ArticlePDF Available

Abstract

Objectives The purpose of this study was to examine the match-to-match variation of match-running in elite female soccer players utilising GPS, using full-match and rolling period analyses. Design Longitudinal study. Methods Elite female soccer players (n = 45) from the same national team were observed during 55 international fixtures across 5 years (2012-2016). Data was analysed using a custom built MS Excel spreadsheet as full-matches and using a rolling 5-min analysis period, for all players who played 90-min matches (files = 172). Variation was examined using co-efficient of variation and 90% confidence limits, calculated following log transformation. Results Total distance per minute exhibited the smallest variation when both the full-match and peak 5-min running periods were examined (CV = 6.8-10%). Sprint-efforts were the most variable during a full-match (CV = 53%), whilst high-speed running per minute exhibited the greatest variation in the post-peak 5-min period (CV = 143%). Peak running periods were observed as slightly more variable than full-match analyses, with the post-peak period very-highly variable. Variability of Accelerations (CV = 17%) and Player Load (CV = 14%) was lower than that of high-speed actions. Positional differences were also present, with centre backs exhibiting the greatest variation in high-speed movements (CV = 41-65%). Conclusions Practitioners and researchers should account for within player variability when examining match performances. Identification of peak running periods should be used to assist worst case scenarios. Whilst micro-sensor technology should be further examined as to its viable use within match-analyses.
Please
cite
this
article
in
press
as:
Trewin
J,
et
al.
The
match-to-match
variation
of
match-running
in
elite
female
soccer.
J
Sci
Med
Sport
(2017),
http://dx.doi.org/10.1016/j.jsams.2017.05.009
ARTICLE IN PRESS
G Model
JSAMS-1527;
No.
of
Pages
6
Journal
of
Science
and
Medicine
in
Sport
xxx
(2017)
xxx–xxx
Contents
lists
available
at
ScienceDirect
Journal
of
Science
and
Medicine
in
Sport
journal
h
om
epage:
www.elsevier.com/locate/jsams
Original
research
The
match-to-match
variation
of
match-running
in
elite
female
soccer
Joshua
Trewina,b,c,,
César
Meylana,b,c,
Matthew
C.
Varleyd,
John
Cronina,e
aSports
Performance
Research
Institute
New
Zealand,
Auckland
University
of
Technology,
New
Zealand
bCanadian
Soccer
Association,
Canada
cCanadian
Sport
Institute—Pacific,
Canada
dInstitute
of
Sport,
Exercise
and
Active
Living,
College
of
Sport
and
Exercise
Science,
Victoria
University,
Australia
eSchool
of
Exercise,
Biomedical
and
Health
Sciences,
Edith
Cowan
University,
Australia
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
9
November
2016
Received
in
revised
form
28
March
2017
Accepted
16
May
2017
Available
online
xxx
Keywords:
GPS
High-speed
running
Accelerations
Player
Load
Accelerometer
a
b
s
t
r
a
c
t
Objectives:
The
purpose
of
this
study
was
to
examine
the
match-to-match
variation
of
match-running
in
elite
female
soccer
players
utilising
GPS,
using
full-match
and
rolling
period
analyses.
Design:
Longitudinal
study.
Methods:
Elite
female
soccer
players
(n
=
45)
from
the
same
national
team
were
observed
during
55
international
fixtures
across
5
years
(2012–2016).
Data
was
analysed
using
a
custom
built
MS
Excel
spreadsheet
as
full-matches
and
using
a
rolling
5-min
analysis
period,
for
all
players
who
played
90-min
matches
(files
=
172).
Variation
was
examined
using
co-efficient
of
variation
and
90%
confidence
limits,
calculated
following
log
transformation.
Results:
Total
distance
per
minute
exhibited
the
smallest
variation
when
both
the
full-match
and
peak
5-min
running
periods
were
examined
(CV
=
6.8–7.2%).
Sprint-efforts
were
the
most
variable
during
a
full-
match
(CV
=
53%),
whilst
high-speed
running
per
minute
exhibited
the
greatest
variation
in
the
post-peak
5-min
period
(CV
=
143%).
Peak
running
periods
were
observed
as
slightly
more
variable
than
full-match
analyses,
with
the
post-peak
period
very-highly
variable.
Variability
of
accelerations
(CV
=
17%)
and
Player
Load
(CV
=
14%)
was
lower
than
that
of
high-speed
actions.
Positional
differences
were
also
present,
with
centre
backs
exhibiting
the
greatest
variation
in
high-speed
movements
(CV
=
41–65%).
Conclusions:
Practitioners
and
researchers
should
account
for
within
player
variability
when
examining
match
performances.
Identification
of
peak
running
periods
should
be
used
to
assist
worst
case
scenarios.
Whilst
micro-sensor
technology
should
be
further
examined
as
to
its
viable
use
within
match-analyses.
©
2017
Sports
Medicine
Australia.
Published
by
Elsevier
Ltd.
All
rights
reserved.
1.
Introduction
The
use
of
technology
is
now
commonplace
in
the
examina-
tion
of
match-running
performance
in
team
sports.1Both
global
positioning
system
(GPS)
and
semi-automated
multi-camera
sys-
tems
(SAMCS)
have
been
used,
with
each
having
advantages
and
disadvantages.1Recent
law
changes
introduced
by
FIFA
have
allowed
the
use
of
GPS
in
official
matches,2allowing
practition-
ers
to
utilise
one
system
in
both
training
and
matches.
This
data
has
however
been
limited
in
the
elite
women’s
game
with
lit-
erature
sparse.3,4 It
is
important
to
therefore
understand
the
match-running
occurring
within
women’s
matches.
These
law
changes
also
bring
in
to
question
the
match-to-match
variation
using
this
technology
and
if
subsequent
inferences
are
meaning-
ful
or
not,
such
as
a
change
in
match-running
in
response
to
an
Corresponding
author.
Tel.:
+64
27
862
9816.
E-mail
address:
jtrewin@aut.ac.nz
(J.
Trewin).
external
factor
(e.g.
altitude
or
temperature).
Biological
variation
occurs
due
to
a
number
of
factors,
with
differences
in
match
situ-
ations
and
environments
a
likely
cause.5Further,
the
reliability
of
devices
to
detect
similar
movement
accurately
can
also
affect
the
match-to-match
variation
observed.
