Professional Documents
Culture Documents
net/publication/43136087
CITATIONS READS
113 2,231
4 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Albin Tenga on 04 June 2014.
To cite this Article Tenga, Albin, Holme, Ingar, Ronglan, Lars Tore and Bahr, Roald(2010) 'Effect of playing tactics on goal
scoring in Norwegian professional soccer', Journal of Sports Sciences, 28: 3, 237 — 244
To link to this Article: DOI: 10.1080/02640410903502774
URL: http://dx.doi.org/10.1080/02640410903502774
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Journal of Sports Sciences, February 1st 2010; 28(3): 237–244
ALBIN TENGA1, INGAR HOLME1,2, LARS TORE RONGLAN1, & ROALD BAHR1
1
Norwegian School of Sport Sciences, Oslo and 2Department of Preventive Cardiology, Ullevål University Hospital, Oslo,
Norway
Abstract
Methods that include an assessment of opponent interactions are thought to provide a more valid analysis of team match
performance. The purpose of this study was to examine the effect of playing tactics on goal scoring by assessing opponent
interactions in Norwegian elite soccer. The sample included 203 team possessions leading to goals (cases) and 1688 random
Downloaded By: [Tenga, Albin] At: 13:11 8 March 2010
team possessions (control group) from 163 of 182 (90%) matches played in the men’s professional league during the 2004
season. Multidimensional qualitative data using ten ordered categorical variables were obtained to characterize each team
possession. The proportion of goals scored during counterattacks (52%) was higher than during elaborate attacks (48%),
while for the control group the proportion using elaborate attacks (59%) was higher than when using counterattacks (41%)
(P ¼ 0.002). Multiple logistic regression analyses showed that, for the main variable ‘‘team possession type’’, counterattacks
were more effective than elaborate attacks when playing against an imbalanced defence (OR ¼ 1.64; 95% confidence interval:
1.03 to 2.61; P ¼ 0.038). Assessment of opponent interactions is critical to evaluate the effectiveness of offensive playing
tactics on the probability of scoring goals, and improves the validity of team match-performance analysis in soccer.
Keywords: Validity, opponent interaction, logistic regression, goal scoring in soccer, match-performance analysis
Correspondence: A. Tenga, Norwegian School of Sport Sciences, Po Box 4014, Ullevål Stadion, N-0806 Oslo, Norway. E-mail: albin.tenga@online.no
ISSN 0264-0414 print/ISSN 1466-447X online Ó 2010 Taylor & Francis
DOI: 10.1080/02640410903502774
238 A. Tenga et al.
(or players), rather than two opposing teams (or sion. To obtain a random sample of 3260 team
players) individually in isolation from the match possessions from which a control group was ex-
context. It is possible to analyse relationships tracted, we assigned each match a computer-gener-
between opponents when using a match-play situa- ated random decimal number between 0 and 1,
tion as the basic unit of analysis rather than a team which was multiplied by 86 to indicate the beginning
(or a player). Therefore, team possession was used as (in minutes) of a match period from which a total of
the unit of analysis in this study. Since such match- 20 consecutive team possessions would be extracted.
play situations emerge from the interplay of play and This was based on the assumption that 20 con-
counter-play produced by the two teams (Grehaigne, secutive team possessions lasts 6.5 min on average,
Bouthier, & Godbout, 1999; Grehaigne & Godbout, and that 2–3 min of extra time is added to each
1995), they allow a breakdown of a match-play match. The random sample obtained was then
action without losing its confrontational nature. analysed for team possession type, namely counter-
Moreover, the use of multidimensional qualitative attack, elaborate attack, and set-play attack. Finally,
data instead of unidimensional frequency data from the 3260 team possessions a total of 1688 team
improves our ability to describe soccer match-play possessions of the counterattack type (n ¼ 686, 41%)
(Grehaigne, Mahut, & Fernandez, 2001; Hughes & and elaborate attack type (n ¼ 1002, 59%) were
Bartlett, 2002; Suzuki & Nishijima, 2004). This is collected and used as a control group to compare
because multidimensional qualitative data permit the these playing tactics.
inclusion of data from the qualitative evaluation of
different dimensions of performance involved in the
Team match-performance analysis
Downloaded By: [Tenga, Albin] At: 13:11 8 March 2010
possession into consideration. For details of how the elaborate attack types were included. The 14 teams
analysis was undertaken as well as the procedures for examined scored a mean number of 15 goals
video editing and analysis, see Tenga et al. (2010). (range ¼ 8–21), and the controls included a mean
The study was approved by the Norwegian Social number of 121 team possessions (range ¼ 101–148).
