Activity Profile of International Rugby Sevens: Effect of Score Line, Opponent, and Substitutes

Abstract

Purpose:
To investigate the influence of score line, level of opposition, and timing of substitutes on the activity profile of rugby sevens players and describe peak periods of activity.

Methods:
Velocity and distance data were measured via 10-Hz GPS from 17 international-level male rugby sevens players on 2-20 occasions over 4 tournaments (24 matches). Movement data were reported as total distance (TD), high-speed-running distance (HSR, 4.17-10.0 m/s), and the occurrence of maximal accelerations (Accel, ≥2.78 m/s2). A rolling 1-min sample period was used.

Results:
Regardless of score line or opponent ranking there was a moderate to large reduction in average and peak TD and HSR between match halves. A close halftime score line was associated with a greater HSR distance in the 1st minute of the 1st and 2nd halves compared with when winning. When playing against higher- compared with lower-ranked opposition, players covered moderately greater TD in the 1st minute of the 1st half (difference = 26%; 90% confidence limits = 6, 49). Compared with players who played a full match, substitutes who came on late in the 2nd half had a higher average HSR and Accel by a small magnitude (31%; 5, 65 vs 34%; 6, 69) and a higher average TD by a moderate magnitude (16%; 5, 28).

Conclusions:
Match score line, opposition, and substitute timing can influence the activity profile of rugby sevens players. Players are likely to perform more running against higher opponents and when the score line is close. This information may influence team selection.

Full text

791
Original investigatiOn
International Journal of Sports Physiology and Performance, 2015, 10, 791 -801
http://dx.doi.org/10.1123/ijspp.2014-0004
© 2015 Human Kinetics, Inc.
Murray is with Aspire Academy, Doha, Qatar. Varley is with the Inst
of Sport, Exercise and Active Living, Victoria University, Melbourne,
Australia. Address author correspondence to Andrew Murray at andrew.
murray@aspire.qa.
Activity Profile of International Rugby Sevens:
Effect of Score Line, Opponent, and Substitutes
Andrew M. Murray and Matthew C. Varley
Purpose: To investigate the inuence of score line, level of opposition, and timing of substitutes on the activity prole of rugby
sevens players and describe peak periods of activity. Methods: Velocity and distance data were measured via 10-Hz GPS from
17 international-level male rugby sevens players on 2–20 occasions over 4 tournaments (24 matches). Movement data were
reported as total distance (TD), high-speed-running distance (HSR, 4.17–10.0 m/s), and the occurrence of maximal accelera-
tions (Accel, ≥2.78 m/s2). A rolling 1-min sample period was used. Results: Regardless of score line or opponent ranking there
was a moderate to large reduction in average and peak TD and HSR between match halves. A close halftime score line was
associated with a greater HSR distance in the 1st minute of the 1st and 2nd halves compared with when winning. When playing
against higher- compared with lower-ranked opposition, players covered moderately greater TD in the 1st minute of the 1st half
(difference = 26%; 90% condence limits = 6, 49). Compared with players who played a full match, substitutes who came on
late in the 2nd half had a higher average HSR and Accel by a small magnitude (31%; 5, 65 vs 34%; 6, 69) and a higher average
TD by a moderate magnitude (16%; 5, 28). Conclusions: Match score line, opposition, and substitute timing can inuence the
activity prole of rugby sevens players. Players are likely to perform more running against higher opponents and when the score
line is close. This information may inuence team selection.
Keywords: GPS, match analysis, physical performance, peak, fatigue
Rugby sevens games are played by 14 players (7 per team)
over 14 minutes (2 × 7-min halves, plus stoppage time). Squads
consist of a maximum of 12 players. All 12 are eligible to play in
each match, but each team may only make 3 substitutions from the
5 nonstarting players. Male international teams compete across 9
tournaments between October and May in the Sevens World Series.
Typically the competition is across 2 days; pool matches are held
on day 1 to determine a seeding for day 2, where teams compete
for World Series points. The World Series is generally scheduled
as sets of 2 tournaments with sets separated by 6 to 8 weeks. These
consecutive tournaments are held in different countries, potentially
on opposite sides of the world.
Data exist on the physical demands of rugby union,1,2 league,3,4
and sevens.5–7 Intensied periods of activity have been investigated
in other football codes, using predened time intervals to identify
peak periods of activity.8–10 However, predened periods can under-
estimate the peak distance covered by up to 25% compared with
using a rolling time scale.11 Thus, the use of rolling time periods
provides a more sensitive method for identifying peak periods of
activity. Currently the peak periods of running performed in rugby
sevens are unknown.
The activity prole of players in various football codes has
shown evidence of pacing within3,12 and between matches.13 There
are a number of variables that may inuence a player’s activity and
pacing strategy during competition. For example, player activity in
Australian football and rugby league can be inuenced by match
score line14 and the ranking of the opposition.15 In rugby sevens,
substitute players cover a greater total distance (TD) and high-
speed-running (HSR) distance per minute than those who play a
full match.6 Knowledge of how the different factors can inuence
match activity in rugby sevens can assist coaches in their tactical
decisions before and during competition (eg, squad rotation and
player substitutions). The aim of this study was to determine how
match score line, the ranking of the opposition, and the use of sub-
stitutes may inuence match activity. A secondary aim was to use
rolling periods to establish peak periods of activity in rugby sevens
players and to determine whether these periods are inuenced by
the aforementioned variables.
Methods
Participants and Experimental Design
Data were collected as part of an applied athlete support package
in elite sport by support staff for the purpose of informing training
and coaching decisions. Data were analyzed retrospectively for this
study; therefore, ethical approval was not obtained but informed
consent was.16
Player velocity was measured via global positioning system
(GPS) units (10-Hz MinimaxX S4, Catapult Innovations, Australia)
from 17 international-level male rugby sevens players on 2 to 20
occasions over 4 international tournaments (24 matches, 143 indi-
vidual les). The GPS units were applied as previously reported.12
The mean ± SD number of available satellite signals during matches
was 11.3 ± 1.4. Although we did not analyze positional differences
of the 17 players primarily, 10 were backs and 7 were forwards. It
is common in sevens for players to play a number of positions, and,
as such, 3 players played across both lines.

