Jour nal of Sports Sciences, 1997, 15, 137 -149

Dynam ic-system analysis of opponent relationships in

collective actions in soccer
JE A N -F R A N C IS G R E H A IG N E , 1 * DA N IE L B O U T H IE R 2 and
B E R N A R D D AV I D 2
Institut U niversitaire de For m ation des M a tres, U niversite de Franche-C om te, Fort Gr iffon, 25042 B esan on
C edex and 2 CEDA PS, U niversit e Paris-Sud, 91405 Orsay Cedex, France
1

Accepted 4 August 1996

The aim of this study was to exam ine the contribution of the systemic approach to the analysis of play in team
sports. We rst focus on the theory of dynamical systems and consider the interactions between the main
variables of the different components of systems and subsystems in soccer. In team sports, these variables
represent uctuating conditions, which momentarily constrain the organization of action for the players. Thus
changes in the momentary con guration of the game have to be exam ined in the light of previous
con gurations, the outline of the defensive strategy and the tactical choices involved. To study this problem, we
analyse the antecedents of goals in soccer. A procedure is proposed which analyses transitions between
con gurations of play, thus allowing time to be taken into consideration when studying the evolution of a
match. To illustrate the use and bene t of the analytic procedure, two goals are described in terms of dynamic
con gurations of play and opportunity of choices made by attackers.
K eywords : Con guration of play, goals, soccer, space, systemic approach, tactics.

Introduction
T here has been growing interest recently in m atch
analysis of football. Generally, notational analysis uses
num erical data to study and assess the quality of a
m atch. H owever, there is a dearth of published
research with regard to the theoretical bases of the
analysis of the tactical aspects of the gam e. To that end,
we examined the contribution of a dynam ic system s
app roach to the analysis of play in team sports and
especially in soccer. T he system s perspective has been
developed as a reaction to the reductionist app roach
that dom inated science for centuries.
The m ain challenge in team sports is that, in an opposition relationship (Deleplace, 1979), a team m ust
coordinate its actions to recapture, conserve and m ove
the ball so as to bring it within the scoring zone and to
score a goal. Brackenbridge (1979; cited in Thorpe et
al., 1986) suggested the following de nition for the
gam e: `a struggle for a territorial dom inance within a
set of rules which includes signi cant strategic and
*Author to whom all correspondence should be addressed.
0264 -0414/97

1997 E. & F.N. Spon

technical aspects and in which coincidence anticipation

is param ount. The struggle for territorial dom inance is
decided by a system of scoring w hich sym bolises the
extent of victory. The code of rules identi es the problem and ensures that both teams or individuals m eet on
an equal basis . Gr e haigne (1991) has indicated `that
this way of view ing team sports brings in three m ain
categories of problems :
1. Problems related to space and time. In an attack, one
m ust nd solutions to problem s of individual and collective handling of the ball to overtake, use or avoid
varying mobile obstacles. In defence, one must bring
forward obstacles to slow down or stop the ball and
one s opponents to allow eventual recapture of the
ball.
2. Problems related to information. Players m ust also
deal with problem s related to the production of uncertainty for opponents and of certainty for their partners
in conditions w hich rem ain fundam entally reversible.
Uncertainty and certainty are related to the quantity
and quality of available inform ation. U ncertainty is the
inform ation that we do not possess about the state of

138
the system (e.g. Atlan, 1979). T he reduction of uncertainty for the team in possession of the ball is a function
of the quality of the comm unication codes and the
choice of explicit tactics, thus allowing app ropriate
choices, understood by all partners, according to
m omentary con gurations of play.
3. Problems related to orga nization. Players m ust
accept the m ove from an individual to a collective project. A player m ust truly m erge the collective project
with his personal actions while giving the best of him self to the group.
To understand better the principles at work in soccer, one approach would be to m odel the interactions
between the players and their environm ent as a complex system . The system s perspective attem pts to
reduce the phenom ena involved in interactions between
the m ain variab les of the different com ponents in order
to study them. In team sports, these variab les represent
uctuating conditions w hich mom entarily constrain the
organ ization of action for the players. (e.g. Walliser,
1977; Bouthier, 1988; Gr e haigne, 1988; Ali and
Farraly, 1990; D avids et al., 1994; Gr e haigne and
G odbout, 1995). According to the space available, the
choices a player in possession of the ball has should be
related to the success of an attack. To im prove our
understanding of such choices, we rst examine the
system ic nature of soccer. Then we present a descriptive research procedure which makes it possible to
study transition between con gurations of play.