Science Data Services (NSD). Differences were observed between proportions of
goals and controls for playing tactics. For the main
variable ‘‘team possession type’’, the proportion of
Statistical analysis
goals scored during counterattacks (52%) was higher
A total of 203 goals and 1688 controls from either than during elaborate attacks (48%), while for the
counterattacks or elaborate attacks were estimated to control group the proportion using elaborate attacks
be sufficient to detect a difference of 27% (D) (59%) was higher than when using counterattacks
between the proportions of the two playing tactics (41%). There were differences in the probability of
ending in the scoring of a goal, assuming an a of 0.05 goal scoring between the playing tactics for all
and a b of 0.10. This was calculated based on the variables including the main variable ‘‘team posses-
sample size formula n ¼ 2(s/D)2f(a, b), assuming a sion type’’ (Table I). For the three defensive zone
standard deviation (s) of 2.22 (Lachin, 1981) and variables combined (‘‘overall defensive score’’), only
given the available sample size of n ¼ 609 (203 goals 2.5% of the goals were scored against a balanced
and twice as many control team possessions, i.e. defence, compared with 31% of the control attacks.
406). The null hypothesis, that there would be no In contrast, 94% of the goals were scored against an
difference in effectiveness between the two main imbalanced defence, compared with 41% of controls
Downloaded By: [Tenga, Albin] At: 13:11 8 March 2010
Table I. Number of goals (n ¼ 203) and controls (n ¼ 1688) plus percentages of goals scored by playing tactics according to offensive and
defensive variables (N ¼ 1891).
Offensive variables
Team possession type
Counterattack (‘‘direct play’’) 792 (41.9) 106 686 13.4 0.002
Elaborate attack (‘‘possession play’’) 1099 (58.1) 97 1002 8.8
Starting zone 50.001
Final third 55 (2.9) 18 37 32.7
Middle third 860 (45.5) 101 759 11.7
First third 976 (51.6) 84 892 8.6
Pass number 50.001
Short possession 884 (47.9) 67 817 7.6
Medium possession 572 (31.0) 68 504 11.9
Long possession 388 (21.0) 56 332 14.4
Pass length 50.001
Long pass 193 (10.5) 5 188 2.6
Mixed 751 (40.7) 94 657 12.5
Short pass 899 (48.8) 92 807 10.2
Pass penetration 50.001
Penetrative pass 173 (9.4) 44 129 25.4
Downloaded By: [Tenga, Albin] At: 13:11 8 March 2010
Note: The variable ‘‘overall defensive score’’ reflects the combined probability scores of the three zone-defence variables.
*Pearson chi-square.
random control team possessions revealed differ- of offensive tactics, and improve the validity of team
ences in the probability of goal scoring between match-performance analysis.
offensive tactics only when playing against an It should be noted that this study has some
imbalanced defence. However, the probability results limitations, which must be taken into account when
for most offensive tactics when playing against a interpreting the results. It is a retrospective study,
balanced defence were not reported because their and the number of team possessions ending in goals
values could not be estimated precisely enough. For being scored when playing against a balanced
the main variable ‘‘team possession type’’, counter- defence was few. Therefore, we have not presented
attack was more effective than elaborate attack when probability results for all offensive tactics when
playing against an imbalanced defence. Thus, these playing against a balanced defence. Thus, while this
findings show that the assessment of opponent paper provides important data on the differences in
interactions is critical to evaluate the effectiveness playing styles on the main outcome variable ‘‘goals
Effect of soccer playing tactics on goal scoring 241
Table II. Odds ratios (OR) for goal scoring by the two playing tactics according to offensive variables.
Note: aThe odds ratio (OR) reflects the chance of goal scoring, compared with the reference category.