IJSPP Vol. 10, No. 6, 2015
792 Murray and Varley
Activity-Profile Measurements
Player-movement data were reported as TD and HSR17 (4.17–
10.0 m/s) distance. In addition, occurrences of maximal accelera-
tions (Accel, ≥2.78 m/s2) were recorded over a minimum duration
of 0.4 seconds.12 Velocity data were calculated using the Doppler
shift method, as opposed to the differentiation of positional data, as
it is associated with a higher level of precision.18 The 10-Hz GPS
is able to detect instantaneous changes of velocity during constant
movement and accelerations, with a percentage bias of –3.6% to
–0.6%, compared against a laser as the criterion measure, and a typi-
cal error expressed as a coefcient of variation of 3.1% to 8.3%.19
When assessed for measuring total distance during a team-sport
simulation circuit these 10-Hz GPS units have a reported error of
<1% compared with actual distance and a typical error of measure-
ment of 1.3% for TD and 4.8% for HSR distance.20 Furthermore,
these devices have an average bias of 6.5% for measuring distance
during a 30-m sprint.21 Raw GPS distance and velocity data were
analyzed using a custom Excel spreadsheet.
The average distance per count of each movement per half was
expressed in meters per count per minute of match time (m · count–1
· min–1). A rolling 1-minute sample period11 was used to identify
the peak 1-minute period for distance per count in each half. This
method provides a more sensitive measure of identifying peak peri-
ods of activity than traditional methods.11 Finally, the rst 1-minute
period at the start of each half for distance per count was recorded
in absolute terms (m/count). Player movement in the initial period
of the second half in comparison with the rst has been used as an
indicator of match fatigue in other football codes.22 Identifying how
this measure is inuenced by variables such as halftime score line
and opposition ranking assists in contextualizing the movements
undertaken by players. This information can assist coaches in their
tactical preparation for the second half at the halftime break and in
analyzing the outcome postmatch.
Relationship of Score Line to the Activity Profile
The relationship between the halftime and full-time score lines
and the activity prole was assessed using data from players who
had completed a full match. The score line was separated into 3
groups, close (7 points or less difference between team scores), lose
(reference team was losing or lost by more than 7 points), and win
(reference team was winning or won by more than 7 points). This
specic score line was chosen as 7 points represents the difference
of a converted (2 points) try (5 points) between teams.
Influence of Level of Opposition on the Activity
Profile
The relationship between the level of opposition and the activity
prole was assessed using data from players who had completed a
full match. The ranking of each opponent relative to the reference
team at the end of the World Series was used to group opponents as
either higher (nished above the reference team) or lower (nished
below the reference team). The reference team nished tenth of
the 12 core teams that contested every World Series tournament.
Of the 22 national teams that competed across the 4 tournaments
investigated, the reference team contested 10 matches against 7
teams that nished the World Series ranked above the reference
team and 14 matches against 10 teams ranked below.
Substitutes
Differences in the activity prole between players who played a full
match and substitutes were examined in the second half of a match.
Only players who were substituted into the match at halftime or
during the second half were included in the substitute group. As the
average duration of the second half was ~8 minutes, substitutes were
divided into those who came on in the rst 4 minutes (early subs)
and those who came on in the last 4 minutes (late subs).
Statistical Analysis
Separate analyses were performed for TD, HSR, and Accel using a
generalized linear mixed model (Proc Glimmix) using the statisti-
cal analysis system (SAS; Version 9.4, SAS Institute, Cary, NC).
Separate analyses were performed using each of the following as
xed main effects; match score line (win, lose, or close), relative
ranking of the opponent (higher or lower), the player’s match status
(full match, early sub, late sub), and match half (rst or second).
The following were analyzed as interactions; match half with match
score line and match half with relative ranking of the opponent. A
random effect for players and for each match was included in the
model to account for repeated measurement within and between
matches. The log-link function and the Poisson distribution were
invoked with an overdispersed residual to account for any cluster-
ing of counts.
An inference about the true value of a given effect (a difference
in means) was based on its uncertainty in relation to the smallest
important difference, which was determined by standardization
as 0.20 of the standard deviation between players in an average
match.23 This standard deviation was derived from the mixed model
by adding the variance for the true differences between players
(provided by the random effect for player identity) with the match-
to-match variance within players (provided by the overdispersion
factor multiplied by the mean, which is the Poisson variance). The
resulting observed between-players variances were different for
each level of a predictor variable, so the variances were averaged
across all levels before taking the square root.
Inferences were nonclinical. An effect was deemed unclear
if the 90% condence interval included the smallest important
positive and negative differences; the effect was otherwise deemed
clear. Quantitative chances of a greater or smaller substantial true
difference between levels of a predictor were calculated using pro-
gramming steps in SAS based on the same sampling theory that
underlies the calculation of traditional P values.24 These chances
were then assessed qualitatively for clear outcomes as follows:
>25% to 75%, possibly; >75% to 95%, likely; >95% to 99%, very
likely; >99%, almost certainly.
To deal with the ination of error in declaring a large number
of effects as substantial, we adopted a strategy adapted from pre-
vious literature.23 First, no adjustment was made for effects that
were only possibly substantial; these are regarded as potentially
substantial but requiring more data from a larger sample before
they can be implemented practically. Second, for each effect with a
higher likelihood of being substantial, the probability that the effect
was substantial but of opposite magnitude was calculated. These
probabilities were then rank ordered, lowest to highest, and were
summed sequentially until the probability exceeded 5%. The last
effect contributing to this sum and all remaining effects were rela-
beled as possibly substantial. The overall type I error for the effects

