01 Sep When Training Matches Reality: Modelling the Space Elite Teams Actually Have to Attack
Reduced-space training is one of the most common tools used in elite football.
Coaches reduce pitch dimensions to increase interaction density, accelerate decisions, constrain passing options and expose players to situations where time and space are limited.
But there is a more interesting question:
What if reduced space is not only created in training? What if it is already one of the problems that competition repeatedly creates?
Data from the full 2025/26 LALIGA EA SPORTS season offer a particularly clear example.
When opponents face FC Barcelona and Real Madrid and defend in their own half, they adopt the two smallest positional structures in the competition when width and depth are considered together.
At the same time, Real Madrid and FC Barcelona also occupy the first two positions for defensive-line height during the finalisation phase.
The opponent contracts.
The attacking team pushes the game higher.
And football is played inside an increasingly compressed spatial environment.
But the rankings only identify the problem.
The real analytical value comes from modelling how that space changes, frame by frame, and understanding how an attacking team can manipulate it.
The competitive problem: opponents become smaller
The first analysis looks specifically at the positional behaviour of the opponents when they are in their own half.
For each team, we have the average opponent width and depth accumulated across the full season.
Combining both dimensions provides a simple rectangular representation:
Opponent positional area = Width × Depth
This is not intended to represent the exact surface occupied by the team. It is not a convex hull. It is a comparative geometric proxy that allows us to identify how wide and how deep the opponent structure tends to be.
The ranking is revealing.
Opponent positional structure in own half
Lower values = smaller opponent structure
| Rank | Team faced | Opponent width | Opponent depth | Area |
|---|---|---|---|---|
| 1 | FC Barcelona | 39.24 m | 27.07 m | 1,062 m² |
| 2 | Real Madrid | 40.81 m | 28.00 m | 1,143 m² |
| 3 | Elche CF | 40.90 m | 30.24 m | 1,237 m² |
| 4 | Atlético de Madrid | 42.00 m | 29.98 m | 1,259 m² |
| 5 | Real Sociedad | 41.95 m | 30.37 m | 1,274 m² |
| 6 | Girona FC | 42.51 m | 30.37 m | 1,291 m² |
| 7 | Athletic Club | 42.85 m | 30.55 m | 1,309 m² |
| 8 | Deportivo Alavés | 42.27 m | 31.00 m | 1,310 m² |
| 9 | RCD Mallorca | 42.67 m | 30.78 m | 1,313 m² |
| 10 | RC Celta | 43.33 m | 30.36 m | 1,315 m² |
| 11 | CA Osasuna | 42.24 m | 31.18 m | 1,317 m² |
| 12 | Valencia CF | 42.60 m | 30.92 m | 1,317 m² |
| 13 | Real Betis | 43.53 m | 30.67 m | 1,335 m² |
| 14 | Rayo Vallecano | 41.64 m | 32.11 m | 1,337 m² |
| 15 | Getafe CF | 43.12 m | 31.20 m | 1,345 m² |
| 16 | Sevilla FC | 42.74 m | 31.89 m | 1,363 m² |
| 17 | Levante UD | 43.82 m | 31.46 m | 1,379 m² |
| 18 | Villarreal CF | 44.60 m | 31.04 m | 1,384 m² |
| 19 | Real Oviedo | 43.80 m | 31.77 m | 1,392 m² |
| 20 | RCD Espanyol | 43.84 m | 32.08 m | 1,406 m² |
The Barcelona and Real Madrid values are particularly interesting because the result is not driven by only one dimension.
Against Barcelona, opponents show both the lowest width and lowest depth.
Against Real Madrid, they show the second-lowest values in both dimensions.
Their opponents therefore do not simply become shorter or narrower.
They become shorter and narrower simultaneously.
That behaviour is compatible with a tactical response familiar to any coach facing a highly dominant attacking team: retreat, protect central areas, reduce distances between units and defend closer to goal.
The data do not establish whether that behaviour is voluntarily selected by the opponent, imposed by Barcelona and Madrid through territorial dominance, or —most likely— produced by an interaction between both mechanisms.
