Analysis of the characteristics of competitive badminton
D Cabello Manrique, J J González-Badillo
Br J Sports Med
Objective: To describe the characteristics of badminton in order to determine the energy requirements,
temporal structure, and movements in the game that indicate performance level. To use the findings to
plan training with greater precision.
Methods: Eleven badminton players (mean (SD) age 21.8 (3.26) years) with international experience
from four different countries (France, Italy, Spain, and Portugal) were studied. Two of the Spanish play-
ers were monitored in several matches, giving a total of 14 samples, all during the 1999 Spanish Inter-
national Tournament. Blood lactate concentration was measured with a reflective photometer.
Maximum and average heart rates were recorded with a heart rate monitor. Temporal structure and
actions during the matches were determined from video recordings. All variables were measured dur-
ing and after the game and later analysed using a descriptive study.
Results: The results confirmed the high demands of the sport, with a maximum heart rate of 190.5
beats/min and an average of 173.5 beats/min during matches over 28 minutes long and performance
intervals of 6.4 seconds and rest time of 12.9 seconds between exchanges.
Conclusions: The results suggest that badminton is characterised by repetitive efforts of alactic nature
and great intensity which are continuously performed throughout the match. An awareness of these
characteristics, together with data on the correlations between certain actions such as unforced errors
and winning shots and the final result of the match, will aid in more appropriate planning and monitor-
ing of specific training.
Previous research has analysed the characteristics of effort
in training and competition in racket sports such as
tennis12and squash.3This has made it possible to create a
proﬁle of the energetic and physiological demands involved,
which has enabled the physical capacity necessary to practise
these sports to be established. However, there are not
sufﬁcient data on badminton to allow a more realistic assess-
ment of energy expenditure in competitive matches.
It is not known which are the most important aspects of
performance in badminton, what should be improved to
increase playing level, and what factors lead to favourable
results when competitors are of a similar level. This is
compounded by ignorance of certain parameters of play that
can be related to performance in competition and their effect
on the ﬁnal result.
Research4–7 has shown the importance of studying physio-
logical variables such as lactate concentration and heart rate
to determine the energy requirements of physical activity and
Although badminton has increased in popularity since its
inclusion as an ofﬁcial sport in the 1992 Olympic Games in
Barcelona, research on performance capacity (the optimum
performance level players can reach, which should be used as
a point of reference for coaches of top players) is still scarce.
However, we should point out the studies of Carlson et al8and
Alvero9on heart rates obtained for young sportsmen during
competition. Minimum, average, and extremely high maxi-
mum heart rates were reported, and it was found that the
maximum heart rate recorded was close to the theoretical
heart rate maximum. With regard to adults, Hughes10 found
maximum heart rates of 186 beats/min in simulated competi-
Other studies include those by Abe et al11 and Gosh et al,12 13
in which the maximum concentrations of lactate were not
above 5 mmol/l. Similarly, Carlson et al8did not ﬁnd any large
differences between top class Australian singles and doubles
players, with a mean maximum concentration of 4.6 mmol/l
obtained at the end of the ﬁrst set of the men’s singles.
Another characteristic of badminton is the execution of
sporadic movements of moderate and high intensity, related to
repetitive actions of short duration but great intensity,14 as
occurs in other sports with similar characteristics (squash,
tennis, and volleyball). These characteristics, together with
highly explosive bursts of play, in the case of badminton tak-
ing place with high speed and technical skill within an 80 m2
court, serve to illustrate the degree of physical exertion in each
match. In terms of data analysis, some authors10 15 16 have
obtained average play values (action intervals) of ﬁve seconds
duration, followed by recovery periods of 5–10 seconds. How-
ever, Cabello et al,17 investigating three top class national play-
ers, found action intervals (performance time) closer to eight
seconds, and rest times were double this (16 seconds). This
ﬁnding does not agree with data obtained from a larger sam-
ple (n = 8) of younger, medium to top level national players,
for whom an average of 3.6 seconds for action intervals and
9.8 seconds for rest times were found,14 suggesting some vari-
ation in play-rest patterns.