IJSPP Vol. 10, No. 6, 2015
7s Activity Profile With Rolling Periods 793
labeled as likely, very likely, and almost certainly was therefore less
than 5% (very unlikely), which is consistent with the acceptable
error rate for a study with only a single effect. The magnitude of a
given clear effect was determined from its observed standardized
value (the difference in means divided by the between-subjects
standard deviation) using the following scale: <0.20, trivial; 0.20
to 0.59, small; 0.60 to 1.19, moderate; ≥1.20, large.22
Results
The activity prole in the rst and second halves for players who
played a full match is presented in Table 1. Players had a moderate
to large reduction in average and peak TD and HSR in the second
half compared with the rst.
Relationship Between Score Line and Activity
Profile
The activity prole in the rst and second halves according to the
score line at halftime is presented in Table 2 and according to the
score line at full time in Table 3. Players had a moderate to large
reduction in average and peak TD and HSR in the second half com-
pared with the rst, irrespective of halftime score line. There was a
small reduction in average Accel in the second half compared with
the rst when the halftime score line was close or win. When the
halftime score line was win there was a moderate increase in rst
HSR and a small increase in rst TD in the second half compared
with the rst. When the full-time score line was close or lose, players
had a moderate to large reduction in average and peak TD and HSR
in the second half compared with the rst. There was a moderate
decrease in the average number of Accel in the second half com-
pared with the rst when the full-time score line was close. When
the full-time score line was close there was a moderate increase in
rst HSR in the second half compared with the rst.
The standardized differences in activity between different score
lines at both halftime and full time are shown for average (Figure
1), peak (Figure 2), and rst (Figure 3) HSR, TD, and Accel. In all
cases data are presented using rolling periods. When the full-time
score line was close the number of average Accel performed in the
second half was lower by a small magnitude compared with when
the full-time score line was win (percentage difference = –21%;
90% condence limits = –35, –4) or lose (–19%; –32, –3) (Figure
1[C]). Players covered a greater peak TD in the second half when the
score line was close or win at halftime compared with lose (22%; 8,
37 and 17%; 4, 31, respectively. Figure 2[B]). When the score line
was close at halftime compared with win, players covered a greater
HSR in the rst minute of the rst half by a large magnitude (96%;
42, 171) and in the rst minute of the second half by a moderate
magnitude (50%; 10, 103) (Figure 3[A]). Similarly, players cov-
ered a greater TD in the rst minute of the rst half by a moderate
magnitude when the halftime score line was close compared with
win (33%; 12, 57) (Figure 3[B]).
Influence of Level of Opposition on the Activity
Profile
The activity prole in the rst and second halves when the reference
team was competing against higher- or lower-ranked opposition
is presented in Table 4. When playing against both higher- and
lower-ranked opponents there was a moderate to large decrease in
peak and average HSR and TD in the second half compared with
the rst. There was a small reduction in peak Accel in the second
half compared with the rst when playing against lower-ranked
opponents. When playing against lower-ranked opponents there
was a moderate increase in rst HSR and a small increase in rst
TD in the second half compared with the rst.
The standardized differences in activity when playing against
a higher-ranked opponent compared with a lower-ranked opponent
in the rst and second halves are shown in Figure 4. When playing
against higher-ranked opposition, players covered a greater rst
TD in the rst half by a moderate magnitude compared with when
playing against lower-ranked opponents (26%; 6, 49). In the second
half average and peak Accel were greater by a small magnitude
when playing against higher- than lower-ranked opponents (24%;
7, 44 and 17%; 2, 36, respectively).
Table 1 Activity Profiles in the First and Second Halves of Male Full-Match Players During International Rugby
Sevens Matches
1st half, mean ± SD 2nd half, mean ± SD 2nd vs 1st half, ES (90% CL)
Average HSR (m/min) 28.3 ± 10.6 19.2 ± 7.6 –1.17 (–1.39 to –0.96)****
Peak HSR (m) 86 ± 30 68 ± 23 –0.80 (–1.05 to –0.55)****
First HSR (m) 31 ± 22 37 ± 25 0.39 (0.07–0.70)**
Average TD (m/min) 103 ± 15 86 ± 19 –1.24 (–1.46 to –1.02)****
Peak TD (m) 183 ± 30 164 ± 28 –0.79 (–1.04 to –0.53)****
First TD (m) 103 ± 34 110 ± 38 0.25 (–0.04–0.53)*
Average Accel (count/min) 1.3 ± 0.6 1.1 ± 0.6 –0.33 (–0.52 to –0.13)**
Peak Accel (count) 3.8 ± 1.6 3.7 ± 1.7 –0.09 (–0.32–0.15)00
First Accel (count) 1.7 ± 1.4 1.9 ± 1.5 0.18 (0.15–0.50)*
Abbreviations: HSR, high-speed running; TD, total distance’ Accel, acceleration; ES, effect size; CL, condence limits. Data = 24 matches, 16 players, 86 individual match
les. Clear substantial effects: *possibly, **likely, ***very likely, ****almost certainly, 00likely trivial.