*Offensive tactics included in the model.
Table III. Odds ratios (OR) for goal scoring by the two playing tactics according to offensive variables when controlling for the effects of the
degree of defensive balance.
Note: aThe odds ratio (OR) reflects the chance of goal scoring, compared with the reference category.
#
Not reported because the odds ratio (OR) could not be estimated precisely enough due to few team possessions.
*Offensive tactics included in the model according to balanced and imbalanced defence subgroups.
242 A. Tenga et al.
scored’’, it also illustrates a limitation, in that goals interactions difficult. Despite this, previous studies
are scored on an infrequent basis. As shown in a generally support the current findings. Harris and
separate study (Tenga et al., 2010), using more Reilly (1988) showed that defence against attacks
frequently occurring end points such as score-box with a shot on target, compared with those without a
possessions can complement the case-control ap- shot, tended to involve higher attacker to defender
proach. Notably, with a cohort-like design and using ratios and greater average distances between the
‘‘score-box possession’’ as an intermediate outcome attacker in possession and the nearest defender
variable, we observed similar results to the present throughout the attack. According to Grehaigne
study. Those showed that counterattacks had a (1991), the overall attacking configuration with
higher probability of producing score-box posses- adequate space and time and opponent’s defence
sions than elaborate attacks when playing against an with its centre of gravity out of position had a positive
imbalanced defence but not against a balanced effect on the scoring of 10 of 33 goals. Elsewhere, it
defence. The case-control design used in the current was reported that the defending performances,
study is generally considered inferior to a cohort directly measured through distances and angles
design in terms of generalizability, but ‘‘goals between attackers and defenders and the number of
scored’’ is obviously the primary outcome of interest players, were significantly related to delaying and
in soccer, not ‘‘score-box possessions’’. Neverthe- diverting attacks, and covering attacking space
less, the two studies produce similar results irrespec- (Suzuki & Nishijima, 2004). Seabra and Dantas
tive of the design and outcome variable used. (2006) reported a higher proportion of successful
Also, use of the variables ‘‘defensive backup’’ and shooting attempts for ball receptions and shots
Downloaded By: [Tenga, Albin] At: 13:11 8 March 2010
‘‘defensive cover’’, which have only fair inter- originating from zones of low defensive confronta-
observer reproducibility (kappa coefficients of 0.24 tion than high defensive confrontation. Moreover,
and 0.27, respectively), represents a limitation. although indirectly, Olsen and Larsen (1997)
However, the combined variable ‘‘overall defensive showed more scoring opportunities and goals from
score’’, the variable used in the subsequent analysis, breakdown attacks (counterattacks) started when the
was still useful. In fact, these variables’ relatively opponent defence was imbalanced rather than
poor inter-observer reproducibility most probably balanced. Similarly, Jones et al. (2004), Bloomfield
stems from the observational limitations experienced et al. (2005), Lago and Martin (2007), and Taylor
when evaluating positions, distances, and angles et al. (2008) reported the influence of score-line
between dynamically interacting players by using status (winning, losing or drawing) and opposition
videotapes recorded from live TV broadcasts. In quality on ball possession.
addition, all the team possessions included were Whether ‘‘possession play’’ or ‘‘direct play’’ is
taken from the Norwegian men’s top professional more effective has long been disputed in the soccer
soccer league, which clearly represents a lower community, including match-performance research-
standard of play than the top leagues in Europe. ers (e.g. Bate, 1988; Hughes & Franks, 2005; Olsen
Consequently, the results obtained could be a & Larsen, 1997; Reep & Benjamin, 1968). For the
reflection of the playing standard or style in this purposes of this paper, the terms ‘‘elaborate attack’’
particular league. Even if the teams in the league and ‘‘counterattack’’ are considered to be synon-
varied in their playing styles and were included with a ymous with ‘‘possession play’’ and ‘‘direct play’’,
similar number of team possessions, care should be respectively. Our overall univariate and multivariate
taken when extrapolating these results to other analyses show that counterattacks were more effec-
soccer leagues or playing standards. tive than elaborate attacks for the scoring of goals.