Maximum effort tests on a treadmill have shown that top
badminton players have a high oxygen consumption: 60.5 and
49.3 ml/kg/min in men and women respectively of the
Australian national team8and 51.5 ml/kg/min for 13 players in
the British team.10
Our aim was to assess the physiological and metabolic bases
of physical effort during badminton competitions and their
possible relation to performance parameters as effective indi-
cators of the ﬁnal result. We also aimed to calculate the volume
and intensity of the work rate in a badminton match, measure
the cardiovascular effort in a top level match, describe the
temporal structure of a badminton competition in relation to
the time of action and recovery, extract quantiﬁable variables
of performance in play, and calculate the relation between the
different variables and the ﬁnal result of the match.
Eleven badminton players (10 men, one woman, mean (SD)
age 21.8 (3.26) years) with international experience were
See end of article for
Mr Cabello, Faculty of
Education, University of
Universitario de La Cartuja,
18071 Granada, Spain;
Accepted 7 May 2002
recruited from four different countries (France, Italy, Spain,
and Portugal). Two of the Spanish players were videorecorded
during several matches in the 1999 Spanish International
Tournament, giving rise to a total of 14 samples. Seven players
were members of their national teams (10 samples), with a
minimum weekly training schedule of 12 hours.
The design is descriptive and comprises measurements taken
before and after the match in each sample and various
comparisons between the subjects. Several variables were
measured; some as used by Dias and Ghosh18 are described as
This was determined, in blood samples (20 µl) taken from the
earlobe, using reactive strips which were immediately
analysed using the lactate-mediator oxidase colour reaction
technique with an Accusport reﬂection photometer. This tech-
nique is considered to be highly accurate for lactate
concentrations below 8 mmol/l.19 20 Several blood samples were
taken: at rest, at the end of the match, and at 1, 3, 5, 7, and 10
minute intervals during recovery.
Heart rate was recorded every ﬁve seconds by telemetry
throughout the match, using Polar Sport Tester 4000 heart rate
monitors and a Polar Interface Unit to introduce the data into
a 486 PC. Polar-HR software was used to determine
parameters such as the maximum and average heart rate of
each subject at different stages of the match. The Polar Inter-
face Unit consists of a belt containing a pair of electrodes and
a wrist receptor. The former is attached to the player’s chest
and sends the information it receives to the wrist receptor.
Total playing time
Total playing time was registered by the heart rate monitor’s
chronometer, as well as by ﬁlming each match with VHS-C
video cameras. From the tapes we were able to determine real
time play (total time that the shuttlecock is moving).
The temporal structure was obtained from subsequent analy-
sis of the videotaped matches by calculation of the average
work interval or performance time, average rest interval or rest
time (both of the latter measured in seconds), and work den-
sity (ratio of performance time to rest time; a non-
Performance rates were analysed by watching the video of
each match and summarising them as follows.
•Unforced errors: errors committed by the player in a situa-
tion where an error is not expected.
•Winning shots: shots that, on account of their effective
execution, score a point.
•Number of shots that occur in each point (number of shots
per rally), in each of the sets and matches (total shots).
•Maximum actions: actions that require highly demanding
physical execution and therefore maximum effort.
•Total number of rallies: number of interventions occurring
throughout the sets and whole match.
Atmospheric conditions (temperature, humidity, and pres-
sure) showed only minimal variations (70–80° humidity and
20–22° outside temperature).
It was necessary to take account of a factor that may have
affected the results obtained—that is, the intensity with
which the match was contested—because, to consider the
results obtained as representative of maximum effort, the
match had to be hard fought with the players performing to
their full capacity. This was controlled by choosing for analy-
sis the matches that a priori were likely to be the most hotly
contested—that is, between players of the same level (using
information obtained from results in other competitions and
from their respective coaches)—as well as the importance of
the result within the competition.
Before the competition, a meeting was held with all those
responsible: club delegates, referee, and members of the com-
petition committee. At this meeting, the bases of the study
were laid out and queries were answered. At the end,
information sheets and agreement forms were handed out to
be ﬁlled in by the players who volunteered to take part.