794 IJSPP Vol. 10, No. 6, 2015
Table 2 Activity Profiles in the First and Second Halves of Male Full-Match Players During International Rugby
Sevens Matches at Different Halftime Score Lines
1st half, mean ± SD 2nd half, mean ± SD 2nd vs 1st half, ES (90% CL)
Average HSR (m/min)
close 31.0 ± 12.3 20.1 ± 6.4 –1.32 (–1.67 to –0.96)****
win >7 26.7 ± 8.9 18.7 ± 8.2 –1.08 (–1.39 to –0.78)****
lose >7 27.4 ± 12.1 18.7 ± 8.5 –1.17 (–1.86 to –0.48)***
Peak HSR (m)
close 94 ± 36 75 ± 21 –0.77 (–1.17 to –0.37)***
win >7 83 ± 26 66 ± 24 –0.77 (–1.12 to –0.42)****
lose >7 79 ± 24 57 ± 16 –1.13 (–1.96 to –0.30)***
First HSR (m)
close 42 ± 20 46 ± 26 0.23 (–0.24–0.70)
win >7 22 ± 18 33 ± 24 0.81 (0.32–1.31)***
lose >7 41 ± 27 31 ± 25 –0.56 (–1.53–0.41)
Average TD (m/min)
close 103 ± 17 84 ± 17 –1.35 (–1.72 to –0.99)****
win >7 102 ± 13 87 ± 20 –1.10 (–1.40 to –0.80)****
lose >7 103 ± 17 83 ± 21 –1.45 (–2.14 to –0.77)****
Peak TD (m)
close 188 ± 35 168 ± 26 –0.78 (–1.20 to –0.36)***
win >7 182 ± 28 165 ± 29 –0.66 (–1.01 to –0.32)***
lose >7 176 ± 20 142 ± 16 –1.48 (–2.30 to –0.66)***
First TD (m)
close 119 ± 33 123 ± 41 0.13 (–0.32–0.58)
win >7 89 ± 31 101 ± 38 0.53 (0.11–0.95)**
lose >7 119 ± 21 105 ± 17 –0.55 (–1.42–0.32)
Average Accel (count/min)
close 1.3 ± 0.6 1.1 ± 0.5 –0.34 (–0.66 to –0.01)**
win >7 1.3 ± 0.6 1.1 ± 0.6 –0.35 (–0.62 to –0.08)**
lose >7 1.1 ± 0.5 1.1 ± 0.6 –0.13 (–0.76–0.49)
Peak Accel (count)
close 3.8 ± 1.6 4.0 ± 1.9 0.11 (–0.28–0.49)
win >7 3.9 ± 1.6 3.5 ± 1.6 –0.22 (–0.55 to –0.10)*
lose >7 3.7 ± 1.4 3.6 ± 1.3 –0.08 (–0.82–0.67)
First Accel (count)
close 1.8 ± 1.3 2.1 ± 1.7 0.19 (–0.33–0.72)
win >7 1.5 ± 1.4 1.8 ± 1.4 0.30 (–0.16–0.77)*
lose >7 1.9 ± 1.7 1.3 ± 1.2 –0.58 (–1.66–0.50)
Abbreviations: HSR, high-speed running; TD, total distance’ Accel, acceleration; ES, effect size; CL, condence limits. Close data = 8 matches, 10 players, 31 individual
match les; win >7 data = 13 matches, 15 players, 46 individual match les; lose >7 data = 3 matches, 6 players, 9 individual match les. Clear substantial effects: *pos-
sibly, **likely, ***very likely, ****almost certainly.

795IJSPP Vol. 10, No. 6, 2015
Table 3 Activity Profiles in the First and Second Halves of Male Full-Match Players During International Rugby
Sevens Matches at Different Full-Time Score Lines
1st half, mean ± SD 2nd half, mean ± SD 2nd vs 1st half, ES (90% CL)
Average HSR (m/min)
close 29.7 ± 10.9 17.5 ± 7.3 –1.62 (–1.97 to –1.28)****
win >7 25.3 ± 8.7 22.2 ± 7.9 –0.40 (–0.82–0.01)*
lose >7 28.8 ± 11.4 19 ± 7.2 –1.27 (–1.61 to –0.93)****
Peak HSR (m)
close 90 ± 34 67 ± 27 –1.04 (–1.44 to –0.63)****
win >7 81 ± 23 70 ± 22 –0.47 (–0.98–0.04)*
lose >7 86 ± 31 69 ± 19 –0.78 (–1.19––0.37)***
First HSR (m)
close 30 ± 17 41 ± 28 0.64 (0.14–1.14)**
win >7 26 ± 21 26 ± 15 0.00 (–0.73–0.73)
lose >7 36 ± 26 41 ± 25 0.31 (–0.18–0.80)*
Average TD (m/min)
close 103 ± 16 82 ± 17 –1.57 (–1.92 to –1.21)****
win >7 105 ± 12 97 ± 17 –0.54 (–0.96 to –0.12)**
lose >7 102 ± 16 82 ± 19 –1.44 (–1.80 to –1.08)****
Peak TD (m)
close 184 ± 35 165 ± 30 –0.73 (–1.14 to –0.33)***
win >7 181 ± 21 174 ± 29 –0.28 (–0.78–0.23)
lose >7 185 ± 30 156 ± 23 –1.21 (–1.63 to –0.79)****
First TD (m)
close 100 ± 33 112 ± 50 0.48 (0.02–0.95)*
win >7 100 ± 38 106 ± 22 0.24 (–0.35–0.82)
lose >7 109 ± 33 109 ± 34 –0.46 (–0.45–0.47)
Average Accel (count/min)
close 1.2 ± 0.5 0.9 ± 0.5 –0.60 (–0.93 to –0.27)***
win >7 1.2 ± 0.6 1.2 ± 0.5 –0.10 (–0.49–0.28)
lose >7 1.3 ± 0.6 1.2 ± 0.6 –0.22 (–0.53–0.08)*
Peak Accel (count)
close 3.8 ± 1.7 3.4 ± 1.6 –0.30 (–0.69–0.09)*
win >7 3.8 ± 1.4 3.8 ± 1.6 0.00 (–0.48–0.48)
lose >7 3.9 ± 1.5 4.0 ± 1.9 0.06 (–0.32–0.44)
First Accel (count)
close 1.6 ± 1.5 2.2 ± 1.6 0.51 (–0.01–1.04)*
win >7 1.4 ± 1.2 1.7 ± 1.3 0.30 (–0.40–1.01)
lose >7 1.9 ± 1.4 1.7 ± 1.4 –0.26 (–0.79–0.27)
Abbreviations: HSR, high-speed running; TD, total distance’ Accel, acceleration; ES, effect size; CL, condence limits. Close data = 8 matches, 11 players, 33 individual
match les; win >7 data = 8 matches, 11 players, 21 individual match les; lose >7 data = 8 matches, 12 players, 32 individual match les. Clear substantial effects: *pos-
sibly, **likely, ***very likely, ****almost certainly.