This
is
particularly
problem-
atic
when
movements
with
a
high-rate
of
change
are
examined,
where
variation
is
known
to
increase.6Therefore,
identification
of
the
match-to-match
variation
must
be
taken
into
account
when
analysing
match-running
performances.
Previous
studies
in
men’s
soccer
have
examined
match-to-
match
variation,5,7 with
researchers
reporting
high-speed
running
to
be
the
most
inconsistent
outcome
variable
of
interest
when
measured
with
SAMCS
(Co-efficient
of
variation,
CV
=
18–20%).
When
the
team
is
in
ball
possession,
variation
further
increased
(CV
=
31–32%),
with
authors
questioning
the
use
of
high-speed
run-
ning
as
a
performance
indicator
for
soccer
match
performances.
However,
CV
using
GPS
in
women’s
soccer
is
relatively
unknown,
it
is
therefore
essential
to
understand
the
match-to-match
variation
of
GPS
metrics
when
attempting
to
justify
with
certainty
changes
http://dx.doi.org/10.1016/j.jsams.2017.05.009
1440-2440/©
2017
Sports
Medicine
Australia.
Published
by
Elsevier
Ltd.
All
rights
reserved.
Please
cite
this
article
in
press
as:
Trewin
J,
et
al.
The
match-to-match
variation
of
match-running
in
elite
female
soccer.
J
Sci
Med
Sport
(2017),
http://dx.doi.org/10.1016/j.jsams.2017.05.009
ARTICLE IN PRESS
G Model
JSAMS-1527;
No.
of
Pages
6
2
J.
Trewin
et
al.
/
Journal
of
Science
and
Medicine
in
Sport
xxx
(2017)
xxx–xxx
in
match-running
in
relation
to
match-factors,
such
as
the
environ-
ment
or
score-line.
Although
this
inherent
variability
exists,
research
has
attempted
to
identify
transient
fatigue,
the
temporary
reduction
in
match-
running
in
response
to
a
period
of
high-intensity.8Researchers
examine
peak-periods
of
match
running
utilising
a
pre-set
time
period
(e.g.
5-min),
to
identify
where
match-running
is
at
its
greatest.4The
fatigue
profile
of
an
athlete
can
help
inform
con-
ditioning
protocols
and
tactical
decisions
made
by
coaches
within
a
match
or
matches.
French
Ligue
1
players
performed
71–121
m
of
high-speed
running
during
a
peak
period,
using
a
pre-set
5-min
period
(e.g.
0–5,
5–10
min),
analysed
using
SAMCS,5with
high
vari-
ation
from
match-to-match
(CV
=
24%).
The
ability
to
quantify
a
change
in
high-speed
running
from
peak-
to
post-periods
indicated
greater
variation
(CV
=
134%),
suggesting
identification
of
transient
fatigue
is
challenging.
The
assumption
that
a
decline
in
match-
running
from
peak
to
post
periods
is
fatigue
does
not
account
for
changes
in
tactical
instructions
or
stoppages
in
play.
Further,
the
use
of
pre-set
time
periods
(e.g.
0–5,
5–10
min,
etc.)
has
also
been
questioned,
with
rolling
periods
shown
to
more
accurately
identify
the
peak-period
of
running
by
as
much
as
25%
compared
to
pre-
set
periods.9However,
to
date
match-to-match
variation
of
this
technique
has
not
been
examined
and
requires
further
research.
Lastly,
with
the
ability
to
utilise
GPS
within
official
matches,
this
opens
up
the
collection
of
data
previously
not
possible,
such
as
accelerations
and
manufacturer
micro-sensor
metrics.
It
has
been
suggested
that
examination
of
total
distance
alone,
without
accounting
for
accelerations
or
decelerations,
may
underestimate
energy
expenditure
by
6–8%.10,11 Further,
maximal
accelerations
have
been
observed
to
occur
up
to
eight
fold
more
in
matches
than
sprinting.12 Micro-sensors
can
also
be
used
when
GPS
satellites
are
not
present,
such
as
in
enclosed
stadia.
This
technology
sam-
ples
at
a
much
greater
rate
and
therefore
may
be
more
sensitive
to
changes
in
performance
but
also
interference
from
noise.13 Under-
standing
the
magnitude
of
variation
with
respect
to
these
metrics
is
important
when
utilising
them
in
match
analyses.
As
match-running
is
of
particular
interest
to
practitioners,
the
quantification
of
match-to-match
variation
in
an
elite
female
pop-
ulation
is
of
importance,
particularly
with
respect
to
GPS
systems.
Researchers
have
suggested
the
use
of
a
single
reference
team
in
the
examination
of
match-to-match
variation.5Therefore,
the
pur-
pose
of
this
study
is
to
examine
the
match-to-match
variation
of
elite
female
soccer
players
from
a
single
national
team
during
full
international
matches
utilising
both
a
full
match
and
a
rolling
5-min
analysis.
2.
Methods
Elite
female
soccer
players
(n
=
45)
from
the
same
senior
national
team
ranked
top
10
in
the
world,
provided
informed
consent
to
participate
in
longitudinal
tracking
and
data
analy-
sis
which
was
approved
by
the
University
of
Victoria
Human
Research
Ethics
Board.
Player
movement
data
was
tracked
across
five
years
(2012–2016)
and
55
International
fixtures
(Files
=
606).
Only
outfield
players
were
included
in
the
current
study,
with
play-
ers
belonging
to
the
following
positional
groups,
forward
(FWD,
n
=
18);
midfield
(MF,
n
=
9);
full
back
(FB,
n
=
11);
and
centre
back
(CB,
n
=
7).
Where
a
player
played
in
multiple
positions,
data
in
each
position
were
analysed
separately
as
two
different
players,
due
to
the
known
positional
differences
in
match-running.14 Dis-
placement
and
velocity
data
was
collected
from
outfield
players
via
GPS
technology
sampling
at
10-Hz
(Minimax
S4,
Catapult
Inno-
vations,
Australia).
Raw
files
were
exported
from
manufacturer
software
(Sprint
5.1,
Catapult
Innovations,
Australia)
and
analysed
using
a
custom
built
MS
Excel
spreadsheet
(2013,
Microsoft,
United
States
of
America).12 Speed
was
calculated
using
the
Doppler
shift
method,
as
opposed
to
the
differentiation
of
positional
data
as
the
Doppler
shift
method
is
associated
with
a
higher
level
of
precision.15 The
average
number
of
satellites
and
horizontal
dilution
of
precision
for
games
was
12.1
±
0.4
and
0.94
±
0.04
respectively.
