionary patterns GPS measures.22 However, it is also known that experience in track-
has, in other sports, been used for a range of applications. These ing plays an important role in the accuracy of the measures.12,13
include estimating future game and training demands,11 assist- Therefore, prior to collecting data for the study, duplicate measures
ing in player selection and talent identification,7 understanding of two games were completed (JW) to assess intra-tester reliability.
injury aetiology,9 and predicting the impact of rule changes and The ball speed values were also compared to duplicate measures of
general drivers of change on the transition of field sports into the another tracker (KN) to quantify inter-tester error.
future.8,10,12 Player density is defined as the congestion of players around the
The aim of this study was to quantify elements of game struc- ball. Creating high density is often a strategy used to place addi-
ture and speed, including play and stop periods, ball speed, player tional defensive pressure on the ball-carrier while low density is
density and passing rates in elite-level soccer played in World Cup preferable when attacking to allow space and time for decision
finals over the period 1966–2010. These patterns of change are used making and ball possession, and when shooting for goal. Player
to build a framework of potential drivers of evolutionary pressures density is operationalised as a measure of the number of players
operating in soccer over the longer term. within an estimated 5 m of the ball. Player density is a fluid and
constantly changing variable within a game that is difficult to mea-
sure continuously with present technologies. Therefore, a count of
2. Methods the number of players within a 5 m radius of the ball was made
at each 15 s increment across the game. If the ball was out of the
The study was a retrospective correlational design to quantify playing field, was not ‘in play’, or if there were no players within
longitudinal changes in the final game of each of the 12 World Cup 5 m of the ball nothing was recorded for that interval.
soccer tournaments between 1966 and 2010. Commercially avail- The 5 m radius required a level of subjectivity and estimation.
able broadcast quality video footage of World Cup finals between To assist, several techniques were used to increase the accuracy
1966 and 2010 was used. Each game was assessed for quantifi- of this judgement. Player heights are known to average approxi-
able variables of game structure and play patterns, with the same mately 1.8 m so players within the vicinity of the ball could be used
observer measuring all games to ensure consistency of measure- as a guide to distance. Also, known distances on the field, such as
ment skill and interpretation of variables recorded. Games last for the central circle or boxes at each end, were also used as these have
approximately 90 min, made up of two 45 min halves and addi- specified dimensions. The consistent patterns of the grass prepara-
tional injury time at the end of each half. Extra time is played in a tion were another method used to judge the 5 m radius. Measures
World Cup final if the scores are tied. Extra time (if applicable) was of intra-tester reliability were made that involved taking five repeat
not included in this analysis. measures of one half, with at least one day between trials.
Play and stop period durations were timed using TrakPerfor- Passing rate is an indicator of game style and skill demands. A
mance software (SportsTec Australia; http://www.sportstec.com). pass was defined as the deliberate act of striking the ball to main-
A play period began at the start of each half and continued until a tain team possession, even without a defined target. This included
stop event occurred such as a free kick or the ball exited the playing attempted passes that were blocked or intercepted. Strikes that
field. Play periods re-started when the ball returned to the field via were intended to clear the ball out of play and shots at goal were not
a kick or throw. The frequency and duration of play and stop periods included. Kicks from set pieces such as corners and free kicks were
were recorded for each game. The totals were also used to assess also not included. Passes were counted manually for each half and
the time the ball spent in and out of play, as well as a percentage of the total was divided by the total play time measured by TrakPerfor-
play time in the final game time. mance to give a rate of passes per minute of play time. Passing rate
Stop periods were further categorised to provide the dura- error assessment involved duplicate measures across a game. The
tion and frequency of different match events. The stop events or mean percent difference between the two samples was calculated.
‘stoppages’ were timed manually on a separate occasion using a The commercial broadcast footage uses replays of specific
stopwatch to the nearest 0.1 s and were coded as throw-ins, corner aspects of the game to enhance viewing experience. The replays
kicks, goal kicks, injuries, substitutions, goals and free kicks. Free occasionally prevented precise measurement so an exclusion crite-
kicks were separated into two groups: a direct shot at goal where ria was used: ball speed and play duration were recorded from the
there was no pass to a teammate, although a touch to set the ball position of the ball when normal viewing returned; player den-
rolling for a team mate was acceptable as part of this definition, and sity intervals were recorded as missing for this interval in line
all other free kicks. An event was recorded as an injury only if a med- with when the ball was out of play; stop period event times were
ical officer or trainer was required, or if the ball was deliberately recorded as missing if either the start or stop times were immea-
put out of play with the sole intention of tending to a player. Play surable.
and stop period intra-tester reliability was assessed using repeat The study is based on only 12 games over a 44-year period
measures of two games. Inter-tester reliability was calculated by and the World Cup finals may not be representative of soccer in
 J.L. Wallace, K.I. Norton / Journal of Science and Medicine in Sport 17 (2014) 223–228 225