The systemic nature of soccer

Theor y of dynam ical systems
System analysis has been developed in the last 30 years
as a result of the interaction between various disciplines, including biology, information theory, cybernetics and the theory of system s. According to Atlan
(1979), it should not be considered a science, a theor y
or a discipline, but rather a new process that allows for
the gathering and organ ization of knowledge with a
view to m ore ef cient action. The system ic app roach
aim s to address three concerns (Walliser, 1977):
1. The need to return (as a reaction to ultra-analytic
tendencies of som e sciences) to a m ore synthetic
approach, which would recognize properties of
dynam ic interactions between elements of a whole,
giving it a `totality character.
2. To be able to conceive and control large and complex wholes, the need to develop a m ethod that
would m ake it possib le to sum m on and organ ize
knowledge so as to relate m eans and objectives
better.

Grehaigne et al.
3. T he need, in the face of fragm entation and dispersal
of knowledge (analytical approach), to prom ote a
utilitarian language that could support the articulation and integration of theoretical models and m ethodological precepts scattered am ong various
disciplines.
General properties of systems. The analytical approach
tries to break a system down into its m ost sim ple constituent elements. T hen, by m odifying one variable at a
time, it tries to deduce general law s which m ake it possible to predict the properties of the system in different
conditions. For such predictions to be feasible, additive
laws of elementary properties m ust com e into play.
However, in the case of high ly com plex systems (such
as a soccer m atch), these additive law s do not work.
Therefore, such system s m ust be approached using
new m ethods, such as those that are part of the systemic approach. Studying a system s behavio ur over
time leads to the determination of action rules, w hich
are used to in uence or m odify the state of the system.
To achieve this, the system ic app roach relies on the
notion of a system or a w hole m ade up of interacting
elem ents:
1. A w hole that has a reciprocal relationship with its
environm ent; such a relationship will provide it w ith
som e autonomy.
2. A whole composed of interacting subsystems; such
interdependence will ensure a degree of coherence.
3. A whole submitted to m ore or less im portant modi cations over time while m aintaining a certain
perm anence.
O ften, in a classical approach, an exp lanation of phenom ena will rely on linear causality: it is an exp lanator y
m ode based on a logical chain of causes and effects.
W ith the system ic approach, m ovem ent replaces perm anence, exibility replaces in exibility and adaptability replaces stability. N otions of ow and ow balance
join those of forces and force balance. By integrating
time, the systemic approach reveals the interdependence of phenom ena and their gradual change. Causality
becom es circular and a regulation loop (see Bertalanffy,
1972; Rosnay, 1975; M orin, 1986; C averni et al.,
1988).
Two main categories of system are de ned: closed
system s and open system s. A closed system exchanges
neither energy nor matter with its environm ent; it is
self-su f cient. O n the other hand, an open system
relates constantly w ith its environment. It exchanges
energy, m atter and inform ation useful for m aintaining
its organization. Its com plexity takes into account variety and interaction between elements. Som e liaisons

139

Dynamic con guration of play

m ay be studied from a causal point of view (balance,
stability, etc.) or from an end-product point of view
(adaptation, learning).
Systems and subsystems. A system is said to be quasidecom posable if it can be decom posed into quasiisolated subsystem s, with som e interaction between
them and with the environm ent (Walliser, 1977). W ith
reference to a given system , one m ay consider:
1. M icrosystems: obtained by retaining only a few subsystem s with all their interactions (e.g. confrontation between two force lines at a given time).
2. Infrasystems: obtained by retaining only a few subsystem s with som e of their interactions (1 vs 1 or 2
vs 2 at som e point in the match).
As w ith the general characteristics of system s, interactions between subsystem s are energy-b ased or inform ation-base d. T hese subsystem s m ay organize
themselves into various types of networks either superim posed upon or m erged inside the system . Each subsystem can generally be in turn decom posed into other
subsystem s according to an interlocking order which
re ects hierarchies or m ultiple-level sets of com binations. Relationships between elements of a given level
differ, in term s of nature and intensity, from those relationships between elements of subsystem s pertaining to
different levels (one m ay not switch from ve-a-sid e
soccer to 11-a-sid e m ode without reorganizing ones
knowledge and ones capacities). T his is so even though
relatively hom ogeneous subsystem s m ay som etimes
occur (e.g. three-player con gurations of play at the
peripher y of the pitch between a defender, a m id eld
player and a winger).
Systems and time. The temporal dim ension is im portant when studying system s because it is the m edium
through which they operate and evolve. All things considered, nothing m ay be fundam entally understood
about soccer if one does not shift from a spatial to a
temporal reference system while processing inform ation. The synchronous properties of a system relate to
the relationships between its various characteristics at a
given time. T he diachronic properties relate to the relationships of those sam e characteristics through m any
successive m om ents. T hey m ake it possible to bring to
light the system s evolutionary trends. In a quasidecom posable system, one may differentiate between
m odi cations m ainly related to its structure, its functioning, its evolution, or to relationships between these
three phenom ena. The system s structure, in a strict
sense, rests upon the whole set of its most unvarying
characteristics; thus the system s structure ensures its
very existence and its perm anence. In a larger sense,
the structure is form ed by all the system s character-