Players
were
required
to
play
a
minimum
of
two
90-min
match
performances,
with
the
analysed
files
ranging
from
of
2
to
21
games
played
per
player
(Mean
±
SD
=
7.0
±
6.2.
Data
were
only
included
from
games
played
in
“normative”
conditions,
consid-
ered
as
near
sea-level
(0–383
m)
and
in
cold/mild
temperature
(5–19 C)
(Files
=
154).
These
criteria
were
used
to
mitigate
the
pos-
sible
effects
that
some
environmental
factors
may
have
on
match
running
performance.16,17 Two
analyses
were
performed:
a
full
match
analysis;
and
a
5-min
rolling
analysis
period,
which
was
observed
as
a
match
maximum
of
both
the
peak
(Peak5)
and
the
subsequent
period
after
the
peak
(Post5).
The
rolling
analysis
cal-
culated
movement
in
5-min
increments
from
each
GPS
time
point,
of
which
there
were
10
per
second.9
Player
movement
categories
were
defined
following
locomo-
tor
analysis
guidelines
developed
using
elite
male
youth
players,
with
thresholds
set
using
pilot
data
of
women’s
players,
which
resulted
in
thresholds
similar
to
that
recommended
in
previ-
ous
research.18,19 High-speed
running
was
defined
as
an
effort
greater
than
4.58
m
s1,
which
represented
the
mean
maximal
aer-
obic
speed
(MAS)
of
the
team
observed
during
piloting.
Sprinting
(Sprint)
was
defined
as
an
effort
exceeding
5.55
m
s1,
a
threshold
representing
the
team
average
in
the
30–15
intermittent
fitness
test20 and
is
also
close
to
the
MAS
plus
30%
of
the
aerobic
speed
reserve
(e.g.
maximal
sprinting
speed
minus
MAS).
This
latter
method
has
been
used
in
previous
literature
to
individualise
maxi-
mal
speed
bands.18 Maximal
accelerations
were
defined
as
an
effort
greater
than
2.26
m
s2,
which
represented
80%
of
a
players
accel-
eration
over
10
m
during
a
40
m
sprint
test
and
was
established
during
piloting.
As
a
player
may
continue
to
accelerate
at
a
sub-
maximal
rate
following
a
maximal
acceleration,
an
acceleration
effort
was
defined
as
beginning
when
the
acceleration
exceeded
the
threshold
of
2.26
m
s2,
and
finishing
when
the
rate
of
acceler-
ation
dropped
below
0
m
s2.12 Acceleration
was
calculated
from
speed
data
over
a
0.3
s
time
interval.
Lastly,
GPS
were
coupled
with
a
100
Hz
accelerometer
and
used
to
estimate
Player
LoadTM,
an
arbi-
trary
value
developed
by
the
manufacturer
to
estimate
total
load
experienced
by
the
athlete.21 Total
distance,
high-speed
running
distance
and
Player
LoadTM were
presented
relative
to
total
match
time,
or
rolling
period
time
(/min).
The
number
of
high-speed
running-efforts,
Sprint-efforts
and
accelerations
were
presented
as
a
count
per
minute
of
match-play.
A
minimum
effort
duration
of
0.30
s
(i.e.
dwell
time)
was
applied
to
speed
data
(high-speed
running
and
Sprints).
Data
are
presented
as
means
±
SD
where
appropriate.
Co-
efficient
of
variation
(CV)
and
90%
confidence
intervals
were
calculated
after
logarithmic
transformation
in
MS
Excel.
Raw
values
were
log
transformed
using
a
natural
logarithm,
allowing
unifor-
mity
of
error.22 Further,
as
players
did
not
play
the
same
number
of
games,
a
weighted
CV
was
calculated
to
account
for
player
con-
tribution
to
the
variance.13 The
smallest
worthwhile
change
(SWC)
was
calculated
as
0.20
of
the
raw
between-player
SD,
prior
to
log
transformation.
The
SWC
can
be
used
to
assess
true
differences
in
performance,
observed
as
a
change
greater
than
the
SWC.23
3.
Results
The
match-to-match
variation
and
SWC
of
full
match
analysis
metrics
are
presented
in
Tables
1
and
2.
CV
values
ranged
from
6.1%
to
53%,
the
lowest
CVs
associated
with
the
total
distance
Please
cite
this
article
in
press
as:
Trewin
J,
et
al.
The
match-to-match
variation
of
match-running
in
elite
female
soccer.
J
Sci
Med
Sport
(2017),
http://dx.doi.org/10.1016/j.jsams.2017.05.009
ARTICLE IN PRESS
G Model
JSAMS-1527;
No.
of
Pages
6
J.
Trewin
et
al.
/
Journal
of
Science
and
Medicine
in
Sport
xxx
(2017)
xxx–xxx
3
Table
1
Match-to-match
variation
of
full
game
analysis
metrics
relative
to
minutes
played
by
position
and
overall,
presented
as
Mean
±
SD.