Table 1
Frequency of stop events in each World Cup final game (1966–2010). Parentheses indicate the number of events that could not be timed accurately due to technical (video)
limitations. The total column includes the number of timed and untimed events.

Year Corners Free kicks Free kick at goal Goal kick Throw in Goal Substitution Injury Total

1966 13 35 (1) 2 30 44 (1) 4 0 0 130

1970 8 28 (1) 10 29 27 4 (1) 2 2 112
1974 19 46 4 16 (1) 26 3 1 1 117
1978 9 56 6 9 (1) 32 (1) 2 4 2 122
1982 7 40 (3) 3 26 (1) 40 0 (4) 4 1 129
1986 13 59 5 12 (2) 36 5 2 1 135
1990 4 40 (3) 4 15 27 (1) 1 1 6 102
1994 5 30 (1) 3 15 (1) 32 (2) 0 2 1 92
1998 12 29 (1) 3 12 (4) 40 (1) 2 (1) 4 7 116
2002 16 27 (8) 3 17 (1) 39 (2) 1 (1) 5 4 124
2006 8 30 (6) 3 16 (5) 37 (6) 0 (2) 3 5 121
2010 10 32 (6) 4 13 (4) 48 (2) 0 3 5 127

general. It is also clear that games can vary widely depending on and 2010 and did not change over time (p = 0.8071). There
environmental conditions, team strategies, refereeing style and were no changes in the number of specific stop events across
score-line.14,15 Even within games variables such as game speed time except for an increase in the number of injury stoppages
and player density can differ considerably. (y = −228.2 + 0.11626 × x; r = 0.71; p = 0.009).
Games were analysed as halves and whole games to determine Across the period 1966 to 2010 there were significant
trends within games and across time. The level of significance was increases in the average time taken for throw-ins (approximately
set at p < 0.05 unless otherwise stated. Variables were analysed 8–12 s; y = −149.15 + 0.080 × x; r = 0.21; p = 0.0001), corner kicks
with paired and independent t-tests, non-parametric tests for the (19–26 s; y = −293.52 + 0.159 × x; r = 0.29; p = 0.0012), and goal
skewed player density data and linear regression analysis to assess kicks (16–24 s; y = −353.45 + 0.187 × x; r = 0.35; p = 0.0001). Regres-
changes across time. sion analyses showed there were no changes in the average time
taken to make substitutions (mean time 46 s; p = 0.663) or for injury
3. Results stoppages (75 s; p = 0.520).
Fig. 2 illustrates a significant increase in the time taken for free
Intra-tester reliability trials showed no significant difference kicks from an average of approximately 13 s in 1966 to 20 s in
between total play time (mean error = 1.8%; p = 0.647) or total stop 2010. It also shows that free kicks for goal increased over time
time (0.5%; p = 0.981). There was a 2.9% difference between the from an average of about 38 s in 1966 to 63 s in 2010. This positive
number of recorded play periods in the repeat trials. The second relationship was unchanged when penalties (n = 5) were excluded
tracker recorded intra-tester reliability of 1.8% for total play time (p = 0.0043).
(p = 0.650) and 2.2% for total stop time (p = 0.355) across repeat Intra-tester reliability showed no significant difference in track-
measures. Inter-tester comparisons showed there were no signif- ing distance (mean difference 2.6%; p = 0.636) or total play time
icant differences between the two trackers for play time (3.2%; (1.8%; p = 0.647). There was no difference in the individual play
p = 0.638) or stop time (6.6%; p = 0.591). period speed (3.4%; p = 0.744). Intra-tester reliability of tracking dis-
The total number of game events between duplicate measures tance and play time between trackers showed no differences (3.8%;