istics at a given time, thus re ecting the state of the

system at that m oment. The system s functioning
relates on the one hand to each subsystem s transform ations and, on the other, to ow s passin g through
linking channels between subsystem s and between the
system and its environment. The system s evolution is
brough t about, on the one hand, by a change in the
subsystem s transfor m ation law s and, on the other, by
changes in the way the system organ izes itself into subsystem s and changes in the linking channels between
the system and its environm ent. For instance, in soccer,
the structural dimension is characterized by:
1. A boundary which establishes the frontiers of the
system (the stadium and playing area).
2. Elements w hich m ay be counted and grouped into
categories. For exam ple, the players (attackers or
defenders), the ball, etc.
3. `C ontainers in w hich energy or inform ation is
stocked (line forces, the goalkeeper, players energy
potential).
4. A com m unication network that allows energy and
information exchanges (the rules, the code of play, a
com m on fram e of reference to read and interpret
plays in the sam e way, etc.).
The functional dim ension is characterized by:
1. F lows of energy, information, or elements that m ove
about (players, the ball, replacements, state of
fatigu e).
2. G ates controlling the rate of various ows (the play
leader, players m om entary tactical choices, the referee, etc.).
3. D elays resulting from ows m oving at different
speeds, or from the gate response time (creating
open space, gaining an interval, restoring a defensive block, etc.).
4. Regulation loops, either proactive or retroactive,
w hich play a large part in the system s behaviour by
m anaging all param eters (taking inform ation to
adjust the gam e plan, setting a defensive reserve,
m odifying the system of play).
Regulation: General principle. The information process
uses an inform ation-collecting m echanism and m akes
use of inform ation to m odify the system . To characterize it, one m ay consider ve types of activity w hich
intervene, in a cyclic way, in its functioning (see M alho,
1974):
c Infor mation activity: translating into a conceptual
form observed, real phenomena (perception).
c Prospecting activity: constructing probable, possible
or desirable schem es about the future (planning).

140

Grehaigne et al.

c Decision activity: translating intents and aspirations

into actions on reality (program m ing and m anagem ent).
c Execution activity: transfo rm ing the system through
voluntary and coordinated actions (execution).
c Control activity: collecting inform ation about the
results of actions w ith a view to pursuing or transform ing the current action or the upcoming action
(regulation).
Soccer: C ontr ibution of system analysis

Figu re 1 Concepts related to the notion of opposition.

To obtain m ore information about the structure and

the functioning of play, we use a system ic approach to
discuss the m odelling of team sports.
In a soccer match, structures and con gurations of
play should be considered as a whole, rather than
exam ined piece by piece. System s with many dynam ically interacting elem ents are capable of rich and varied patterns of behavio ur, which are clearly different
from the behavio ur of each com ponent considered separately. T he in uence of general system s theory is now
clearly evident and one m ust analyse the performances
of the players as a system in synergy w ith the environm ent.
Indeed, in a match, the opposition creates the unexpected, necessitating constant adaptation to constraints
brought about by the confrontation. A m atch rarely
relies upon the sim ple application of schem a of play
learned previously during training. Thus during the
gam e, one can foresee only probab ilities of evolution
for the attack and defence con gurations, hence the
im portance of heuristics to quickly solve the problem
inherent in speci c interactions between two teams.
In a classic learning approach, one tries before anything else to teach students technical skills and to m aintain order on the playing eld by, for exam ple, use of
form al groupings. H owever, it could be argued that it is
as important, and m aybe even m ore so, to get the players to m anage disorder optim ally (Villepreux, 1987;
G r e haigne, 1989, 1992a). T his type of app roach, w hich
advocates `opposition and `disorder m anagement as
the basis for any progress, brings to bear new concepts
which appear fundam ental for a renewal of team sport
teaching. Figure 1 identi es som e concepts which
com e into focus when one points out opposition as a
fundam ental elem ent of the m odelling process in team
sports.
The central notion of opposition leads us to consider
the two team s as interacting organized systems. The
structural characteristics of these system s consist of a
prog ram m e that can be m odi ed according to acquired
exp erience; their m ain functional property is learning.
T he operational conditions of such system s in team
sports require the m anagem ent of disorder, while pre-