Before the on court warm up, a blood sample was taken for
analysis of basal lactate concentration, and each player was
Table 1 Detailed individual results, after elimination of data that, on account of
abnormal circumstances of play, may have altered the coherence of the result—for
example, three set matches influenced the mean values of some variables, causing
Variable (n=14) Mean SD CV Max Min
Age (years) 21.79 3.26 14.97 28 17
Weight (kg) 67.54 8.70 12.88 84 56
Height (cm) 175.21 6.83 3.90 182 165
Max heart rate (beats/min) 190.57 5.50 2.89 201 186
Average heart rate (beats/min) 173.43 8.86 5.11 187 162
Max lactate (mmol/l) 3.79 0.91 24.11 5.1 2.4
Total time (seconds) (n=12) 1689.33 312.89 18.52 2308 1320
Real time (seconds) (n=12) 548.75 98.62 17.97 696 387
Performance time (seconds) 6.40 1.25 19.61 8.86 4.57
Rest time (seconds) 12.93 2.68 20.76 18.7 9.2
Work density 0.49 0.06 11.42 0.61 0.4
Winning shots (number) 18.85 8.51 45.17 32 8
Unforced errors (n=12) 22.46 7.68 34.18 32 7
Shots per rally 6.06 1.08 17.86 7.82 4.6
Maximum actions 43.42 16.59 38.22 87 24
Total rallies 83.33 11.03 13.24 99 64
Total shots 510.75 109.76 21.49 774 354
Average/max heart rate coefficient 91.00 2.50 2.75 94.44 86.63
Number of shots/real time 0.93 0.11 11.40 1.13 0.69
CV, Coefficient of variation; max, maximum; min, minimum.
Badminton characteristics 63
provided with a heart rate monitor and reminded of the pro-
cedure to be adhered to.
Just before the start of the match, the player switched on
the heart rate monitor and so initiated the heart rate register.
At the same time, the video recorder started ﬁlming. A blood
sample was taken and heart rate was recorded immediately
after the ﬁrst set. At the end of the match, blood samples were
taken immediately and at 1, 3, 5, 7, and 10 minute intervals
Over the next few days the videotaped matches were
watched to determine each player’s temporal structure.
We carried out a bivariate correlational statistical analysis
(Pearson’s coefﬁcient) between all the variables using the Sta-
tistical Package for Social Sciences (SPSS) 7.5, and those vari-
ables, such as total time, that could affect correlations were
subject to partial control through this programme. For the
relation between unforced errors or winning shots and the
ﬁnal result, we used a binomial analysis of ratios for
non-parametric measures with c2contingency tables. The level
of signiﬁcance used in the statistical analysis was p<0.05.
Table 1 shows the different variables analysed for each player,
as well as their descriptive analysis.
The very high degree of uncertainty in the course of the
game of badminton is clearly shown by the great variability in
the different variables for each player, with performance times
varying between 4.57 and 8.86 seconds and lactate concentra-
tions varying between 2.4 and 5.1 mmol/l. However, some
variables are more constant such as maximum heart rate dur-
ing the match (186–201 beats/min, which is very close to the
real maximum of each player), the average heart rate
(162–187 beats/min), work density (0.4–0.6), and the relation
between the number of shots and real time as a possible indi-
cator of effort (1.1–0.7).
The tendency was for all the players who were analysed to
increase their average heart rate in relation to their maximum
heart rate of the playing interval as the match progressed
(>3% in all cases). This indicates that cardiovascular demand
gradually increased as the game progressed.
It is clear from ﬁg 1 that the most frequently occurring ral-
lies in all the matches analysed are those that have a perform-
ance time of 3–6 seconds, representing 40% of the total moves,
which, together with the rallies of 0–3 seconds (19%) and
Figure 1 Mean percentage of playing intervals (performance time) and recovery (rest time) of all the matches.