showed no difference (0%), while the summation of stop time of the p = 0.754 and 1.8%; p = 0.650, respectively). There was no difference
game events showed an average difference of 6.4%. There were no in individual play period speed (8.7%; p = 0.240).
significant differences in times recorded for throw-ins (mean dif- The inter-tester reliability showed no differences for tracking
ference = 0.1%; p = 0.877), corner kicks (1.0%; p = 0.927), goal kicks distance (3.2%; p = 0.638), total play time (0.4%; p = 0.326) and indi-
(0.4%, p = 0.950), or free kicks (0.2%; p = 0.957) across the duplicate vidual play period speeds (4.1%; p = 0.161). The difference in mean
measures. speed averaged 6.3%. The results both within and across testers
The percentage of game time where the ball was in play showed acceptable reliability using the computer-based tracking
averaged 5.1% less in the second half compared to the first half technology and are in line with previous findings.13
(mean = 61.4% v 56.3%; t = 3.08; p = 0.011). This was due to both a Fig. 1 illustrates game speed increased significantly across the
decrease in play time in the second half (mean = 1653 s v 1522 s; study period. The regression line shows game speed increased from
t = 2.81; p = 0.071), and a similar increase in stop time (1040 s v an average of approximately 8.0 m/s to 9.2 m/s, a 15% elevation
1192 s; t = 1.16; p = 0.009). Regression analysis showed total play across the period 1966–2010. Paired t-tests showed no signifi-
time remained relatively constant (r = 0.306; p = 0.145), while total cant difference between halves for average game speed (t = 0.88,
stop time increased (r = 0.502; p = 0.012) over the period 1966- p = 0.398).
2010. These changes resulted in a decreased percentage of play There were no significant differences in player density meas-
time from 70.1% in the earliest game to 51.8% in the most recent ures across the repeat reliability trials (p = 0.7128). Player density
World Cup (predicted average of 64.2% down to 53.6%; Fig. 1). measures are markedly skewed. Spearman’s rank correlation anal-
Regression analysis of all play periods (n = 1635) across the ysis of all player density data points within each game (n = 2184)
period 1966 and 2010 showed a decrease in the average play over time showed an increasing trend (p < 0.0001).
duration equal to about 4 s from 29 to 25 s (y = 206.5 − 0.09 × x; Reliability using repeated measures showed an error of 3.0%
p = 0.037). Analysis of all stoppages (n = 1670) showed an increase difference in the number of passes recorded across the game mea-
in the average stop period duration by about 7 s to around 17 s in sured. In the World Cup finals analysis there was a significant
the most recent game (y = −287.8 + 0.154 × x; p < 0.0001). increase in passing rate from 1966 to 2010. The regression line
Table 1 shows the type and frequency of specific stop events. shows an increase from about 11.3 to 15.3 passes/min playing time
The number of stops per game averaged 119 ± 12 between 1966 shown in Fig. 1.
226 J.L. Wallace, K.I. Norton / Journal of Science and Medicine in Sport 17 (2014) 223–228

Proportion of game time in play Duration of stop events

140
80
120
75
100

Free kick (s)

Play time (%)

70
80
65
60
60
40
55
20
50
0
45

40
140
Ball speed 120

Free kick for goal (s)

100
11
Play period speed (m/s)