serving a certain order and thus allowing decisions to

be m ade in an environment that can be com pletely
foreseen a priori. For instance, let us analyse the organ izational level `match - the confrontation between two
team s - by looking at its structural and functional characteristics. The structural aspects are the spatial organ ization of the constituent elem ents of the system
(synchronic and topological properties), whereas the
functional aspects refer to the various time-related processes, such as exchanges, regulations and reorganizaton of the elements (diachronic and kinetic
properties).
For the level `match , structurally the elements of the
system are represented by the two opposing team s and
the com m unication network between the two is de ned
by the rules of the sport. At this level, the idea is to
characterize from a space standpoint the opposition
rapp ort, and to analyse the relationships between the
strong points of the attack system on the one hand and
those of the defensive system on the other. Notions at
stake here include `in block , `in pursuit , centre of
gravity, circulation of the ball, etc. (e.g. W inkler, 1984;
Gr e haigne, 1992b; Bouthier et al., 1994; Gr e haigne et
al., 1994). Functionally, one is dealing with the evolution in time of the opposing relationship between the
two teams (advance, delay, breaking, continuity, etc.).
In this case, each m atch provides a phenomenal datum
- that is, som ething original and unique - thus reducing the ef ciency of ready-m ade m otor or strategic
solutions.
A m atch constitutes a complex system . Analysis of
part of the gam e reveals a large num ber of interacting
variab les. On the eld, a non-h om ogeneous distribution of the players results in a non-h om ogeneous distribution of the energy state of the players. A certain
kind of hom ogeneous scattering characterizes the equilibrium state towards which soccer system s always
evolve. It corresponds, therefore, to a hom ogeneity of
the players distribution on the different energetic
states. The degree of hom ogeneity of the con gurations
of play can also be explain ed by a distribution of the

141

Dynamic con guration of play

probabilities of the presence of the players in certain
areas of the pitch.
Another way to show this consists in de ning the
m icro-states of the system of attack and defence. Each
m icro-state is de ned by a distribution of the players on
the pitchs playing area according to their positions,
orientations and speeds. O ne can call this kind of distribution a `dynam ical con guration of play . T he
app arent disorder indicates a hom ogeneity m ore general than the sim ple spatial distribution of the players
on the ground because the latter distribution in uences
only two energetical levels: potential and speed. In
those cases of opposition, the kinetic interactions lead
to the stabilization of the spatially non-hom ogeneous
states by other non-h om ogeneous distributions which
app ear hom ogeneous if one looks for certain energetic
states. It m eans that those states would seem to be
m ore hom ogeneous for an observer w ho is able to recognize the different kinetic states. C onversely, a classic
observation would stress the heterogeneous aspects by
dealing only with positions and geom etric shape s. T hat
is how, we suggest, the dialectic equilibrium and disequilibrium of the gam e operates. O n the one hand,
very stable structures m ake one think of a crystalline
structure, seen as being rigid and w ith few chances of
evolution, as for exam ple in set-plays. On the other
hand, the dynam ical con gurations of play have within
them a num ber of transfor m ations, lim ited according
to the different possibilities of the continuous evolution
of the gam e, but nevertheless im portant if one chooses
a break in m odifying the m ovem ent in process. This
group of elem ents form s a foundation for our analytical
fram ework. D ynam ical system s are able to attain multiple patterns of stability in achieving a state of coordination with the opposition. A theoretical fram ework that
focuses on the achievem ent of coordination within a
dynam ical system m ay be appropriate for the study of
con gurations of play.