But they clearly identify the spatial problem.
The other side of the problem: pushing the game upwards
Opponent compactness alone does not determine how compressed the game becomes.
The attacking team’s own structure matters too.
During the finalisation phase, Real Madrid and FC Barcelona have the two highest defensive lines in the competition.
Defensive-line height during the finalisation phase
Own-team values; higher = more advanced defensive line
| Rank | Team | Defensive-line height |
|---|---|---|
| 1 | Real Madrid | 62.10 m |
| 2 | FC Barcelona | 61.15 m |
| 3 | Girona FC | 59.39 m |
| 4 | Valencia CF | 59.27 m |
| 5 | Elche CF | 59.19 m |
| 6 | Villarreal CF | 58.92 m |
| 7 | Atlético de Madrid | 58.88 m |
| 8 | RC Celta | 58.26 m |
| 9 | Real Sociedad | 58.12 m |
| 10 | Athletic Club | 58.11 m |
| 11 | RCD Mallorca | 58.00 m |
| 12 | RCD Espanyol | 57.88 m |
| 13 | Getafe CF | 57.61 m |
| 14 | CA Osasuna | 57.57 m |
| 15 | Real Oviedo | 57.44 m |
| 16 | Deportivo Alavés | 57.17 m |
| 17 | Real Betis | 57.13 m |
| 18 | Levante UD | 56.39 m |
| 19 | Rayo Vallecano | 56.02 m |
| 20 | Sevilla FC | 55.89 m |
Now both findings can be interpreted together.
The opponent reduces its own structure close to goal.
Barcelona and Real Madrid simultaneously position their defensive line exceptionally high.
The space in which possession is sustained therefore becomes compressed from both directions.
This is the competitive context.
But knowing that this happens is only the first step.
From ranking the problem to modelling the solution
Season averages can tell us which teams encounter the smallest opponent structures.
They cannot tell us exactly how those structures change during an attack.
For that we need a dynamic model.
The model used in this analysis operates at 25 Hz: 25 observations every second.
At each instant it calculates:
- the width of both teams;
- the depth of both teams;
- defensive-line height;
- the spatial relationship between the attacking and defending structures;
- and the intersection area in which both structures interact.
The result is not a static rectangle describing an entire match.
It is a moving representation of how space is continuously created, reduced and redistributed during the action.
That distinction is fundamental.
A low defensive block is not permanently compact in exactly the same way.
It breathes.
It moves laterally.
It stretches when the ball travels.
Its lines separate and reconnect.
One player jumps to press.
Another covers.
The weak side contracts.
The attacking team changes orientation.
And, for a few frames, a previously unavailable space may appear.
Those moments are precisely what an attacking model should help identify.
What can the model tell a coach?
Imagine an attacking sequence against a compact low block.
At one moment, the opposition may occupy 40 metres of width and 27 metres of depth.
Those numbers describe its dimensions.
But for coaching purposes, the more important questions are different:
What caused those dimensions to change?
Did moving the ball to one side increase the distance between defenders?
Did maximum attacking width force the weak-side full-back to move away from a central position?
Did a player dropping between lines increase the opponent’s depth?
Did an inside run force the back line towards goal?
Did switching the point of attack temporarily enlarge the central corridor?
And, critically:
How long was the resulting space available before the block closed again?
A 25 Hz model allows those questions to be studied temporally rather than only descriptively.
Instead of saying:
The opponent defended compactly.
we can begin asking:
How was the opponent’s compactness broken?
That is much closer to the language of coaching.
The intersection area: where the game actually has to be solved
The video accompanying this analysis shows an FC Barcelona attacking sequence.
Both teams’ width and depth are represented dynamically.
Their spatial structures overlap.
That intersection is especially useful because it approximates the zone in which attacking and defensive occupation compete directly for the same space.
For an attacking player, this is where many of the most difficult actions occur:
receiving between opponents;
playing with limited time;
turning under pressure;
finding a third-player connection;
breaking a line;
attacking an interval;
or recognising when circulation should continue instead of forcing penetration.