Positional
role
FB
CB
MF
FWD
Overall
Match
observations
N
=
24
N
=
44
N
=
56
N
=
30
N
=
154
Total
distance
(m
min1)
Mean
±
SD 110
±
9.2 100
±
7.3 115
±
7.9
108
±
10
108
±
10
CV
(90%
CI)
7.7
(6.1,
11)
7.2
(6.1,
8.9)
5.6
(4.8,
6.6)
7.9
(6.4,
10)
6.8
(6.2,
7.6)
SWC
1.7%
1.5%
1.4%
1.9%
1.9%
Low-speed
running
(m
min1)
Mean
±
SD
98
±
7.2
93
±
6.6
104
±
6.1
98
±
8.4
99
±
8.3
CV
(90%
CI)
6.7
(5.3,
9.2)
6.6
(5.2,
7.6)
5.2
(4.5,
6.2)
8.2
(6.7,
11)
6.5
(5.9,
6.8)
SWC
1.5%
1.4%
1.2%
1.7%
1.7%
High-speed
running
(m
min1)
Mean
±
SD
12.5
±
3.3
6.9
±
2.3
10.2
±
3.5
10.8
±
3.2
9.7
±
3.7
CV
(90%
CI)
31
(24,
42)
41
(34,
51)
31
(26,
36)
28
(23,
37)
33
(30,
37)
SWC
5.3%
6.7%
6.9%
6.0%
7.5%
Accelerations
(count
min1)
Mean
±
SD
1.95
±
0.29
1.96
±
0.35
1.65
±
0.34
1.81
±
0.28
1.82
±
0.35
CV
(90%
CI)
13
(10,
18)
22
(18,
27)
15
(13,
18)
13
(11,
17)
17
(15,
19)
SWC
3.0%
3.5%
4.1%
3.1%
3.8%
High-speed
running-efforts
(count
min1)
Mean
±
SD
0.78
±
0.17
0.46
±
0.15
0.70
±
0.20
0.70
±
0.18
0.64
±
0.21
CV
(90%
CI)
27
(21,
37)
45
(38,
55)
26
(22,
31)
28
(23,
36)
32
(29,
36)
SWC
4.3%
6.3%
5.6%
5.2%
6.5%
Sprint-efforts
(count
min1)
Mean
±
SD
0.28
±
0.10
0.14
±
0.06
0.20
±
0.09
0.26
±
0.09
0.21
±
0.10
CV
(90%
CI)
49
(39,
68)
65
(55,
81)
54
(46,
64)
35
(29,
46)
53
(49,
60)
SWC
6.9%
8.5%
9.1%
7.1%
9.4%
Player
Load
(AU
min1)
Mean
±
SD
10.6
±
1.5
10.3
±
1.7
13.2
±
2.5
10.6
±
2.4
11.5
±
2.5
CV
(90%
CI)
9.1
(7.2,
13)
14
(12,
17)
12
(10,
14)
20
(16,
25)
14
(13,
16)
SWC
2.9%
3.3%
3.8%
4.6%
4.5%
CV
=
Co-efficient
of
Variation;
CI
=
Confidence
Intervals;
min1=
per
minute
of
match-play;
SWC
=
smallest
worthwhile
change
as
a
percentage
of
the
mean.
Table
2
Absolute
mean
±
SD
and
CV
of
full
game
match-running
metrics.
Positional
Role FB
CB
MF
FWD
Overall
Match
observations
N
=
24
N
=
44
N
=
56
N
=
30
N
=
154
Total
Distance
(m)
Mean
±
SD
10,496
±
822
9533
±
650
10,962
±
750
10,380
±
893
10,368
±
952
CV
(90%
CI)
6.8
(5.4,
9.4)
6.6
(5.6,
8.2)
5.7
(4.9,
6.8)
7.1
(5.8,
9.2)
6.4
(5.8,
7.1)
SWC
1.6%
1.4%
1.4%
1.7%
1.8%
Low-speed
running
(m)
Mean
±
SD 9304
±
629
8872
±
594
9990
±
588
9343
±
739
9437
±
771
CV
(90%
CI)
5.5
(4.4,
7.7)
6.1
(5.2,
7.6)
5.5
(4.8,
6.6)
7.6
(6.2,
9.9)
6.1
(5.6,
6.8)
SWC
1.4%
1.3%
1.2%
1.6%
1.6%
High-speed
running
(m)
Mean
±
SD
1191
±
314
661
±
221
973
±
334
1037
±
305
930
±
348
CV
(90%
CI)
30
(24,
42)
40
(34,
49)
30
(26,
36)
28
(23,
36)
33
(30,
36)
SWC
5.3%
6.7%
6.9%
5.9%
7.5%
Accelerations
(count)
Mean
±
SD
185
±
27
187
±
33
158
±
33
174
±
27
174
±
33
CV
(90%
CI)
12
(9.5,
17)
21
(18,
26)
15
(13,
18)
13
(11,
17)
16
(15,
18)
SWC
2.9%
3.5%
4.2%
3.1%
3.8%
High-speed
running-efforts
(count)
Mean
±
SD
74
±
16
44
±
14
67
±
19
67
±
17
62
±
20
CV
(90%
CI)
27
(21,
37)
44
(37,
55)
25
(22,
30)
27
(22,
35)
32
(29,
35)
SWC
4.3%
6.3%
5.6%
5.1%
6.5%
Sprint-efforts
(count)
Mean
±
SD
26
±
9
14
±
6
20
±
9
25
±
9
20
±
9
CV
(90%
CI)
49
(39,
68)
65
(54,
80)
53
(46,
65)
35
(28,
45)
53
(48,
59)
SWC
6.9%
8.5%
9.2%
7.1%
9.4%
Player
Load
(AU)
Mean
±
SD
1007
±
147
982
±
159
1265
±
237
1016
±
226
1096
±
239
CV
(90%
CI)
8.5
(6.7,
12)
14
(11,
17)
12
(10,
14)
19
(16,
25)
14
(12,
15)
SWC
2.9%
3.2%
3.7%
4.4%
4.4%
CV
=
Co-efficient
of
Variation;
CI
=
Confidence
Intervals;
SWC
=
smallest
worthwhile
change
as
a
percentage
of
the
mean.
(6.4–6.8%)
and
low-speed
running
(6.1–6.5%).
High-speed
move-
ments
were
found
to
have
the
highest
variation,
as
observed
in
high-speed
running
(33%)
and
Sprint-efforts
(53%).
The
MF
group
were
the
least
variable
for
both
total
distance
(5.6–5.7%)
and
low-
speed
running
(5.2–5.5%),
compared
to
all
other
positions
for
both
metrics
(6.4–7.9%
and
6.1–8.2%,
respectively).
The
CB
group
indi-
cated
the
greatest
variation
for
all
high-speed
movements
(40–65%)
and
accelerations
(21–22%).
CV
values
during
the
rolling
5-min
analysis
ranged
from
7.2%
to
143%
for
all
players
(Table
3).
The
Peak5values
for
all
met-
Please
cite
this
article
in
press
as:
Trewin
J,
et
al.
The
match-to-match
variation
of
match-running
in
elite
female
soccer.
J
Sci
Med
Sport
(2017),
http://dx.doi.org/10.1016/j.jsams.2017.05.009
ARTICLE IN PRESS
G Model
JSAMS-1527;
No.
of
Pages
6
4
J.
Trewin
et
al.