approach in the precise determ ination of the different

areas led us to search for a new form of analysis.
Pur pose and under lying assumptions
D uring the gam e, a unit of play evolves from a state 1
to a state 2 and so on to a state n. As in a photograph,
the con guration of play is de ned by the positions of
the players at a m om ent M . T hrough analysis of the
tactical choice of the different ball-holders in a collective movem ent before a goal, one m ay understand better how goals are scored (Gr e haigne and Bouthier,
1994). In short, the aim is to describe the `dynam ic
states of the players w ho participate in the attack and
those of their opponents. To that end, we devised a
num ber of diagram s representing the few seconds preceding a goal. For each player, one can note (a) the
player s position, (b) the direction of m ovem ent and (c)
the speed of m ovem ent. T hese param eters de ne the
potential turning angle and the am ount of ground that
can be covered. To represent those kinetic data in a
plane, we propose the notions of a `sector of play for
the attackers and a `sector of intervention for the
defenders. T hese sectors spatially de ne the lim its of
possible actions for the different players, w ithin 1 s,
when considering the three variab les m entioned above.
These three variab les and the param eters discussed
below were selected based on the following assum ption: a collective offensive m ovem ent leads to a goal
when the ball-holders respect the offensive param eters
im posed by the state of play and, in particular, pass the
ball into an open space. W hat are these param eters?
Based on eld experience, we believe that a num ber
of param eters m ust be respected in a dynamical configuration of play for a goal to be scored. Figure 2 shows
three cases. First, for a ball to be intercepted ( < 2 m ),
there m ust not be any sector of intervention w hich can

A procedure to analyse transitio ns between

con gurations of play
D uring a soccer m atch, the player w ith the ball is faced
with the problem of having to make choices. In a prelim inar y approach to this problem , we tried to m odel
these aspects of the gam e by considering the possib ilities that exist for an attacker to release the ball. From
the notion of a hypo thetical sphere of operations (the
area of play in attack which is composed of free space +
interacting space + surfaces hidden by the defenders),
we tried to determ ine the available area for the attackers (m ovem ent + available space) to obtain the probabilities of passes. D if culties w ith the previous

Figu re 2 Con gurations where exchange of the ball is into a

free sector. The shapes of play and sectors of intervention are
in accordance with the players speeds. The trajectory of the
ball is a function of the height of the ball.

Grehaigne et al.

142

F igure 3 Con gurations where defenders do not have suf cient time to intercept the trajectory of the ball.
Figu re 6 Example of a grid overlay, each square representing
2 m 2 on the ground.

act as a block between two sectors of play where the

players exchange the ball. Second, the ball m ust not
cross a sector of intervention. T hird, the player in possession must be ahead of the defender and be in a sector that is free or nearly free. Figure 3 shows that, if an
interceptable ball crosses (entering and going out of) a
sector of intervention, the time between the ball being

kicked and its clearing the sector m ust be less than the
time taken by the defender to cut the trajectory of the
ball. It m ust arrive in a free sector of play or a partially
free one in the case of a pass. Figure 4 shows that an
accurately played ball will leave a free or partly free
sector of play and reach another sector of play, also free
or partially free. In the case of a successful shot on
target, only the strikers sector of play need be free or
partly free. Figure 5 shows that w hen a player decides
to dribble the ball, that sector of play m ust be free or
partly free. T he difference in speed is also im portant.
Tools used for analysis
To analyse con gurations of play, we used a video
recorder to note the positions and speeds of the players,
together with the types of m oves. Since the recording
gave 25 fram es per second, it was easy to construct an
accurate representation of the action. D ifferent lines on
the ground were used to estim ate the position of the

F igure 4 Con gurations where the sectors of play of the

attackers are partly free.

F igure 5 Con gurations where players can dribble the

ball.

Figu re 7 Experim ental set-up to calculate angles according

to the players speeds.

143

Dynamic con guration of play

Table 1 Turning angles according to the speed of the players

Stop
M ean
Range
S .D .