10.5 80

10 60

9.5 40
9 20
8.5
0
8 1970 1978 1986 1994 2002 2010
1966 1974 1982 1990 1998 2006
7.5

7 Year
Fig. 2. Duration of free kick stop events in World Cup finals from 1966 to 2010.
Passing rate Top panel shows free kicks not for goal (y = −277.50 + 0.148 × x; r = 0.18; p = 0.0001);
bottom panel shows free kicks for goal (y = −1088.4 + 0.573 × x; r = 0.36; p = 0.0104).
16.5
16
4. Discussion
15.5
Passes per min

15 Ball tracking and notational analysis were used to assess the evo-
14.5 lution of World Cup soccer finals played in the period 1966–2010.
14 Specifically, the study quantified ball speed, the duration and fre-
13.5 quency of play and stop periods, player density and passing rates.
13 The study has demonstrated elite level soccer has undergone sub-
12.5 stantial changes in game speed, structure and play patterns. These
changes are not random but rather illustrate models of game speed
12
and patterns that are likely to confer an advantage in terms of
11.5
success at the highest level.
11 Field sports have game structures involving patterns of attack-
1970 1978 1986 1994 2002 2010 ing and defending play periods interspersed with stoppages. The
1966 1974 1982 1990 1998 2006 results show the total number of stoppages was unchanged over
time but there were consistent increases in the average duration
Year of almost all stop events. Therefore, the increase in total stop time
found was due to increased duration of stop events rather than
Fig. 1. Play structure patterns in World Cup finals from 1966 to 2010. Top frequency of stoppages. The way in which these play and stop
panel shows regression of play time as a percentage of overall game time
periods ‘unfold’ during games and interact is important for a num-
(y = 539.8 − 0.2419 × x; r = 0.44; p = 0.032). Centre panel is game speed using ball
tracking (y = −42.958 + 0.026 × x; r = 0.446; p = 0.029) and bottom panel shows pass- ber of reasons. They directly impact potential work-to-recovery
ing rate per minute (y = −166.194 + 0.090 × x; r = 0.796; p < 0.0001). Darker circles ratios, influence opportunities to rest, and determine the intensity
are first half measures. with which players can compete during subsequent play periods.16
These fatigue-related interactions have changed over the 44-year
period such that an average work: recovery ratio (play period dura-
tion:stop period duration) has decreased from around 4:1 in 1966
to 1.5:1 in 2010 and is also reflected in the lower play time relative
to the total game time (Fig. 1). Game structure impacts physical
demands on the players, for example, the trend towards shorter,
 J.L. Wallace, K.I. Norton / Journal of Science and Medicine in Sport 17 (2014) 223–228 227