Table 2 Summary of the correlational analysis between the studied variables
Value p Value
Average heart rate & max heart rate 0.91 0.000
Average heart rate & lactate 0.55 0.051
Average heart rate & lactate (controlling total time) 0.48 0.127
Real time & rest time 0.62 0.031
Real time & performance time 0.73 0.007
Performance time & rest time 0.87 0.000
Shots in rally & real time 0.63 0.028
Shots in rally & performance time 0.75 0.007
Real time & total time 0.69 0.013
Max heart rate & work density −0.57 0.041
Max heart rate & work density (controlling total time) −0.54 0.082
Average heart rate & work density −0.57 0.038
Average heart rate & work density (controlling total time) −0.55 0.079
Max heart rate & maximum actions −0.62 0.028
Max heart rate & maximum actions (controlling total time) −0.66 0.025
Average heart rate & maximum actions −0.59 0.042
Average heart rate & maximum actions (controlling total time) −0.64 0.033
NB some correlations that originally seemed to be significant have stopped being so after controlling total
64 Cabello, González-Badillo
those of more than 6–9 seconds (20.3%), constitute more than
80% of the total rallies. Moreover, there is a progressive
decrease in the frequency with which rallies occur as the per-
formance time of each rally increases, with the rallies that last
more than 21 seconds being less than 1%. However, for the rest
intervals, it is between the third and ﬁfth duration—that is,
between more than 6 and 15 seconds—that 80% of the
Table 2 gives the results of a bivariate correlational statisti-
cal analysis (Pearson’s coefﬁcient) between all the variables
analysed. The goal was to determine the signiﬁcance of the
relation between pairs of variables by controlling total time in
those cases where this could affect the relation. Most of the
positive correlations were related to the temporal structure:
total and real time; performance and rest time. However, in the
case of the negative correlations, the variables that were found
to be related were the different expressions of heart rate, as
well as temporal structure and playing performance variables.
When the total time was controlled, it was found that the
maximum and average heart rate were no longer signiﬁcant in
relation to the work density, leaving only the relation between
maximum heart rate, average heart rate, and maximum
actions as signiﬁcant.
The binomial analysis of ratios for the study of unforced
errors and winning shots with the ﬁnal study of each set and
match shows us the degree of relation between the different
variables. In most cases, a high number of unforced errors
resulted in loss of the set or match. In six of the 14 matches
analysed using a χ2value of 7.095 and a level of signiﬁcance of
p<0.01 (0.00773), the difference between the number of
unforced errors and players winning or losing the match was
signiﬁcant. In these matches, the players with a higher
number of unforced errors were the ones who lost the match.
By analysing the correlations in table 2 and comparing the
results with those from other studies, a speciﬁc analysis of
each variable can be made.
Maximum lactate concentrations showed a mean of 3.8
mmol/l, with a maximum value of 5.1 and a minimum of 2.4,
which are similar to those found by Ghosh et al12 in top level
13–14 year old players. However, they are lower than those
obtained by Cabello et al in three top level senior Spanish play-
ers (mean 7.1 mmol/l17) and eight medium to top level players
(mean 5.7 mmol/l).14 These differences may be explained by
the differences in age, ﬁtness, and training levels of the
subjects. Although there is no correlation between lactate
concentration and the other variables, it is in the matches
where the smallest differences between work and rest (work
density) with longer work intervals occurred that lactate con-
centrations were highest (in four of the cases in which work
density was >0.5, lactate concentration was >3.5 mmol/l,
whereas in three of the cases in which work density was <0.5,
lactate concentration was <3 mmol/l). These lower values may
also be due to a greater aerobic work capacity conditioned by
a higher degree of previous training.10
One noteworthy aspect is the contrast between the low lac-
tate production, with an average over all the samples lower
than 3.8 mmol/l, and the high intensities shown in both
maximum (190 beats/min) and average (173 beats/min) heart
rate. This ﬁnding has no obvious explanation.