/
Journal
of
Science
and
Medicine
in
Sport
xxx
(2017)
xxx–xxx
Table
3
Match-to-match
variation
of
Peak5and
Post5running
metrics
by
position
and
overall
as
Mean
±
SD.
Positional
Role
FB
CB
MF
FWD
Overall
Match
observations
N
=
27
N
=
43
N
=
59
N
=
32
N
=
161
Total
distance
Peak5
Metres
718
±
46 658
±
49 732
±
50
707
±
61
704
±
59
m
min1144
±
9.1
132
±
9.8
146
±
9.9
141
±
12
141
±
12
CV
(90%
CI)
6.8
(5.4,
9.4)
7.7
(6.5,
9.5)
7.1
(6.1,
8.5)
6.9
(5.6,
8.9)
7.2
(6.5,
8.0)
SWC
1.3%
1.5%
1.4%
1.7%
1.7%
Total
distance
Post5
Metres
551
±
88
498
±
64
152
±
21
543
±
82
540
±
84
m
min1110
±
18 100
±
13 113
±
17 109
±
16 108
±
16
CV
(90%
CI) 18
(14,
25)
14
(12,
17)
15
(13,
18)
15
(12,
19)
15
(14,
17)
SWC
3.2%
2.6%
3.1%
3.0%
3.1%
High-speed
running
Peak5
Metres
153
±
39
101
±
45
126
±
34
127
±
31
123
±
41
m
min130.7
±
7.9
20.1
±
9.0
25.2
±
6.7
25.4
±
6.1
24.6
±
8.2
CV
(90%
CI)
28
(22,
38)
44
(37,
54)
25
(21,
29)
21
(18,
28)
31
(28,
34)
SWC
5.1%
9.0%
5.3%
4.8%
6.7%
High-speed
running
Post5
Metres
48
±
25 24
±
19 43
±
25 44
±
22
38
±
24
m
min19.7
±
4.9
4.8
±
3.7
8.5
±
5.0
8.7
±
4.4
7.7
±
4.9
CV
(90%
CI)
64
(50,
88)
262
(221,
326)
113
(97,
136)
78
(64,
102)
143
(130,
159)
SWC
10%
16%
12%
10%
13%
Acceleration
Peak5
Count
18
±
2.7
17
±
2.9
15
±
2.6
17
±
3.7
17
±
3.2
count
min13.58
±
0.54
3.44
±
0.59
2.99
±
0.52
3.44
±
0.74
3.30
±
0.63
CV
(90%
CI) 16
(12,
21) 21
(18,
26) 17
(14,
20)
21
(17,
27)
19
(17,
21)
SWC
3.0%
3.4%
3.5%
4.3%
3.8%
Acceleration
Post5
Count
11
±
2.1
11
±
3.1
8.9
±
2.7
9.9
±
3.1
10
±
2.9
count
min12.18
±
0.42
2.16
±
0.63
1.79
±
0.54
1.99
±
0.62
2.00
±
0.59
CV
(90%
CI) 27
(21,
38) 37
(32,
46) 35
(30,
42) 34
(28,
45)
35
(28,
39)
SWC
3.9%
5.8%
6.0%
6.3%
5.9%
Player
Load
Peak5
AU
71
±
11
70
±
11
87
±
16
72
±
16
77
±
16
AU
min114.1
±
2.3
14.0
±
2.1
17.5
±
3.2
14.3
±
3.2
15.3
±
3.2
CV
(90%
CI)
9.3
(7.3,
13)
13
(11,
16)
12
(11,
15)
20
(16,
25)
14
(13,
15)
SWC
3.2%
3.0%
3.7%
4.5%
4.2%
Player
Load
Post5
AU
46
±
18
51
±
14
65
±
16
52
±
15
55
±
17
AU
min19.3
±
3.5
10.1
±
2.7
13.0
±
3.3
10.4
±
3.0
11.1
±
3.4
CV
(90%
CI)
19
(15,
25)
20
(16,
24)
26
(22,
31)
26
(22,
34)
23
(21,
26)
SWC
7.6%
5.4%
5.0%
5.8%
6.2%
CV
=
Co-efficient
of
Variation;
CI
=
Confidence
Intervals;
Peak5=
Peak
5-min
period;
Post5=
Post
5-min
period;
min1=
per
minute
of
match
play;
SWC
=
smallest
worthwhile
change
as
a
percentage
of
the
mean;
AU
=
Arbitrary
Units.
rics
were
similar
to
that
of
the
full
game
analysis,
within
plus
or
minus
3.0%.
The
Post5CV
values
increased
for
all
metrics,
princi-
pally
due
to
reduced
mean
values.
The
most
substantial
increase
in
the
CV
value
occurred
in
high-speed
running/min
during
Post5
compared
to
Peak5:
143%
versus
31%
respectively.
This
was
partic-
ularly
noticeable
for
the
CB
group,
where
the
Post5CV
was
262%,
compared
to
the
Peak5of
44%.
4.
Discussion
This
study
is
the
first
to
examine
the
match-to-match
variabil-
ity
of
match-running
performance
in
elite
female
soccer
players
using
GPS
technology.
This
is
also
the
first
study
to
include
findings
on
the
use
of
5-min
rolling
analysis
periods
and
maximal
acceler-
ations
within
an
elite
female
population.
The
major
findings
from
the
repeated
measures
analysis
were:
(1)
compared
to
high-speed
running
and
Sprints,
the
greater
occurrence
and
lower
variability
of
accelerations
warrants
addition
within
match
analyses;
(2)
Player
Load/min
may
be
used
with
relative
certainty,
however,
only
within
player
comparisons
should
be
considered;
(3)
Peak5periods
are
no
more
variable
than
full
match
analyses,
highlighting
their
ability
to
identify
the
period
of
greatest
match-running;
(4)
Post5periods
for
all
movements
are
substantially
more
variable
than
both
peak
peri-
ods
and
full
match
analysis,
limiting
their
use
in
identifying
within
match
transient
fatigue.
Researchers
have
suggested
a
possible
link
exists
between
accel-
erations
and
decelerations
and
post-match
neuromuscular
fatigue
and
energy
cost,11,24 which
suggest
particular
relevance
of
such
metrics
in
match
analysis.
The
lower
variability
observed
with
accelerations/min
in
this
study
(CV
=
17%)
compared
to
high-speed
running-efforts
and
Sprint-efforts
(CV
=
34%
and
56%
respectively),
contradicts
the
findings
of
previous
research25 and
demonstrate
its
suitability
to
be
tracked
from
match
to
match.