The value of the model is therefore not simply visual.
It provides a way of quantifying the spatial problem the player is being asked to solve.
From match analysis to training design
This is where the connection with reduced-space training becomes particularly relevant.
A coach can always create a 30 × 25 m possession game.
But why 30 × 25?
Why not 34 × 28?
Why should the pitch remain constant?
Should the task reproduce the dimensions of the opponent’s low block?
The interaction area?
The space available immediately before a successful line break?
The duration for which that space remains open?
Once competition can be described dynamically, training-space design no longer needs to be arbitrary.
The model opens the possibility of moving from generic reduced-space training towards opponent- and problem-specific spatial design.
For example, if the next opponent consistently defends with a very short and narrow structure in its own half, training could reproduce comparable spatial constraints.
If the analysis shows that its block tends to expand laterally after fast switches, the task can reward precisely that behaviour.
If central space appears only briefly after one defender jumps out, the exercise can reproduce not only the dimensions, but also the temporal window players need to recognise.
This changes the question from:
Should we train in reduced spaces?
to:
Which reduced spaces should we train in, and why?
Training the opponent you are going to face
The practical application could extend directly into match preparation.
Before facing an opponent, analysts and coaches could establish:
- its typical width and depth in a low block;
- how those dimensions change according to ball location;
- how frequently the block enters highly compressed states;
- which attacking behaviours increase its width;
- which movements increase separation between lines;
- how quickly it recovers compactness;
- and which spatial configurations precede successful penetrations.
Those outputs can then inform the training process.
Not by attempting to reproduce every detail of the match.
But by creating representative spatial problems.
This may be especially relevant for dominant teams.
Their competitive reality is often different from the league average because opponents adapt to them.
The ranking of Barcelona and Real Madrid illustrates precisely this point.
Their opponents do not occupy the same structures against them that might be expected from an average match environment.
The opponent changes because the opponent is also responding to who is in front of them.
Training design should be capable of acknowledging that.
A different way of thinking about reduced-space tasks
Reduced-space exercises are usually discussed in terms of their effect on players:
more touches;
more accelerations;
less decision time;
greater technical involvement;
higher interaction density.
All of those dimensions are useful.
But there is another perspective.
A reduced-space task can also represent a tactical environment.
For FC Barcelona or Real Madrid, attacking in a reduced spatial environment may not simply be something coaches artificially impose during training.
It may resemble what happens when an opponent acknowledges their attacking threat, retreats towards its own area, reduces width and depth, and forces them to solve football inside a compact block.
That is why the most interesting finding here is not that Barcelona is first or Madrid second in a ranking.
The rankings reveal the problem.
The model allows us to study how that problem can be solved.
From spatial description to spatial intelligence
Previous LALIGA research has already shown that collective spatial behaviour contains meaningful tactical information. Longitudinal analysis found that team width remained relatively stable across seasons while team length progressively decreased, suggesting greater longitudinal density in team structures.
The current approach moves one step further.
Instead of asking only:
How much space does a team occupy?
we can ask:
How does one team’s behaviour modify the space available to the other?
And then:
Which actions successfully transform that space?
That transition —from spatial description to spatial interaction— is where the model becomes useful for football practice.
Because the final objective of football analytics is not simply to measure what happened.
It is to help coaches and players understand what problem the game is presenting and what behaviours can solve it.
Practical takeaway
The 2025/26 season data identify FC Barcelona and Real Madrid as the two teams whose opponents occupy the smallest structures in their own half. They are also the two teams with the highest defensive lines during the finalisation phase.
Together, those findings describe a highly compressed attacking environment.
But season averages are only the entry point.
At 25 frames per second, the analytical model can reveal how width, depth, defensive-line height and the interaction area evolve throughout each attacking sequence.
That makes it possible to study not only how small the space is, but:
when it opens, why it opens, how long it remains available and what the attacking team does with it.
And that leads to a much more useful connection between analysis and training:
Do not simply train in small spaces. Train the spatial problem you are actually going to face.