Walk

180
180-180
0

120
105-140
9.5

players. These positions were then m arked on an

index card depicting a standardized half pitch. International m atch pitches are 112 70 m . The data collection diag ram represents proportionally half a pitch
and is covered w ith a pattern of squares (Fig. 6). T he
squares, each representing 2 m 2 on the ground, allow
for greater accuracy when plotting the players positions (for m ore precise location, we used a 1 m 2 pattern). W hen the pitch sizes were different, the grid was
adjusted to the dim ensions set out by FIFA in its World
C up presentation book.
The video recorder was used to obtain the positions,
the m ovem ents and the speeds of the players. Second
by second, the positions taken by players involved in
attack and defence during the pre-go al phase were
noted. Successful attacks which resulted in a goal generated 4 -10 topographic diagram s of the precise spots
occupied by the players and the ball. For each diag ram ,
the speed calculated was the average speed. By using
the diagram with the grid, the distance covered by a
player could be determ ined. Sim ilarly, the video
recorder can be used to obtain the time interval
between two con gurations of play. It is then possib le
to estimate the average speed of a player.
Searching for a procedure that would m ake it possible to pattern the potential sphere of operation of a
given player and draw the players sector of play, we
tested 20 physical education students registered on a
soccer course at the U niversity of Burgundy. Each
player ran, without a ball, from m arker A to m arker B
(see Fig. 7). Upon reaching B, he was asked to turn to

Jog
51
40 -60
6.7

Cruise
42
30-60
6.9

Sprint
21
10-30
6.9

his right as sharply as possible. This task was carried

out at a speed that would be found in a soccer m atch
(Reilly and Thom as, 1976; Lacour 1983; D ufour,
1989). T he speed was m onitored by photoelectric eyes
at marker B and the references used were: stop, 0 m s-1 ;
walk, 0 -2 m s -1 ; jog , 2 -4 m s -1 ; cruise, 4 -6 m s -1 ;
spr int, > 6 m s -1 . Table 1 shows the average of the
angles for the 20 players according to their speeds.
D uring the test, the players diverted their run to one
side only. Assum ing sim ilar angles for a turn to the
other side, the following angles for the sector of play
were obtained by doubling the results of our test: stop,
360; walk, 240; jog, 100; cruise, 80; spr int, 40.
Knowing the speed of a player - and consequently the
distance he can cover in 1 s - the depth of his potential
sph ere of action can be determ ined (Table 2) and his

Table 2 The potential sphere of action

according to the speed of the players

Speed
(m s -1 )

>

0
1
2
3
4
5
6
7
8

Angle
()

Depth of
angle (cm)

360
240
240
100
100
80
80
40
40

0.35
0.7
1.0
1.4
1.7
2.1
2.45
2.8

Figu re 8 Sectors of play or intervention according to

speed.

Grehaigne et al.

144

M - 5 s and M - 4 s (Fig. 9c). PE chose to dribble

past his opponent on a one-to-one basis. C onsidering
the m oves open to defender d2, PE had som e room to
m anoeuvre. H is sector of play was partly free, since he
was using a body screen. Again the fourth param eter
was respected.
M - 3 s (Fig. 9d). PE now had just two possibilities
open to him : m ake an accurate pass to C A, who would
be m ade to feel very uncomfortable by defenders d4,
d5 and d6, or to pass to PSA, who was close to being
offside in a free sector of play. PE chose the latter
option, thus respecting the third param eter (Fig. 4).
M - 2 s (Fig. 9e). PA received the ball on his head,
redirecting it in the direction of his m ovem ent. The
only possibility open to him was to keep the ball and go
for the Swedish goal. A pass to CA was not possible
because of the attention of defenders d5 and d6.
D efender d4 was also in front of CA. T he choice m ade
by PA respected the fourth param eter.
M - 1 s (Fig. 9f). The choice m ade by PA was a consequence of CAs position in Fig. 9e. Being ahead of his
closest opponent, PA had just one possibility open to
him : keep the ball and head for the target. This was a
good decision, since his sector of play was partially free
(he was ahead of defender d5). O nce again , the fourth
param eter is respected.

sector of play or sector of intervention can be drawn

(Fig. 8).
An illustration
To illustrate, we will analyse Jean-Pierre Papins goal,
scored after 59 m in, in the France vs Sweden m atch at
the European Cham pionships in Sweden in 1992.
M - 8 s and M - 7 s (8 and 7 seconds before the goal
was scored) (Fig. 9a). PE recovered the ball on the lefthand side of the French half. He had m any options
open to him : m ake a lateral pass, a set back pass to AM ,
or keep the ball. H e chose the third option, thus
respecting the fourth param eter: his sector of play was
partly free despite the intervention of defender d3. H e
still had the option of m oving to the left or right. H e
chose to dribble past defender d3 on the right and was
successfu l in doing so. H e had just the necessar y space
to dribble the ball, thus respecting the fourth param eter
(Fig. 5).
M - 6 s (Fig. 9b). PE has now dribbled past defender
d3. His sector is now free and he has a lot of space in
which to m anoeuvre. H e has three options: keep the
ball, m ake a low pass between defender d1 and d2 to
C A, or ight the ball over defenders d1 and d2 to CA.
H e decided to keep the ball, once again respecting the
fourth param eter.