This study also has strengths worthy of considera- Similar results were reported by Olsen and Larsen
tion. It is a case-controlled study using a large sample (1997), who found that breakdown attacks (counter-
size of controls randomly extracted from matches attacks) resulted in more scoring opportunities and
played in the Norwegian professional league. More- goals than longer attacks (elaborate attacks). Further
over, we used logistic regression analysis, the univariate and multivariate analyses reveal that
appropriate statistical method for comparisons of counterattacks were more effective than elaborate
categorical differences associated with binary re- attacks when playing against an imbalanced defence.
sponse variables (Nevill et al., 2002). The use of However, the multivariate analysis results for coun-
multidimensional qualitative evaluation allowed us in terattack versus elaborate attack when playing against
the current study to analyse different factors of match a balanced defence were not reported. That counter-
performance that usually are difficult to measure attacks are more effective than elaborate attacks
directly, as well as their interdependency. when playing against an imbalanced rather than a
The differences in study design and variable balanced defence should not be surprising, since the
types and their definitions make a direct comparison main objective of counterattacking is to exploit
between studies that have assessed opponent imbalances in the opponent’s defence to achieve
Effect of soccer playing tactics on goal scoring 243
penetration. But, it might also be that teams choose the variable ‘‘pass length’’, our analyses show that
to play directly precisely because they can take short passes were more effective than long passes in
advantage of imbalances in the opponent’s defence. univariate analysis but not in multivariate analysis. In
That elaborate attacks were found to be less effective contrast, for the variables ‘‘starting zone’’ and ‘‘pass
than counterattacks when playing against an im- penetration’’, our overall analysis shows that the final
balanced defence might also be due to differences third was more effective in goal scoring than first
in defensive balance that our variables failed to third and penetrative passes were more effective than
uncover. non-penetrative passes. That team possessions ori-
Similarly, for the variable ‘‘pass number’’, un- ginating from the final third rather than first third
adjusted overall univariate and multivariate analyses have considerably higher effectiveness for goal
show that long possessions were more effective than scoring was also reported by Bate (1988), Hughes
short possessions. However, further univariate and (1990), and Hughes and Snook (2006). Our
multivariate analyses show that long possessions were subgroup analyses show that the final third was more
more effective than short possessions when playing effective than the first third and penetrative passes
against an imbalanced defence, but not against a more effective than non-penetrative passes when
balanced defence. It would appear that a relatively playing against an imbalanced defence. However, the
high number of consecutive passes (five passes or univariate and multivariate analysis results for
more) is more effective in exploiting imbalances in penetrative passes versus non-penetrative passes
the opponent’s defence than in creating space by and multivariate analysis results for final third versus
dislocating defenders in a balanced defence. first third when playing against a balanced defence
Downloaded By: [Tenga, Albin] At: 13:11 8 March 2010
The current finding that long possessions (five are not reported.
passes or more) are more effective than short It is important to realize that goal scoring is often a
possessions (two passes or less) for scoring goals is result of a combination of factors, including technical
both supported (e.g. Hughes & Churchill, 2004; (e.g. passing precision), psychological (e.g. coping
Hughes & Franks, 2005; Hughes, Robertson, & with stress), physical (e.g. endurance), social (e.g.
Nicholson, 1988; Hughes & Snook, 2006) and cooperation), and tactical (e.g. exploitation of
contested (e.g. Bate, 1988; Hughes, 1990; Reep & imbalances in the opponent’s defence) (Burwitz,
Benjamin, 1968) in the literature. Using data from 1997). Nevertheless, the tactical approach employed
the study by Reep and Benjamin (1968), Hughes and is an important factor. One aspect that should be
Franks (2005) demonstrated that more goals were explored further is the sequential analysis of playing
indeed scored from shorter passing sequences, but tactics, especially when playing against a balanced
also that there were many more of the shorter passing defence. For example, analysing tactics as typically
sequences than the longer ones. Thus, consistent occur in a competitive match such as long passes and
with our results, longer passing sequences were space passes versus short passes and foot passes
considered to be more effective than shorter ones might uncover other effective combinations of
(Hughes & Churchill, 2004; Hughes & Franks, 2005; playing tactics (Seabra & Dantas, 2006). Another
Hughes & Snook, 2006). However, the present study aspect worthy of exploration is expanding the scope
also reveals that long possessions were more effective of analysis variables. It should be possible to apply
than short possessions when playing against an multidimensional qualitative evaluation in the ana-
imbalanced defence but not against a balanced lysis of off-the-ball movements involving one or more
defence. It is also apparent that using short possessions players. In so doing, effective offensive movement
and long possessions interchangeably with direct play tactics for creating and utilizing space according to
and possession play, as did Hughes and Franks (2005), the degree of defensive balance may be revealed.