higher intensity play periods. The current pattern of shorter play The results showed player density has increased over time.
periods with longer recovery periods facilitates higher game inten- Player density is an indicator of congestion and this plays an impor-
sity when play resumes.17 As a consequence, players impact game tant part in successful and unsuccessful patterns of team play. For
structure through their control over stop duration where they will example, the distance between attackers and defenders is a crit-
often extend this time to aid in recovery and set up structures. As ical feature of goal success where attacking players try to open
a spectacle this is often exciting and adds to the drama but there space or create space between them and defenders while at the
are obvious drawbacks. These include an increased probability of same time successful defenders try to close space to constrain
injuries through high-speed collisions involving greater kinetic scoring angles and force rushed skill execution.4 Creating space
energy9 and when moving at sprint speeds more frequently.18 requires speed, early movement and anticipatory skill, and it also
Interestingly the only game stoppage event to increase in frequency involves decreasing player density when in possession of the ball.
over time was for injuries although these stop events can also be Decreased player density allows players more time and space and
used strategically. the probability of scoring in soccer doubles if there is more than
Stoppages have become critical elements of soccer games where one metre of free space around the kicker when shooting for
set plays and tactics enhance goals scoring opportunities.19 This is goal.24 Defenders, on the other hand, try to increase player den-
reflected by ‘set piece specialists’, or players who have skills and sity around the attacker and this involves matching attackers for
abilities in particular events such as kicking with a significant curve speed and early anticipatory movements,4 so it is not surprising
or taking long throw-ins. Fig. 2 shows the trends in stop duration for that attackers and defenders show the highest levels of sprinting
direct and indirect free kicks. While they are both increasing over in games.17
time, the rate of change as indicated by the slopes of the regres- Raising player density leads to elevated levels of skill, speed
sion lines, is almost 4-fold higher for the set shots, reinforcing the and precision required to move through the player traffic, increases
importance of these relatively rare game events in goal-scoring.19 the number of variables included in decision-making, while at the
Using the ball speed in a field sport such as soccer to quantify same time forces players to hurry selections.24 Greater congestion
game speed has only rarely been reported.10,12,20 This is despite obliges players to accelerate, decelerate and change direction more
the fact ball speed has a fundamental bearing on the speed of often to avoid other players, and to find space and break away from
player movement and decision-making, for example to intercept or contests. Not only is this physically demanding but may lead to
reach a pass, determines a skill set that the game demands, and is more contests and associated collisions,14,15 and may be a factor
routinely measured and reported in many other sports such as rac- behind increased injury rates in soccer.25–27 Higher player density
quet (tennis and badminton), and throwing sports (cricket, softball and associated man-to-man marking induces higher blood lactate
and baseball). This may be because there is no general consen- levels and leads to greater fatigue compared to a zoning defensive
sus of what game speed is, how it should be measured, and until strategy.28
recently, technical difficulties in measuring it.2 Notwithstanding, Player density, therefore, is a balance between attacking forces
in this study the game speed increased by 15% across the period to create space and defensive actions to close space, adjusted to
reviewed. In a similar way, Australian football (AF) has also had levels of fitness and fatigue within the game. The long-term trend
increases in game speed, doubling over a 50 yr period to the late showing increasing player density at the highest level suggests
1990s.12 an evolution towards greater defensive strategies matched with
The question becomes what drives this evolutionary pattern? superior skill and fitness. This is supported more broadly, although
That is, what is the advantage of playing faster that might lead to anecdotally, whereby contemporary play styles promote ‘compact-
this change over time? There is evidence to suggest a performance ness’ where players are concentrated in a relatively narrow band in
advantage in playing with speed in invasion-style field sports, par- the central regions of the field, particularly when defending, com-
ticularly at times when moving closer to goal and during phases pared to games of yesteryear. Both elite-level AF and World Cup
of transition from defending to attacking.3 Using a metric called Rugby Union have also demonstrated increases in player density in
a centroid, or vector-averaged position of players within a team, recent decades suggesting the success of these common defensive
the attacking teams were at an advantage and more likely to score threads among field sports.29
when their centroid moved faster in a transition period relative Passing rate increased linearly by around 35% across the time
to the defender’s centroid. Vilar et al.4 used an ecological dynam- period studied. This measure illustrates the increased intent or
ics model to help explain playing patterns and interactions among need to move the ball more often, rather than holding the ball up
soccer players. They found attackers were constantly attempting or playing long passes, a style more characteristic of the earlier
to break symmetry with their nearest opponent while defenders World Cup games. Fast ball movement or game speed is gener-
tried to maintain symmetry by staying between the attacker and ated through a combination of high passing rates and high ball
the goal. In situations where goals were scored an advantage was speed (in addition to faster player movement when in possession
achieved when attackers moved faster, both before and after receiv- of the ball), and has been shown to be critical for success,3 and to
ing the ball, than their immediate defender. Successful defenders differentiate elite from non-elite players in tests of soccer skill.30
also achieved higher velocities relative to the attackers to intercept Successful teams at the World Cup level have also been found
the ball trajectory.13 Others have found the amount of high speed to have significantly greater passes per game than less successful
running intensity activity is related to the overall success of the teams. Moreover, since previous work suggests player proximity is
team.21,22 This is consistent with reports showing the need for high a key factor in skill execution,24 the higher passing rate and player
intensity running and repeat sprint ability in the modern game and congestion in modern games, mean today’s elite players require
how these have increased compared to games in previous decades faster decision-making, and skill. This is undoubtedly selected for
and as the level of competition increases.17,23 Finally, ball speed in through the developmental years and then rehearsed as full-time
AF was shown to be positively correlated with playing level, larger athletes. These attributes will be even more critical in the future
grounds and dry conditions.9 if game speed and passing rates continue along the evolutionary
Together, these results reinforce the advantage in faster game trajectories found.
speed. The high-intensity movements impact the game demands Comparing distances covered across soccer studies shows it
and speed attributes of elite players and create a relentless has become more physically demanding over time,2 although it
search for faster players and playing styles, and concomitant skills has been difficult to quantify this in consistent and comparable
required for these demands. ways. The changes in World Cup soccer finals shown in the current
228 J.L. Wallace, K.I. Norton / Journal of Science and Medicine in Sport 17 (2014) 223–228