F igure 9a Jean-Pierre Papins goal for France vs Sweden: M

-8

s and M

-7

s.

145

Dynamic con guration of play

F igure 9b Jean-Pierre Papins goal for France vs Sweden: M

-6

s.

F igure 9c Jean-Pierre Papins goal for France vs Sweden: M

-5

s and M

-4

s.

Grehaigne et al.

146

F igure 9d Jean-Pierre Papins goal for France vs Sweden: M

-3

s.

F igure 9e Jean-Pierre Papins goal for France vs Sweden: M

-2

s.

147

Dynamic con guration of play

F igure 9f Jean-Pierre Papins goal for France vs Sweden: M

-1

F igure 9g Jean-Pierre Papins goal for France vs Sweden: M

- 0.

s.

Grehaigne et al.

148
M - 0 (Fig. 9g). PA, who was ahead of defender d5,
decided to shoot at goal. T he trajectory of the ball
crossed the sector of play of the Swedish goalkeeper,
G K. However, the time elapsing between the striking of
the ball and the ball entering the target was less than
0.5 s. Consequently, the last defender could not change
the trajectory of the ball because the time he needed to
cut the trajectory inside his sector of play was longer
than the time between the ball leaving PAs boot and it
crossing his sector (thus respecting the second param eter; F ig. 3). As a consequence, a goal was scored.

D iscussion
T he analysis of collective m ovements show s that, generally, different holders of the ball respect the param eters of success. C hoices are m ade based on position,
m ovem ent and the speed of ones team m ates and
opponents. A prelim inar y analysis of 110 dynam ical
con gurations of play showed that 102 respected the
principle of passin g the ball into an open space. This
principle can be seen as a constant of the system
(G r e haign e and Bouthier, 1994).
N evertheless, at the end of the collective m ovem ents
to which they belonged, eight goals were scored,
although they did not respect this principle. U pon
closer analysis of those eight con gurations, holders of
the ball m ade bad choices (no respect for the param eters). T he defender should have intercepted the ball,
but because of his clum siness the ball was intercepted
by the attacker.
For example, during the Sweden vs D enm ark gam e,
the collective m ovem ent w hich led to a goal in the 58th
m inute showed a m istake at M - 1 (Fig. 10). A1 had
elected to pass the ball to A2 (M - 2). T he ball was
interceptable and reached defender d1. D efender d1

F igure 10 From M

-2

s to M

-1

s and then M

- 0:

tried to get the ball, but failed and sent the ball into
A2s sector of play. A2, with a location advantage over
d2, seized the opportunity and scored.
T he contexts in which actions occur in soccer are so
varied that, in most instances, the concepts of `principle
of play and `constant of the system are useful to
understand how players treat inform ation. O ur results
show that the position and the m ovem ent of players are
reliable and faithfu l param eters for the analysis of play.
M odels of the sectors of play and sectors of intervention can be used to pattern attacks and obtain m ore
inform ation about the functioning of the system . In
addition to its reliability, the analysis shows that a goal
has to be constructed and the tactical choices of the
players optim ized. A goal can only be scored if the principles of play are m et; if they are not, the only way to
score is by virtue of a m istake on the part of the
defenders.
T his type of analysis can be considered as a starting
point because, from a m ethodological point of view, it
is a rst step to obtaining tools which allow sim ulation
of m ovem ent in soccer and the production of a m odel
with good predictive power. M oreover, this m odel
m aterializes and objecti es abstract param eters, such
as the assessm ent of relative speed and orientation of
ones opponents.
T he organizational level `match or `system m atch (a
group of players playing together as a w hole) is an
interesting way of analysing soccer. W ith the opposition
relationship, order and disorder can emerge from the
play at any m om ent. In this way, the energy and
choices of the players serve to create the conditions for
transitions between con gurations of play and thus
transfor m the play. However, in soccer, the different
states of the system operate fundamentally according
to states of equilibrium . If a goal is scored, it m eans
that the organ izational level `defence has not respon-

Sweden vs Denmark.