796 IJSPP Vol. 10, No. 6, 2015
Figure 1 — Standardized differences between average high-speed running
(HSR), total distance covered (TD), and count of accelerations (Accel)
of male full-match players during international rugby sevens matches
at various halftime (HT) and full-time (FT) score lines in the rst and
second halves. (A) Average HSR (m/min), (B) average TD (m/min),
and (C) average Accel (counts/min). Close = 7 points or less difference
between team scores, lose >7 = reference team was losing by more than
7 points, win >7 = reference team was winning by more than 7 points.
Shaded column represents the smallest important difference. Quantita-
tive chances of higher or lower differences are evaluated according to
thresholds identied in statistical analysis: *possibly, **likely, ***very
likely, ****almost certainly.
Figure 2 — Standardized differences between peak high-speed running
(HSR), total distance covered (TD), and number of accelerations (Accel)
performed in a 1-minute period (peak) of male full-match players during
international rugby sevens matches at various halftime (HT) and full-time
(FT) score lines in the rst and second halves. (A) Peak HSR (m), (B)
peak TD (m), (C) peak Accel (count). Close = 7 points or less difference
between team scores, lose >7 = reference team was losing by more than
7 points, win >7 = reference team was winning by more than 7 points.
Shaded column represents the smallest important difference. Quantitative
chances of higher or lower differences are evaluated according to thresh-
olds identied in statistical analysis: *possibly, **likely, ***very likely,
****almost certainly.

IJSPP Vol. 10, No. 6, 2015
7s Activity Profile With Rolling Periods 797
Substitutes
The activity-prole data for players who played a full match or
were early or late subs are presented in Table 5. Early subs were on
the eld for 430 ± 86 seconds (ie, came on at or around halftime)
while late subs were on for 196 ± 79 seconds (ie, nal 3–4 min).
Players who came on as late subs performed greater average HSR
and Accel of a small magnitude and a greater average TD of a
moderate magnitude than those who played a full match (31%; 5,
65, 34%; 6, 69 and 16%; 5, 28). Late subs performed a lower peak
HSR by a moderate magnitude compared with those who played a
full match (–24%; –38, –7) and compared with early subs (–28%;
–42, –12). Late subs performed a lower peak TD by a moderate
magnitude compared with early subs and those who played a full
match (–10%; –18, –2 and –15%; –22, –7).
Discussion
In this study we characterized the activity proles of international-
level rugby sevens players using rolling periods and for the rst
time described peak periods of activity. In addition, the results
demonstrate that score line, the level of opposition, and the time
at which a player is introduced into the match can inuence match
activity prole.
Full-Match Activity Profile
Using rolling periods we have shown, for the rst time, the peak TD
and HSR distances covered in a 1-minute period (183 ± 30 and 86
± 30 m/min, respectively). These peak distances are inuenced by
variables such as the quality of the opponent (higher against better
opponents, Table 4). Thus, while the peak periods represent only
the average of the true peak distances a player may cover and are
likely to be subject to variation across individuals and positions,
they offer insight into the high running demand for which rugby
sevens players must be physically prepared.
The Scottish international players in this study covered an
average TD similar to those of Australian international25 and Span-
ish domestic26 rugby sevens players (~100 m/min). Furthermore,
when the distances covered above ~4 m/s in the aforementioned
studies were combined to represent a threshold similar to that used
in this study, all players covered a similar average HSR distance.
The average number of Accel per minute reported in this study (1.3/
min) was similar to that in Australian international players (~1.6/
min6; using a threshold of >2 m/s2 compared with 2.78 m/s2 in this
study). Notably, the players in this study performed a threefold
higher number of Accel per minute in a peak period than the aver-
age (Table 1). As it is more energetically demanding to accelerate
than it is to move at a constant velocity,27 this information further
highlights the high-intensity requirements of international rugby
sevens. Further work may investigate the impact of collisions or
high-intensity bouts28 during peak periods of activity.
Relationship Between Score Line and Activity
Profile
Research is equivocal in other football codes as to whether a close
score line leads to an increased amount of running. In rugby league,
moderate and large winning margins were associated with greater
relative distances compared with losing.15 In Australian football,
HSR distance per minute, sprints per minute, and peak speed were
higher for players in losing quarters.14 Furthermore, smaller score
Figure 3 — Standardized differences between high-speed running (HSR),
total distance covered (TD), and count of accelerations (Accel) performed
in the rst 1-minute period at the start of each half (rst) of male full-match
players during international rugby sevens matches at various halftime
(HT) and full-time (FT) score lines in the rst and second halves. (A) First
HSR (m), (B) rst TD (m), (C) rst Accel (count). Close = 7 points or less
difference between team scores, lose >7 = reference team was losing by
more than 7 points, win >7 = reference team was winning by more than
7 points. Shaded column represents the smallest important difference.
Quantitative chances of higher or lower differences are evaluated according
to thresholds identied in statistical analysis: *possibly, **likely, ***very
likely, ****almost certainly.