might be inappropriate. This is because simply The current findings have some practical implica-
counting the number of passes excludes other essential tions. The information obtained about the relative
features in the analysis of these styles of attack (Franks, effectiveness of offensive playing tactics, especially
1988; Olsen, Larsen, & Semb, 1994). when playing against an imbalanced defence, can be
Furthermore, our overall and subgroup analyses used to improve a team’s goal-scoring and goal-
showed no difference in effectiveness between space preventing abilities effectively. This information can
passes and foot passes on the probability of goal be used when coaches and players plan and practise
scoring. However, these tactics were most effective in how to take advantage of an opponent’s choice of
goal scoring when used in combination, as indicated playing tactics in a competitive match.
by the results of the category ‘‘mixed’’ for ‘‘space
utilization’’ in Table I. Scoulding and colleagues
Conclusions
(Scoulding, James, & Taylor, 2004) also failed to
distinguish between space passes and foot passes This study shows that counterattacks were more
used by a successful and an unsuccessful team. For likely than elaborate attacks to lead to the scoring of a
244 A. Tenga et al.
goal when playing against an imbalanced defence. Hughes, M., & Franks, I. (2005). Analysis of passing sequences,
Thus, the assessment of opponent interactions is shots and goals in soccer. Journal of Sports Sciences, 23, 509–
514.
crucial to evaluate differences in the probability of Hughes, M., Robertson, K., & Nicholson, A. (1988). Comparison
goal scoring between different offensive playing of patterns of play of successful and unsuccessful teams in the
tactics, and hence improves the validity of team 1986 World Cup for soccer. In T. Reilly, A. Lees, K. Davids, &
match-performance analysis in soccer. W. J. Murphy (Eds.), Science and football (pp. 363–367).
London: E & FN Spon.
Hughes, M., & Snook, N. (2006). Effectiveness of attacking play
References in the 2004 European Championships. In H. Dancs,
M. Hughes, & P. G. O’Donoghue (Eds.), Proceedings of the
Bate, R. (1988). Football chance: Tactics and strategy. In World Congress of Performance Analysis of Sport 7 (pp. 46–62).
T. Reilly, A. Lees, K. Davids, & W. J. Murphy (Eds.), Science Cardiff: CPA UWIC Press.
and football (pp. 293–301). London: E & FN Spon. Jones, P. D., James, N., & Mellalieu, S. D. (2004). Possession as a
Bloomfield, J. R., Polman, R. C. J., & O’Donoghue, P. G. (2005). performance indicator in soccer. International Journal of
Effects of score-line on team strategies in FA premier league Performance Analysis in Sport, 4 (1), 98–102.
soccer. Journal of Sports Sciences, 23, 192. Lachin, J. M. (1981). Introduction to sample size determination
Burwitz, L. (1997). Developing and acquiring football skills. In and power analysis for clinical trials. Controlled Clinical Trials, 2,
T. Reilly, J. Bangsbo, & M. Hughes (Eds.), Science and football 93–113.
III (pp. 201–206). London: E & FN Spon. Lago, C., & Martin, R. (2007). Determinants of possession of the
Elias, N., & Dunning, E. (1966). Dynamics of group sports with ball in soccer. Journal of Sports Sciences, 25, 969–974.
special references to football. British Journal of Sociology, 17, Nevill, A. M., Atkinson, G., Hughes, M. D., & Cooper, S.-M.
388–402. (2002). Statistical methods for analysing discrete and catego-
Ensum, J., Pollard, R., & Taylor, S. (2004). Applications of rical data recorded in performance analysis. Journal of Sports
Downloaded By: [Tenga, Albin] At: 13:11 8 March 2010