study all indicate the physical demands are increasing. Defensive 3. Frencken W, Lemmink K, Delleman N et al. Oscillations in centroid position
actions in a dense player environment where the emphasis is placed and surface area of soccer teams in small-sided games. Eur J Sports Sci 2011;
11(4):213–223.
on number superiority, the need for speed in the transitions, the 4. Vilar L, Araujo D, Davids K et al. The role of ecological dynamics in analyzing
advantage of rapid attacker ‘spread’ when there is a defensive-to- performance in team sports. Sports Med 2012; 42(1):1–10.
offensive turnover, all demand superior fitness. In response, and 5. Glazier P. Game, set and match? Substantive issues and future directions in
performance analysis. Sports Med 2010; 40(8):625–634.
in line with the human power curve, players will extend recovery 6. Gregson W, Drust B, Atkinson G et al. Match-to-match variability of high-speed
periods when they can to help mitigate fatigue. Although specu- activities in Premier League Soccer. Int J Sports Med 2010; 31:327–342.
lative, if left unchecked the game will likely become even more 7. Burgess D, Naughton G, Norton K. Quantifying the gap between under 18 and
senior AFL football: 2003–2009. J Sports Physiol Perform 2011; 7(1):53–58.
intense characterised by shorter, intense play periods. In turn, this 8. Norton K. Modelling the effect of altering the interchange patterns in the AFL. AFL
game style will suit bigger, more powerful athletes who will have Research Report. Melbourne: AFL House; 2010.
time to recover in the longer stoppages. 9. Norton K, Schwerdt S, Lange K. Evidence for the aetiology of injuries in Australian
football. Br J Sports Med 2001; 35(6):418–423.
10. Norton K, Olds T. Morphological evolution of athletes over the 20th century:
5. Conclusions causes and consequences. Sports Med 2001; 31:763–783.
11. Wisbey B, Montgomery P, Pyne D et al. Quantifying movement demands of AFL
football using GPS tracking. J Sci Med Sports 2010; 13:531–536.
This study on performance analysis has described emergent
12. Norton K, Craig N, Olds T. The evolution of Australian football. J Sci Med Sport
team and game structure patterns over the longer term. Soccer at 1999; 2(4):389–404.
the highest level has evolved significantly and will likely continue 13. Edgecomb S, Norton K. Comparison of global positioning and computer-based
to change. While the causative factors are not known, the func- tracking systems for measuring player movement distance during Australian
football. J Sci Med Sport 2006; 9(1):25–32.
tional interaction of players and environments form predictable 14. Bradley P, Carling C, Archer D et al. The effect of playing formation on high-
and successful patterns. The persistent trend towards these sta- intensity running and technical profiles in English FA Premier League soccer
ble blueprints, such as increasing player density, game speed and matches. J Sports Sci 2011; 29(8):821–830.
15. Carling C. Influence of opposition team formation on physical and skill-related
passing rates, within a sport having very few rule changes and performance in a professional soccer team. Eur J Sports Sci 2011; 11(3):
with parallels in other field sports, suggest common evolutionary 155–164.
pressures are operating. To increase the probability of game success 16. Glaister M. Multiple sprint work: physiological responses, mechanisms of
fatigue and the influence of aerobic fitness. Sports Med 2005; 35(9):757–777.