IJSPP Vol. 10, No. 6, 2015
798 Murray and Varley
Table 4 Activity Profiles in the First and Second Halves of Male Full-Match Players During International Rugby
Sevens Matches When Playing Against a Higher- or Lower-Ranked Opponent
1st half, mean ± SD 2nd half, mean ± SD 2nd vs 1st half, ES (90% CL)
Average HSR (m/min)
higher-ranked opponent 31.5 ± 9.8 20.5 ± 6.4 –1.31 (–1.63 to –0.99)****
lower-ranked opponent 25.8 ± 10.7 18.2 ± 8.3 –1.07 (–1.38 to –0.77)****
Peak HSR (m)
higher-ranked opponent 95 ± 27 72 ± 20 –0.98 (–1.35 to –0.61)****
lower-ranked opponent 80 ± 31 66 ± 25 –0.67 (–1.02 to –0.32)***
First HSR (m)
higher-ranked opponent 39 ± 24 40 ± 26 0.09 (–0.36–0.54)
lower-ranked opponent 26 ± 18 35 ± 24 0.68 (0.23–1.14)***
Average TD (m/min)
higher-ranked opponent 107 ± 14 89 ± 15 –1.25 (–1.58 to –0.92)****
lower-ranked opponent 99 ± 15 83 ± 21 –1.23 (–1.53 to –0.93)****
Peak TD (m)
higher-ranked opponent 192 ± 26 166 ± 28 –1.00 (–1.38 to –0.61)****
lower-ranked opponent 177 ± 31 162 ± 28 –0.62 (–0.96 to –0.28)***
First TD (m)
higher-ranked opponent 118 ± 33 116 ± 37 –0.08 (–0.49–0.34)
lower-ranked opponent 92 ± 31 105 ± 40 0.54 (0.14–0.94)**
Average Accel (count/min)
higher-ranked opponent 1.2 ± 0.5 1.2 ± 0.6 0.04 (–0.25–0.32)
lower-ranked opponent 1.3 ± 0.6 1.0 ± 0.5 –0.63 (–0.90 to –0.37)
Peak Accel (count)
higher-ranked opponent 3.7 ± 1.4 4.1 ± 1.8 0.25 (–0.10–0.60)*
lower-ranked opponent 3.9 ± 1.7 3.4 ± 1.5 –0.36 (–0.68 to –0.05)**
First Accel (count)
higher-ranked opponent 2.0 ± 1.5 1.9 ± 1.5 –0.09 (–0.57–0.38)
lower-ranked opponent 1.4 ± 1.3 1.8 ± 1.54 0.42 (–0.04–0.88)*
Abbreviations: HSR, high-speed running; TD, total distance’ Accel, acceleration; ES, effect size; CL, condence limits. Higher-ranked opponent data = 10 matches, 11
players, 37 individual match les; lower-ranked opponent data = 14 matches, 15 players, 49 individual match les. Clear substantial effects: *possibly, **likely, ***very
likely, ****almost certainly.
differentials were associated with increased TD and HSR distance.
This pattern is similar to our nding that players covered the greatest
peak HSR in the second half when the halftime score differential
was close (Figure 2[A]).
Notably, when the halftime score was close, players performed
more HSR in the rst minute of the second half than when winning
(Figure 3[A]). Similarly, players performed greater peak HSR in
both the rst and second halves when the halftime score was close.
This nding supports the idea that a close score line will encourage
players to work harder, as they may feel that the outcome of the
match is still in contention. This higher work rate may not be to
the team’s benet as it may result in players fatiguing earlier in the
second half. This may explain why a higher HSR pattern was not
associated with a close full-time score. Thus, the effect of score line
on match running may be inuenced by a combination of the time at
which a certain score line occurs and the match running performed to
that point (eg, the longer a score line stays close the longer a player
may run at a greater intensity, which may inuence the player’s
ability to respond to further changes in score line). There may be
implications for the use of tactical substitutions late in the match
to maintain a high level of match running that pressures opponents
in both attack and defense.6 There were not enough data available
to explore the relationship between the activity prole across the
whole match based on the real-time relative score line.
Effect of Opposition
When playing against higher- compared with lower-ranked oppo-
nents the players in this study covered a greater TD in the rst minute
of the rst half (Figure 4). Regardless of opponent ranking, peak and
average TD and HSR were reduced in the second half compared with
the rst (Table 4). However, when playing lower-ranked opponents,

IJSPP Vol. 10, No. 6, 2015
7s Activity Profile With Rolling Periods 799
Figure 4 — Standardized differences between total distance (TD), high-
speed-running distance (HSR), and the count of accelerations (Accel) of
male full-match players during international rugby sevens matches when
playing against an opponent ranked higher or lower than the reference team
in the rst and second halves. Opponents ranked higher are compared with
those ranked lower. Shaded column represents the smallest important dif-
ference. Quantitative chances of higher or lower differences are evaluated
according to thresholds identied in statistical analysis: *possibly, **likely,
***very likely, ****almost certainly.
TD and HSR were greater in the rst minute of the second half than
in the rst half (Table 4).
Player movement in the opening stages of a half may be inu-
enced by the level of the opposition. The greater distances may be
due to the ability of the higher-ranked opposition to dictate play by
maintaining possession of the ball29 and being direct,30 thus forcing
the reference team to run more. Greater average and peak HSR in
the rst half when playing higher-ranked opponents support the
effect of opposition on player movement, although the likelihood
of these small effects is only possible (Figure 4). In contrast, rugby
league players perform more HSR against bottom- compared with
top-ranked opponents.18 However, in this study a relative ranking
was not used, so it is unclear where the reference team was ranked
compared with their opponents.
Movement patterns similar to those in our study have been
observed in Italian soccer players who perform more TD and HSR
against better compared with poorer opponents.31 It is likely that
the players in our study did not have to run as hard when playing
lower-ranked opponents compared with higher-ranked opponents,
possibly reducing the level of fatigue experienced when commenc-
ing the second half. This result is supported by the increase in TD
and HSR at the start of the second half when playing lower-ranked
opponents (Table 4).
As players did not experience a substantial reduction in TD
and HSR in the rst minute of the second half in comparison
with the rst minute of the match, variables other than fatigue (ie,
opponent and halftime score line) may have a greater inuence on
movement at the start of the second half. Future research should
look at the interaction between score line and opposition ranking
on match running.
Impact of Substitutes
Players who were late subs covered a greater average TD than
players who played a full match. Late subs had higher average
HSR and Accel than players who played a full match. This pat-
tern is in agreement with previous ndings that substitutes (on the
eld <4 min) covered a greater average TD and HSR distance per
minute (24% and 123%, respectively) than players who played a
full match.6
The peak TD, HSR, and Accel were lower for late subs than
for players who played a full match and early subs. There are
several reasons that may explain these differences. Many require
additional contextual information to make conclusive inferences,
but we believe that the most likely is that late subs will have less
time on the eld and therefore less time over which a peak period
can be calculated.
Limitations
Players were not divided into positional groups to maintain group
sample size to investigate several levels of variables. Differences
in the distances covered at medium and sprinting speed have been
observed between forwards and backs in rugby sevens.26 Exploring
how external variables may inuence the movement of different
positions warrants further investigation in studies with larger data
sets. Differences in movement patterns have also been observed
between matches across a tournament. The competition structure
results in teams playing opponents that are more similarly matched
as the tournament progresses. This study did not account for the
order of the match in each tournament, which is a potential con-
founding variable that warrants further investigation.
Practical Applications
This study provides an average of the peak activity prole that male
players may experience during international level rugby sevens.
These peak periods of activity are inuenced by variables such as
opponent ranking, match score line, and when a player enters the
eld. Thus, coaches should be aware that these physically demand-
ing periods of activity in terms of both HSR and Accel are subject
to variation. As activity is reduced from the rst to second half
regardless of score line or opposition, there is a need for specic
conditioning to mitigate this decline in physical performance. We
encourage sport practitioners to be aware of the patterns identied
in this study and to determine whether they are directly applicable
to their own athletes.
Players are likely to perform greater peak periods of running
against higher-ranked opponents; this may affect a coach’s choice
of team. Furthermore, the timing of substitutions and match score
line at halftime and full time can inuence a player’s activity prole.
Teams and players may differ in their response to these variables.
Sport practitioners should determine how their team responds to
these variables and use this information to improve their decision
making when making tactical decisions for their own team during
rugby sevens tournaments.