Differences
in
analysis
methods
could
result
in
the
lower
variation
observed
in
the
current
study,
such
as
the
definition
of
when
an
accelera-
tion
starts
and
finishes.
Acceleration/min
maybe
more
sensitive
to
worthwhile
changes
in
performance,
with
a
smaller
SWC
(3.8%),
compared
to
higher-speed
metrics
(6.4–9.2%).
However,
researchers
and
practitioners
are
still
advised
to
apply
caution
when
using
acceleration/min
and
interpreting
training
or
match
data.
Practitioners
should
also
be
aware
of
how
GPS
systems
are
detecting
and
defining
the
start/stop
of
an
acceleration
(e.g.
once
a
player
stops
accelerating
or
once
they
drop
below
the
threshold).12
This
point
in
particular
may
play
an
important
role
in
the
variation
of
accelerations.
To
the
authors
knowledge,
this
is
the
first
study
to
present
the
match-to-match
variation
of
Player
Load
in
soccer
match-play,
however,
a
strong
relationship
has
been
observed
between
Player
Load
and
total
distance
(r
=
0.70).26 Recently
researchers
of
field
hockey
have
report
the
relationship
between
Player
Load
and
total
Please
cite
this
article
in
press
as:
Trewin
J,
et
al.
The
match-to-match
variation
of
match-running
in
elite
female
soccer.
J
Sci
Med
Sport
(2017),
http://dx.doi.org/10.1016/j.jsams.2017.05.009
ARTICLE IN PRESS
G Model
JSAMS-1527;
No.
of
Pages
6
J.
Trewin
et
al.
/
Journal
of
Science
and
Medicine
in
Sport
xxx
(2017)
xxx–xxx
5
distance
in
both
training
and
match
scenarios.21 A
strong
rela-
tionship
(r
=
0.63–0.74)
was
observed
in
training
for
absolute
and
relative
metrics,
but
weakened
during
matches
when
examined
relative
to
minutes
played
(r
=
0.49).
Strength
of
association
was
primarily
different
by
position,
with
strikers
showing
the
weak-
est
relationship
in
both
relative
and
absolute
terms
(r
=
0.13–0.69)
during
matches.21 Practitioners
should
be
aware
of
the
challenges
associated
with
Player
Load,
with
accelerometer
data
sensitive
to
collisions,
tackles
and
jump
landings,
which
may
affect
the
mag-
nitude
of
such
data.27 With
that,
positional
demands
should
be
considered
when
interpreting
data
and
between
player
compar-
isons
avoided.28 It
would
seem
from
this
study
that
Player
Load/min
could
potentially
be
used
interchangeably
with
total
distance/min
with
relative
confidence
as
a
global
individual
load.
The
most-novel
aspect
of
this
study
was
the
examination
of
peak-periods
of
play,
using
a
rolling
5-min
analysis
period.
These
periods
were
first
introduced
in
an
attempt
to
identify
within
match
transient
fatigue.8,29 The
use
of
a
rolling
period
has
been
shown
to
better
observe
the
most
intense
period
in
a
match,
compared
to
using
pre-set
periods,30 however,
researchers
have
yet
to
define
the
match-to-match
variation
using
such
an
analysis.
Utilising
pre-
set
5-min
analysis
periods,
researchers
observed
Peak5of
total
high-speed
running
(>19.8
km
h1)
to
be
more
variable
(CV
=
24%)
than
the
full-match
analysis
(CV
=
20%).5In
this
study
the
oppo-
site
was
observed,
the
Peak5of
high-speed
running/min
exhibiting
slightly
less
variability
than
the
full
match-analysis,
whilst
acceler-
ation/min
and
Player
Load/min
were
similar
in
variability.
It
would
seem
that
Peak5may
be
a
useful
metric
to
identify
worst
case
sce-
narios
to
inform
conditioning
protocols.31 However,
the
variation
of
Peak5values
would
appear
specific
to
the
population
examined,
with
practitioners
advised
to
determine
population
specific
CVs.
Such
data
could
be
used
to
identify
peak
periods
of
play
with
the
aim
of
creating
drills
designed
to
replicate
this
intensity,
to
optimise
athlete
preparation
and
possibly
minimise
injury
risk.
Greater
uncertainty
surrounds
the
use
of
the
Post5metrics
as
a
means
to
identify
transient
fatigue
during
soccer
matches,
con-
sidering
the
variability
associated
with
this
method,
particularly
high-speed
running/min
(CV
=
133%,
SWC
=
12%).
Researchers
have
reported
changes
in
high-speed
running
(29–58%)
from
Peak5to
Post5in
female
soccer.3,4,32 However,
variability
of
the
change
between
Peak5and
Post5has
been
reported
as
high
as
134%
using
pre-set
periods.5Situational
factors,
such
as
tactics,
score-line
and
stoppages,9could
explain
the
large
variation
observed.
Further,
the
Post5period
may
not
be
the
period
of
lowest
intensity,
with
the
Post5
higher
than
the
mean
for
some
metrics
examined.
These
find-
ings
highlight
the
challenge
of
trying
to
examine
transient
fatigue
using
current
analytic
techniques.
With
the
possibility
that
match
situations
might
dictate
longer
or
shorter
periods
of
maximal
work-
rate,
5-mins
may
not
be
the
optimal
observational
period.33 Further
research
is
required
to
refine
the
use
of
peak-period
analyses
and
the
diagnostic
value
of
this
measure.
The
positional
findings
of
the
current
study
indicate
the
need
to
assess
variability
and
changes
within
player.
Whilst
the
larger
spread
of
the
CI
suggests
a
larger
data
set
is
required
to
exam-
ine
positional
data.
The
high-speed
running
and
Sprint-efforts
of
the
CB
group
observed
the
greatest
variation
compared
to
all
other
positional
groups
(CV
=
41–65%).
Whereas,
the
total
distance
and
low-speed
running
of
the
MF
group
indicated
the
least
vari-
able
metric
examined
(CV
=
5.2–5.6%).