Dynamic con guration of play

ded correctly to the existing conditions of play im posed
by the opposition attack. T he use of such a m odel of
the analysis of team sports seem s to provide a viab le
and pertinent basis for exp laining the system capabilities underlying persistence and change during a
m atch.

Acknowledgements
The authors wish to thank Paul Godbout and Lew Hardy
for their help, suggestions and feedback throughout this
project.

References
Ali, A.H. and Farraly, M. (1990). An analysis of patterns of
play in soccer. Science and Football , 3 , 37 -44.
Atlan, H. (1979). E ntre le Cr istal et la Fum e e . Paris: Seuil.
Bertalanffy, L.V. (1972). The or ie G e n e rale des Syst`e mes . Paris:
Dunod.
Bouthier, D. (1988). Les conditions cognitives de la formation d actions sportives collectives. Unpublished thesis,
Universite Paris V, EPH E.
Bouthier, D., Barthes, D., David, B. and Grehaigne, J.F.
(1994). Tactical analysis of play combinations in rugby
with video-computer: Rationalising `French- air . Communication to the Second World C ong ress of Notation al
Analysis , Cardiff, November.
Caverni, J.P., Bastien, C., M endelsohn, P. and Tiberghien, G.
(1988). Psychologie Cognitive: M od e` les et M e th odes . Grenoble: Presses Universitaires.
Davids, K., Handford, C. and Williams, M . (1994). The natural alternative to cognitive theories of motor behaviour:
An invitation for interdisciplinary research in sport science? Jour nal of Sports Science s, 12 , 495 -528.
Deleplace, R. (1979). Rugby de M ouvem ent-R ugby Total . Paris:
EPS.
Dufour, W. (1989). Les techniques d observation du comportement moteur. E ducation Physique et Sport , 217 ,
68 -71.

149
Grehaigne, J.F. (1988). Game systems in soccer. In Science
and Football (edited by T. Reilly, A. Lees, K. Davids and
W.J. Murphy), pp. 316 -321. London: E. and F.N.
Spon.
Grehaigne, J.F. (1989). Football de mouvement: Vers une
approche system ique du jeu. Unpublished thesis, U niversite de Bourgogne.
Grehaigne, J.F. (1991). A new method of goal analysis. Science and Football , 5 , 10 -16.
Grehaigne, J.F. (1992a). L organisation du Jeu en Football.
Paris: ACTIO.
Grehaigne, J.F. (1992b). A weighted model to analyse the
conditions of scoring in soccer. Comm unication to the
First World C ong ress of Notationa l A nalysis , Liverpool,
November.
Grehaigne, J.F. and Bouthier, D. (1994). Analyse des e volutions entre deux con gurations du jeu en football. Science
et M otricite , 24 , 44 -52.
Grehaigne, J.F. and Godbout, P. (1995). Tactical knowledge
in team sports from a constructivist and cognitivist perspective. Q uest , 47 , 490 -505.
Grehaigne, J.F., Bouthier, D. and David, B. (1994). The players action zone in soccer. Com munication to the Second
World C ong ress of Nota tional Analysis , Cardiff, November.
Lacour, J.R. (1983). Aspects physiologiques du football.
L Entra neur Fran ais , 183 , 1 -6.
Malho, F. (1974). L acte Tactique en Jeu . Paris: Vigot.
Morin, E. (1986). La C onna issance de la Connaissance . Paris:
Seuil.
Reilly, T. and Thomas, V. (1976). A motion analysis of workrate in different positional roles in professional football
m atch play. Jour na l of H um an M ovem ent Studies , 2 ,
87 -97.
Rosnay, J. de (1975). Le M acroscope . Paris: Seuil.
Thorpe, R., Bunker, D. and Almond, L. (1986). R ethink ing
G am es Teaching . Loughborough: Loughborough University of Technology, Department of Physical Education,
Sport Science and Recreation M anagement.
Villepreux, P. (1987). Rugby de M ouvem ent et D isponibilite du
Joueur . Paris: M e m oire IN SEP.
Walliser, B. (1977). Syst`e mes et M od`e les: Introdu ction C ritique a`
l Analyse de Syst`e m es . Paris: Seuil.
Winkler, W. (1984). Sport Strategie als Lehrfach. Leitungss port , 2 , 5 -13.