800 IJSPP Vol. 10, No. 6, 2015
Table 5 Activity-Profile Differences Between Starting Players and Substitutes in the Second Half
Full match, n = 86,
mean ± SD Early sub, n = 34,
mean ± SD Late sub, n = 21,
mean ± SD Early sub vs full match,
ES (90% CL) Late sub vs full match,
ES (90% CL) Late sub vs early sub,
ES (90% CL)
Average HSR (m/min) 19.2 ± 7.6 21.1 ± 13.1 25 ± 15.2 0.31* (–0.11–0.73) 0.58** (0.10–1.06) 0.27 (–0.25–0.79)
Peak HSR (m) 68 ± 23 64 ± 28 48 ± 26 0.15 (–0.23–0.53) –0.70** (–1.21 to –0.20) –0.86*** (–1.39 to –0.32)
Average TD (m/min) 86 ± 19 92 ± 19 100 ± 28 0.14 (–0.26–0.54) 0.69*** (0.22–1.16) 0.55* (0.04–1.07)
Peak TD (m) 164 ± 28 154 ± 22 138 ± 37 –0.27* (–0.65–0.11) –0.88*** (–1.35 to –0.41) –0.61* (–1.12 to –0.09)
Average Accel (count/min) 1.1 ± 0.6 1.2 ± 0.8 1.5 ± 1.2 0.22* (–0.18–0.62) 0.55** (0.10–0.99) 0.32* (–0.16–0.81)
Peak Accel (count) 3.7 ± 1.7 3.7 ± 2.1 3.0 ± 2.4 0.17 (–0.21–0.56) –0.48* (–0.98–0.02) –0.65* (–1.19 to –0.12)
Abbreviations: HSR, high-speed running; TD, total distance’ Accel, acceleration; ES, effect size; CL, condence limits. Clear substantial effects: *possibly, **likely, ***very likely, ****almost certainly.