The high values for maximum and average heart rate obtained
here are similar to those obtained in other studies on heart
rate,91112in most cases approaching the theoretical heart rate
maximum (220−age). A possible explanation21 is that they are
due to neurophysiological factors that affect the heart. The
vegetative nervous system (speciﬁcally the sympathetic nerv-
ous system) may increase heart beat frequency by acting on
the sinoatrial node. The main neurotransmitter released by
the postganglionic ﬁbres of the sympathetic system during
physical activity is usually norepinephrine (noradrenaline),
which, by means of messages sent by the nervous system to
the heart, increases the heart rate to the limits required by the
activity itself. Certain characteristics of badminton (speed,
reﬂexes, precision, high level of concentration) produce a high
level of stress, which may give rise to further epinephrine
(adrenaline) secretion at the suprarenal gland level, in
addition to that required by the nervous system. This may
cause the heart to accelerate over and above that provoked by
the actual effort.10 15 16
The correlations between the different variables relating to
playing time (total playing time, real time, average work, and
rest interval) may explain the importance of events that occur
throughout a game of badminton, a competitive sport in
which individual differences and the game’s dynamics can
vary greatly from one game to another. However, the total
playing time may modify the relation between work density
and average and maximum heart rate, because a player does
not start playing until he or she feels ready to do so. This may
result in games with a long total playing time but quite a
short real playing time. A work density such as this (0.48)
may explain why maximum lactate concentration is not
obtained at the end of the game, because the rest intervals,
especially those in the second set, may be sufﬁcient to estab-
lish a partial decrease between one point and the next. It is
logical to assume that, in more highly demanding games in
which the variables related to playing times are considerably
greater, the physiological response will also be higher,
because the opponent’s level and the importance of the com-
petition can inﬂuence the intensity of the game, as well as the
values obtained. From an analysis of a video of the men’s sin-
gles ﬁnal at the 1992 Olympic Games in Barcelona, it is inter-
esting to note that, although the ﬁnal result of the game was
two sets to love, the total and real playing times (55 minutes
and 25 minutes 26 seconds for the ﬁrst and second set
respectively) are far longer than in the games we studied:
double the total time (28 minutes 9 seconds) and nearly three
times the real time (9 minutes 9 seconds). Moreover, there are
large differences in the work interval (6.4 seconds in our
study compared with 12.3 seconds in the Olympic ﬁnal) and
rest time (12.9 and 20.4 seconds respectively). The work/rest
ratio (0.62) was more than 20% higher than in our study
(0.49), which means that the chances of recovery are less and
therefore the degree of accumulated fatigue is greater.
The high positive correlation (r= 0.87) found between
performance time and rest time (p<0.001) conﬁrms that the
longer the point, the greater the time interval required for
recovery, something that is fairly logical and on occasions is
limited by the intervention of the umpire, who is trying to
avoid time wasting.
Similarly, the high correlation between work interval and
the number of shots in a rally shows that the number of shots
that can be played, usually one shot per second, is quite
limited by the shuttlecock ﬂight time. This is borne out by the
correlation (r= 0.75) between the two, as shuttlecock ﬂight
trajectories vary between 0.2 seconds for a short range smash
and 1.5 seconds for a defensive clearance from one end of the
court to the other or a high lob. Thus, the ﬁnal average result
for average ﬂight trajectories is close to one second, because of
the higher incidence of clearances and lobs.
Badminton characteristics 65
Actions during play
Because there are no studies on these variables, it is difﬁcult to
discuss their importance. However, the results showing the
relation between winning and losing a set and the higher or
lower number of unforced errors and/or winning shots—that
is, between these ratios and the ﬁnal result—are of great
interest because there is no known case of a player winning a
set with a signiﬁcantly higher number of unforced errors than
his/her opponent. In this study, there were six matches in
which in 76.9% of the cases the player with the fewer unforced
errors won the set, and in only 23.1% of the cases did the
player with the most unforced errors win. However, the
apparent contradiction implied by these data is not signiﬁcant
in terms of statistical analysis. With respect to maximum
actions, the negative correlation found between the number of
maximum actions and the average and maximum heart rate (r
=−0.59, p<0.05 and r=−0.62, p<0.03 respectively) can be
explained because the higher the heart rate the lower the
capacity to perform at maximum intensity and therefore the
maximum actions are fewer, because the decrease in the level
of response does not allow them to be carried out.