These
findings
are
similar
to
recent
male
research,
which
found
Sprint
distance
(>25.2
km
h1)
to
be
the
most
variable
in
CB
(CV
=
45–58%)
compared
to
that
of
FB
(CV
=
22–31%).5This
data
highlights
the
need
to
use
advanced
sta-
tistical
methods
to
analyse
variation
and
changes
within
player.34
Whilst
practitioners
should
also
be
aware
of
analysing
individual
player
variation
when
making
inferences
on
player
match
perfor-
mances.
The
reader
needs
to
be
cognizant
of
a
number
of
limitations
when
reading
this
study.
Due
to
the
nature
of
elite
international
soccer,
only
55
games
were
played
throughout
a
five-year
period
limiting
the
sample
size
of
the
current
study.
Within
this
time
period,
player
physical
capabilities
may
have
changed,
which
may
have
increased
individual
variation
from
match-to-match,
however
this
was
not
examined
in
the
current
study
with
further
inves-
tigation
suggested.
This
limited
data
set
may
also
influence
the
positional
data
and
CI
observed.
With
researchers
and
practitioners
advised
to
assess
the
variation
of
their
own
data
sets
before
mak-
ing
inferences.
Further,
situational
and
environmental
factors
are
known
to
alter
player
performance,16,17,35 although
extreme
envi-
ronmental
data
was
excluded
from
this
study
in
order
to
limit
their
effects,
situational
factors
were
not
accounted
for.
Future
research
should
examine
the
effects
of
different
factors
on
the
variation
of
player
match-running.
The
examination
of
multiple
teams
is
also
suggested,
with
positional
variation
reported
to
further
highlight
the
need
to
distinguish
by
position
rather
than
generalise
by
team.
Finally,
the
results
are
only
relevant
to
the
team
investigated,
with
practitioners
advised
to
examine
variation
within
their
own
team.
5.
Conclusion
From
the
results
of
this
study
it
appears
that
the
match-to-match
variation
of
total
distance
is
acceptable
to
quantify
the
external
load
of
elite
female
soccer
players.
High-speed
activities
are
sus-
ceptible
to
higher
variation,
reducing
their
use
as
indicators
of
performance,
whilst
accelerations
should
be
included
in
match
analyses
to
account
for
the
most
energetically
demanding
activ-
ities.
The
use
of
Peak5to
identify
worst-case
scenarios,
the
most
intense
physical
load,
is
suggested.
Albeit
the
identification
of
tran-
sient
fatigue
using
the
Post5period
should
be
avoided.
Further,
examination
of
micro-sensor
technology,
including
relationships
with
GPS
metrics,
is
suggested
based
on
the
variability
of
the
mea-
sures
observed.
Practitioners
are
advised
to
examine
the
individual
variance
of
their
players
where
possible,
to
best
make
inferences
on
changes
in
match-running
performances
observed.
Practical
applications
Data
presented
is
specific
to
the
team
examined
and
should
not
be
generalised.
Researchers
should
explore
the
CV
of
their
reference
team
as
results
will
likely
differ,
whilst
interpretation
of
findings
will
also
be
strengthened.
Researchers
should
look
to
examine
a
large
sample
of
teams
using
mixed
linear
modelling
to
better
generalise
findings
whilst
accounting
for
within
player
variation.
Monitoring
of
accelerations/min
is
more
stable
than
that
of
high-
speed
running-efforts
and
Sprint-efforts
during
match
play.
The
greater
occurrence
of
accelerations
warrants
inclusion
of
such
data
by
practitioners
to
better
indicate
total
work/effort
and
energy
consumption.
Micro-sensor
use
appears
a
viable
alternative
when
GPS
technol-
ogy
is
not
available.
A
strong
link
has
been
indicated
between
Player
Load
and
total
distance,
however,
practitioners
should
be
aware
of
the
sensitivity
of
Player
Load
magnitude
to
a
variety
of
other
factors,
such
as
collisions.
Identification
of
Peak5match-running
periods
appears
similar
in
variability
to
that
of
full
match
analyses.
This
method
can
help
inform
conditioning
protocols
to
prepare
players
for
the
worst
case
scenarios
observed
within
match.
Please
cite
this
article
in
press
as:
Trewin
J,
et
al.
The
match-to-match
variation
of
match-running
in
elite
female
soccer.
J
Sci
Med
Sport
(2017),
http://dx.doi.org/10.1016/j.jsams.2017.05.009
ARTICLE IN PRESS
G Model
JSAMS-1527;
No.
of
Pages
6
6
J.
Trewin
et
al.
/
Journal
of
Science
and
Medicine
in
Sport
xxx
(2017)
xxx–xxx
Acknowledgements
There
has
been
no
financial
assistance
with
this
project.
References
1.
Buchheit
M,
Allen
A,
Poon
TK
et
al.
Integrating
different
tracking
systems
in
football:
multiple
camera
semi-automatic
system,
local
position
measurement
and
GPS
technologies.
J
Sports
Sci
2014;
32(20):1844–1857.
[2].
FIFA,
Accessed
2016
09th
May
http://www.fifa.com/about-fifa/news/
y=2015/m=10/news=fifa-and-ifab-to-develop-global-standard-for-electronic-
performance-an-2709918.html.
3.
Andersson
HA,
Randers
MB,
Heiner-Møller
A
et
al.
Elite
female
soccer
players
perform
more
high-intensity
running
when
playing
in
international
games
com-
pared
with
domestic
league
games.
J
Strength
Cond
Res
2010;
24(4):912–919.
4.
Datson
N,
Drust
B,
Weston
M
et
al.
Match
physical
performance
of
elite
female
soccer
players
during
international
competition.
J
Strength
Cond
Res
2016.
Ahead
of
Print.
5.
Carling
C,
Bradley
PS,
McCall
A
et
al.
Match-to-match
variability
in
high-speed
running
activity
in
a
professional
soccer
team.
J
Sports
Sci
2016:1–9.
Ahead
of
Print.
6.
Scott
MT,
Scott
TJ,
Kelly
VG.
The
validity
and
reliability
of
global
positioning
sys-
tems
in
team
sport:
a
brief
review.
J
Strength
Cond
Res
2016;
30(5):1470–1490.
7.
Gregson
W,
Drust
B,
Atkinson
G
et
al.
Match-to-match
variability
of
high-speed
activities
in
premier
league
soccer.
Int
J
Sports
Med
2010;
31(4):237–242.
8.
Waldron
M,
Highton
J.
Fatigue
and
pacing
in
high-intensity
intermittent
team
sport:
an
update.