within a complex team sport requires rapid self-organisation in a 17. Mohr M, Krustrup P, Bangsbo J. Match performance of high standard soccer
dynamical system, underpinned by prediction speed and accuracy, players with direct reference to development of fatigue. J Sports Sci 2003;
skill and fitness. The trends found over the longer term suggests 21(7):519–528.
18. Elliott MCCW, Zarins B, Powell JW et al. Hamstring muscle strains in professional
professionals are getting better at these through the processes
football players: a 10-year review. Am J Sports Med 2011; 39(4):843–850.
of self-selection, conditioning, and positive feedback of successful 19. Grant AG, Williams AM, Reilly T. Analysis of goals scored in the 1998 World Cup.
strategies. J Sports Sci 1999; 17(10):826–827.
20. Luhtanen P, Korhonen V, Ilkka A.A new notational analysis system with special
reference to the comparison of Brazil and its opponents in the World Cup 1994, In:
Practical implications Reilly T, Bangsbo J, Hughes M, editors, London, Science and Football III, E & FN
Spon, 1997, p. 229–232.
21. Di Salvo V, Gregson W, Atkinson G et al. Analysis of high intensity activity in
• Selection needs to prioritise speed as a critical element in elite- premier league soccer. Int J Sports Med 2009; 30:205–212.
level soccer. 22. Rampinini E, Coutts AJ, Castagna C et al. Variation in top level soccer match
• Future game demands are likely to require faster decision-making performance. Int J Sports Med 2007; 28:1018–1024.
23. Bangsbo J, Iaia F, Krustrup P. The Yo-Yo intermittent recovery test: a useful tool
athletes. for evaluation of physical performance in intermittent sports. Sports Med 2008;
• Current trends suggest larger, powerful athletes will be tolerated 38(1):37–51.
by the sport because of the longer stoppage breaks facilitating 24. Pollard R, Ensum J, Taylor S. Estimating the probability of a shot resulting in a
goal: the effect of distance, angle and space. Int J Soccer Sci 2004; 2(1):50–55.
more complete recovery. 25. Arnason A, Sigurdsson S, Gudmundsson A et al. Physical fitness, injuries and
team performance in soccer. Med Sci Sports Exerc 2004; 36(2):278–285.
Acknowledgement 26. Junge A, Dvorak J. Soccer injuries: a review on incidence and prevention. Sports
Med 2004; 34(13):929–938.
27. Rahnama N, Reilly T, Lees A. Injury risk associated with playing actions during
No sources of funding were used to assist in this study. competitive soccer. Br J Sports Med 2002; 36(5):354–359.
28. Gerisch G, Rutemoller E, Weber K. Sport medical measurements of performance
in soccer, in Science and football, Reilly T, Lees A, Davids K, Murphy W, editors,
References London, E & FN Spon, 1988, p. 60–67.
29. Norton K. Match analysis in AFL, Soccer and rugby union: patterns, trends and
1. Drust B, Atkinson G, Reilly T. Future perspectives in the evaluation of the phys- similarities, in Chapter 24 in Science and Football VII: The Proceedings of the Seventh
iological demands of soccer. Sports Med 2007; 37(9):783–805. World Congress on Science and Football, Hiroyuki Nunome, Barry Drust, Brian
2. Carling C, Bloomfield J, Nelsen L et al. The role of motion analysis in elite soccer: Dawson, editors, London, Routledge, 2013.
contemporary performance measurement techniques and work rate data. Sports 30. Ali A, Williams C, Hulse M et al. Reliability and validity of two tests of soccer
Med 2008; 38(10):839–862. skill. J Sports Sci 2007; 25(13):1461–1470.