From these results, we can conclude that badminton is based
on fast movements, with a great demand on the alactic
anaerobic system and, to a lesser degree, on the lactic anaero-
bic metabolism. The high frequency and intensity of play
throughout a match, together with the high maximum and
minimum average heart rates, indicate that badminton is a
sport that, at competition level, demands a high percentage of
individual aerobic power and that high levels of aerobic power
allow players to maintain this type of effort during a total time
of about 30 minutes. Coaches should therefore base training
on a large number of competitive actions of high intensity but
short duration. Moreover, they should train speciﬁc endurance
by means of actions and moves performed at short (15–20
seconds) and very short (6–10 seconds) intervals.
Certain factors, such as the number of unforced errors,
seem to affect the ﬁnal result and could therefore be used to
predict the outcome of the match and a player’s performance
We wish to acknowledge the following people who assisted in the
translation and revision of this article: Tony Harris, Francisco Javier
Morales Robles, Inmaculada Roldán Miranda, Inmaculada Sanz
Sainz, and Gerald Smith.
D Cabello Manrique, Faculty of Education, University of Granada,
J J González-Badillo, Higher Education Olympic Centre
1Galiano D, Escoda J, Pruna R. Aspectos fisiológicos del Tenis.
2Christimass MA, Richmond SE, Cable NTT,
. A metabolic
characterisation of single tennis. In: Reilly T, Hughes M, Lees A, ed.
Science and rackets sports
. London: E & FN Spon, 1994.
3Sanchis J, González JC, López JA,
. Propuesta de un modelo de
entrenamiento de squash a partir de parámetros obtenidos durante la
4Anderson GS, Rhodes EC. A review of blood lactate and ventilatory
methods of detecting transition thresholds.
5Mader RA, Neck N. A theory of the metabolic origin of anaerobic
Int J Sports Med
6Stainsby WN. Biomechanical and physiological bases for lactate
Med Sci Sports Exerc
7Wassermann W. Mechanisms and patterns of blood lactate increase
during exercise in man.
Med Sci Sports Exerc
8Carlson J, Tyrrell J, Naughton G,
. Physiological responses during
badminton games by elite Australian players.
9Alvero JR. Valoración telemétrica de la frecuencia cardíaca. Estudio de
campo en jóvenes deportistas de élite de bádminton.
III Jornadas de
Iniciación al Bádminton
10 Hughes MG. Physiological demands of training in elite badminton
players. In: Reilly T, Hughes M, Lees A, eds.
Science and rackets sports
London: E & FN Spon, 1995.
11 Abe K, Haga S, Nakatani T,
. The work intensity of a badminton
match in Japanese top male players.
Bulletin of Institute of Health and
Sports Sciences - University of Tsukuba
12 Ghosh AK, Goswami A, Ahuja A. Evaluation of a sports specific training
programme in badminton players.
Indian J Med Res
13 Ghosh AK, Mazumdar P, Goswami A,
. Heart rate and blood
lactate response in competitive badminton.
Annals of Sports Medicine
14 Cabello D, Tobar H, Puga E,
. Determinación del metabolismo
energético en bádminton.
Archivos de Medicina del Deporte
15 Coad D, Rasmussen B, Mikkelsen F. Physical demands of recreational
badminton. In: Terauds J, ed.
Science in racket sports
Academic Publishers Del Mar, 1979.
16 Docherty D. A comparison of heart rate responses in racquet games.
J Sports Med
17 Cabello D, Cruz JC, Padial P. Estudio de la frecuencia cardíaca y ácido
láctico en bádminton. In:
VIII Congreso Europeo de Medicina del
Deporte, Granada, October, 1995
18 Dias R, Ghosh AK. Physiological evaluation of specific training in
badminton. In: Reilly T, Hughes M, Lees A, eds.
Science and racket
. London: E & FN Spon, 1995.
19 Bosquet L, Mercier D, Léger L. Validité de l’analyseur de lactate portatif
Science & Sports
20 Fell JW, Rayfield JM, Gulbin JP,
. Evaluation of the Accusport Lactate
Int J Sports Med
21 Chin M-K, Wong AS, So RCH,
. Sport specific fitness testing of elite
Br J Sports Med
Take home message
To improve badminton results, it is necessary to plan train-
ing according to the characteristics of the sport—that is, to
work on specific endurance and highly intensive competi-
66 Cabello, González-Badillo