IJSPP Vol. 10, No. 6, 2015
7s Activity Profile With Rolling Periods 801
Conclusions
The data in this study demonstrate that match score line, opposition,
and timing of substitutions can inuence the activity prole of rugby
sevens players. Players are likely to perform more running when
playing against higher-ranked opponents and when the score line
is close. Finally, the activity prole of players is likely to decrease
across halves regardless of these factors.
Acknowledgments
We gratefully acknowledge the cooperation of players and interdisciplinary
support staff. We would like to thank Prof Will Hopkins for his assistance
with the statistical analyses. No external nancial assistance was obtained
for this study.
References
1. Roberts SP, Trewartha G, Higgitt RJ, El-Abd J, Stokes KA. The physi-
cal demands of elite English rugby union. J Sports Sci. 2008;26:825–
833. PubMed doi:10.1080/02640410801942122
2. Austin D, Gabbett TJ, Jenkins D. The physical demands of Super
14 rugby union. J Sci Med Sport. 2011;14:259–263. PubMed
doi:10.1016/j.jsams.2011.01.003
3. Waldron M, Highton J, Daniels M, Twist C. Preliminary evidence of
transient fatigue and pacing during interchanges in rugby league. Int
J Sports Physiol Perform. 2013;8:157–164. PubMed
4. Austin DJ, Kelly S. Positional differences in professional rugby
league match-play through the use of global positioning systems.
J Strength Cond Res. 2013;27:14–19. PubMed doi:10.1519/
JSC.0b013e31824e108c
5. Rienzi E, Reilly T, Malkin C. Investigation of anthropometric and
work-rate proles of rugby sevens players. J Sports Med Phys Fitness.
1999;39:160–164. PubMed
6. Higham DG, Pyne DB, Anson JM, Eddy A. Movement patterns in
rugby sevens: effects of tournament level, fatigue and substitute
players. J Sci Med Sport. 2012;15:277–282. PubMed doi:10.1016/j.
jsams.2011.11.256
7. Granatelli G, Gabbett TJ, Briotti G, et al. Match analysis and tem-
poral patterns of fatigue in rugby sevens. J Strength Cond Res.
2014;28:728–734. PubMed doi:10.1519/JSC.0b013e31829d23c3
8. Dellal A, Lago-Peñas C, Rey E, Chamari K, Orhant E. The effects
of a congested xture period on physical performance, technical
activity and injury rate during matches in a professional soccer
team. Br J Sports Med. 2015;49(6):390–394. PubMed doi:10.1136/
bjsports-2012-091290
9. Johnston RD, Gibson NV, Twist C, Gabbett TJ, Macnay SA, Macfar-
lane NG. Physiological responses to an intensied period of rugby
league competition. J Strength Cond Res. 2013;27:643–654. PubMed
doi:10.1519/JSC.0b013e31825bb469
10. Dupont G, Nedelec M, McCall A, McCormack D, Berthoin S, Wisloff
U. Effect of two soccer matches in a week on physical performance
and injury rate. Am J Sports Med. 2010;38:1752–1758. PubMed
doi:10.1177/0363546510361236
11. Varley MC, Elias GP, Aughey RJ. Current match-analysis techniques’
underestimation of intense periods of high-velocity running. Int J
Sports Physiol Perform. 2012;7:183–185. PubMed
12. Aughey RJ. Australian football player work rate: evidence of fatigue
and pacing? Int J Sports Physiol Perform. 2010;5:394–405. PubMed
13. Aughey RJ. Increased high-intensity activity in elite Australian football
nals matches. Int J Sports Physiol Perform. 2011;6:367–379. PubMed
14. Sullivan C, Bilsborough JC, Cianciosi M, Hocking J, Cordy J, Coutts
AJ. Match score affects activity prole and skill performance in profes-
sional Australian football players. J Sci Med Sport. 2014;17:326–331.
PubMed doi:10.1016/j.jsams.2013.05.001
15. Gabbett TJ. Inuence of the opposing team on the physical demands
of elite rugby league match-play. J Strength Cond Res. 2013;27:1629–
1635. PubMed doi:10.1519/JSC.0b013e318274f30e
16. Leeder J, Glaister M, Pizzoferro K, Dawson J, Pedlar C. Sleep duration
and quality in elite athletes measured using wristwatch ActiGraphy.
J Sports Sci. 2012;30:541–545. PubMed doi:10.1080/02640414.201
2.660188
17. Abt G, Lovell R. The use of individualized speed and intensity
thresholds for determining the distance run at high-intensity in
professional soccer. J Sports Sci. 2009;27:893–898. PubMed
doi:10.1080/02640410902998239
18. Townshend AD, Worringham CJ, Stewart IB. Assessment of speed
and position during human locomotion using nondifferential GPS.
Med Sci Sports Exerc. 2008;40:124–132. PubMed doi:10.1249/
mss.0b013e3181590bc2
19. 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;30:121–127. PubMed doi:10
.1080/02640414.2011.627941
20. Johnston RJ, Watsford ML, Kelly SJ, Pine MJ, Spurrs RW. The valid-
ity and reliability of 10 Hz and 15 Hz GPS units for assessing athlete
movement demands. J Strength Cond Res. 2014;28:1649–1655.
PubMed
21. Castellano J, Casamichana D, Calleja-Gonzalez J, Roman JS, Ostojic
SM. Reliability and accuracy of 10 Hz GPS devices for short-distance
exercise. J Sports Sci Med. 2011;10:233–234. PubMed
22. Lovell R, Barrett S, Portas M, Weston M. Re-examination of the post
half-time reduction in soccer work-rate. J Sci Med Sport. 2013;16:250–
254. PubMed doi:10.1016/j.jsams.2012.06.004
23. Hopkins WG, Marshall SW, Batterham AM, Hannin J. Progres-
sive statistics for studies in sports medicine and exercise science.
Med Sci Sports Exerc. 2009;41:3–13. PubMed doi:10.1249/
MSS.0b013e31818cb278
24. Batterham AM, Hopkins WG. Making meaningful inferences about
magnitudes. Int J Sports Physiol Perform. 2006;1:50–57. PubMed
25. Higham DG, Pyne DB, Anson JM, Hopkins WG, Eddy A. Comparison
of activity proles and physiological demands between international
rugby sevens matches and training [published online ahead of print
November 20, 2013]. J Strength Cond Res. PubMed doi:10.1519/
JSC.0b013e3182a9536f
26. Suarez-Arrones L, Arenas C, López G, Requena B, Terrill O, Mendez-
Villanueva A. Positional differences in match running performance and
physical collisions in men rugby sevens. Int J Sports Physiol Perform.
2014;9(2):316–323. PubMed doi:10.1123/IJSPP.2013-0069
27. Osgnach C, Poser S, Bernardini R, Rinaldo R, Di Prampero PE.
Energy cost and metabolic power in elite soccer: a new match
analysis approach. Med Sci Sports Exerc. 2010;42:170–178. PubMed
doi:10.1249/MSS.0b013e3181ae5cfd
28. Gabbett TJ, Jenkins DG, Abernethy B. Physical demands of pro-
fessional rugby league training and competition using microtech-
nology. J Sci Med Sport. 2012;15:80–86. PubMed doi:10.1016/j.
jsams.2011.07.004
29. Higham DG, Hopkins WG, Pyne DB, Anson JM. Patterns of play
associated with success in international rugby sevens. Int J Perform
Anal Sport. 2014;14:111–122.
30. Higham DG, Hopkins WG, Pyne DB, Anson JM. Relationships
between rugby sevens performance indicators and international tour-
nament outcomes. J Quant Anal Sports. 2014;10:81–87.
31. Rampinini E, Coutts AJ, Castagna C, Sassi R, Impellizzeri FM.
Variation in top level soccer match performance. Int J Sports Med.
2007;28:1018–1024. PubMed doi:10.1055/s-2007-965158