Sports
Med
2014;
44(12):1645–1658.
9.
Varley
MC,
Elias
GP,
Aughey
RJ.
Current
match-analysis
techniques’
underesti-
mation
of
intense
periods
of
high-velocity
running.
Int
J
Sports
Physiol
Perform
2012;
7(2).
10.
Gaudino
P,
Iaia
FM,
Alberti
G
et
al.
Monitoring
training
in
elite
soccer
players:
systematic
bias
between
running
speed
and
metabolic
power
data.
Int
J
Sports
Med
2013;
34(11):963–968.
11.
Osgnach
C,
Poser
S,
Bernardini
R
et
al.
Energy
cost
and
metabolic
power
in
elite
soccer:
a
new
match
analysis
approach.
Med
Sci
Sports
Exerc
2010;
42(1):170–178.
12.
Varley
MC,
Aughey
RJ.
Acceleration
profiles
in
elite
Australian
soccer.
Int
J
Sports
Med
2013;
34(1):34–39.
13.
Meylan
C,
Trewin
J,
McKean
K.
Quantifying
explosive
actions
in
International
Women’s
Soccer.
Int
J
Sports
Physiol
Perform
2016;
12:310–315.
14.
Dellal
A,
Chamari
K,
Wong
DP
et
al.
Comparison
of
physical
and
technical
per-
formance
in
European
soccer
match-play:
FA
Premier
League
and
La
Liga.
Eur
J
Sport
Sci
2011;
11(1):51–59.
15.
Townshend
AD,
Worringham
CJ,
Stewart
IB.
Assessment
of
speed
and
position
during
human
locomotion
using
nondifferential
GPS.
Med
Sci
Sports
Exerc
2008;
40(1):124.
16.
Aughey
RJ,
Hammond
K,
Varley
MC
et
al.
Soccer
activity
profile
of
altitude
versus
sea-level
natives
during
acclimatisation
to
3600
m
(ISA3600).
Br
J
Sports
Med
2013;
47(Suppl.
1):i107–i113.
17.
Carling
C,
Dupont
G,
Le
Gall
F.
The
Effect
of
a
cold
environment
on
physical
activity
profiles
in
elite
soccer
match-play.
Int
J
Sports
Med
2011;
32(7):542–545.
18.
Mendez-Villanueva
A,
Buchheit
M,
Simpson
B
et
al.
Match
play
intensity
distri-
bution
in
youth
soccer.
Int
J
Sports
Med
2013;
34(2):101–110.
19.
Bradley
PS,
Vescovi
JD.
Velocity
thresholds
for
women’s
soccer
matches:
sex
specificity
dictates
high-speed
running
and
sprinting
thresholds—Female
Ath-
letes
in
Motion
(FAiM).
Int
J
Sports
Physiol
Perform
2015;
10:112–116.
20.
Buchheit
M.
The
30–15
intermittent
fitness
test:
10
year
review.
Myorobie
J
2010;
1(9):278.
21.
Polglaze
T,
Dawson
B,
Hiscock
D
et
al.
A
comparative
analysis
of
accelerometer
and
time-motion
data
in
elite
men’s
hockey
training
and
competition.
Int
J
Sports
Physiol
Perform
2015;
10(4):446–451.
22.
Hopkins
WG.
Measures
of
reliability
in
sports
medicine
and
science.
Sports
Med
2000;
30(1):1–15.
23.
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.
24.
Akenhead
R,
Hayes
PR,
Thompson
KG
et
al.
Diminutions
of
acceleration
and
deceleration
output
during
professional
football
match
play.
J
Sci
Med
Sport
2013;
16(6):556–561.
25.
Buchheit
M,
Al
Haddad
H,
Simpson
BM
et
al.
Monitoring
accelerations
with
GPS
in
football:
time
to
slow
down.
Int
J
Sports
Physiol
Perform
2014;
9(3):442–445.
26.
Casamichana
D,
Castellano
J,
Calleja-Gonzalez
J
et
al.
Relationship
between
indi-
cators
of
training
load
in
soccer
players.
J
Strength
Cond
Res
2013;
27(2):369–374.
27.
Arruda
A,
Carling
C,
Zanetti
V
et
al.
Effects
of
a
very
congested
match
schedule
on
body
load
impacts,
accelerations,
and
running
measures
in
youth
soccer
players.
Int
J
Sports
Physiol
Perform
2014;
10(2):248–252.
28.
Malone
JJ,
Lovell
R,
Varley
MC
et
al.
Unpacking
the
black
box:
applications
and
considerations
for
using
GPS
devices
in
sport.
Int
J
Sports
Physiol
Perform
2016:1–30.
Ahead
of
Print.
29.
Bradley
PS,
Sheldon
W,
Wooster
B
et
al.
High-intensity
running
in
English
FA
Premier
League
soccer
matches.
J
Sports
Sci
2009;
27(2):159–168.
30.
Varley
MC,
Fairweather
IH,
Aughey
RJ.
Validity
and
reliability
of
GPS
for
mea-
suring
instantaneous
velocity
during
acceleration,
deceleration,
and
constant
motion.
J
Sports
Sci
2012;
30(2):121–127.
31.
Gabbett
TJ,
Kennelly
S,
Sheehan
J
et
al.
If
overuse
injury
is
a
‘training
load
error’,
should
undertraining
be
viewed
the
same
way?
Br
J
Sports
Med
2016;
50:1017–1018.
[32].
Mohr
M,
Krustrup
P,
Andersson
H
et
al.
Match
activities
of
elite
women
soccer
players
at
different
performance
levels.
J
Strength
Cond
Res
2008;
22(2):341–349.
[33].
Varley
M,
Aughey
R.
One
minute
rolling
sampling
periods:
the
most
sensitive
method
for
identifying
transient
fatigue
in
soccer.
J
Sci
Med
Sport
2010;
13:e53.
[34].
Murray
AM,
Varley
MC.
Activity
profile
of
international
rugby
sevens:
effect
of
scoreline,
opponent
and
substitutes.
Int
J
Sports
Physiol
Perform
2014;
10(6):791–801.
[35].
Castellano
J,
Blanco-Villase˜
nor
A,
Álvarez
D.
Contextual
variables
and
time-
motion
analysis
in
soccer.
Int
J
Sports
Med
2011;
32(6):415–421.
<