Territory Progression: why the work before the final pass matters
A football possession can be valuable long before it becomes a shot. This is the case for Territory Progression: a diagnostic built to catch the players who move the game from safe areas into pressure, danger and better attacking states.
The easiest football actions to remember are the ones closest to the finish. The goal. The assist. The shot. The final ball. They matter, obviously, but they are also the part of the possession that already survived everything else. Before that final action, somebody had to move the ball away from pressure, cross a line, enter the final third, or force the defending team to protect a different space.
That middle part is where a lot of football analysis becomes lazy. We say a player “progresses play” or “gets the team up the pitch”, but then the evidence often collapses back into assists, chances created, touches, pass completion or possession share. Those are useful numbers, but they do not quite answer the question I wanted this metric to ask: who repeatedly changes where the opponent has to defend next?
Territory Progression is my first diagnostic for that problem. It is not trying to replace xG, xA or shot-based models. It is aimed at the possessions that improve the next football problem before a shot ever appears.
Territory is not the same thing as possession
Possession usually tells us who had the ball and, sometimes, how securely they kept it. Territory asks a different question: where did the ball move, and did that movement make the defending team protect a more dangerous part of the pitch?
A team can have a long possession that never changes the opponent's problem. Centre-back to full-back, back to centre-back, across the goalkeeper, safe recycle, repeat. That can be useful game control, but it is not automatically attacking value. The opposite can also happen: one carry through midfield, one vertical pass into the half-space, or one switch that opens the weak side may create territorial stress in two seconds.
This distinction matters for player comparison. If two midfielders both complete ninety percent of their passes, the pass completion number hides the question we actually care about. Did those passes move the game into better territory, or did they simply maintain the same possession state?
Basketball gives analysts dense endings.
Most basketball possessions end in a shot, turnover or free throw. If a player improves possessions, the feedback comes quickly. Plus-minus and possession-value models get a lot of repeated endings to learn from.
Football hides value in the middle.
Football is different. Many good possessions never become shots. They end in a recycle, clearance, foul, blocked lane or reset after the player has already changed the field position and pressure.
The shot bias problem
Shot-based metrics are strongest near the end of a move. That is exactly why xG and xA are so valuable. They describe the part of the possession where the link to scoring is clearest. The weakness is also obvious: football contains far more useful actions than shots, and most possessions do not reach a shot at all.
This is where football differs from basketball in a way that matters statistically. Basketball possessions usually produce a recorded ending. Football possessions often produce an intermediate advantage that disappears from the final log. A midfielder receives under pressure and turns out. A full-back carries thirty metres and pins the winger back. A forward drops into a pocket and lays the ball into the final third. If the next action is blocked or the cross is cleared, a pure shot lens can miss almost all of that work.
So the metric should not ask only, “did this possession become a shot?” It should also ask, “did this player move the possession into a zone where better things become possible?”
The metric
Territory Progression per 90 combines four event-data signals. Forward territory gain measures how much useful ground a player's actions move. Final-third entries capture actions that take possession into the attacking third. Box-zone entries capture actions ending in the central high-value lane around the penalty area. Successful involvement keeps the model grounded in actions that actually stayed alive.
The score is intentionally readable. It rewards forward movement, then gives extra weight to entries into more dangerous zones. A successful action matters because failed progression can still be interesting, but it should not be treated the same as value that lets the team keep attacking.
The important bit is not the single number by itself. The number is the doorway. The split underneath is the explanation. Antoine Griezmann topping this World Cup 2022 progression sample is not the same story as Luka Modrić ranking second, and that difference is exactly why the ingredient view matters.
Why use the 2022 World Cup sample?
The World Cup is not a perfect club-football sample. Roles are compressed, team styles are less rehearsed, and a few matches can swing perception. That is exactly why it is useful for a case study. The tournament gives us a clean event-data slice where the best players are asked to solve different problems in a short, high-pressure environment.
For this article I filtered to players with at least 450 minutes. That leaves twenty players in the view below. It removes cameo noise while still keeping enough role variety to make the metric interesting: creators, tempo midfielders, centre-backs, wing-backs, forwards and even one goalkeeper whose distribution profile shows up strongly.
A. Griezmann
Griezmann leads the sample because his value is not just volume. He combines final-third access with a tournament-high box-zone profile: 9.89 box-zone entries per 90. This is the player who keeps appearing in the part of the pitch where the next action can hurt you.
- TV p90
- 32.45
- Box p90
- 9.89
- Success
- 77.9%
L. Modrić
Modrić is almost level with Griezmann, but the route is different. His 521.9 territory gain per 90 and 9.47 final-third entries per 90 describe a midfielder constantly moving Croatia into the next zone rather than simply showing up for the last pass.
- TV p90
- 32.24
- Gain p90
- 521.9
- F3 p90
- 9.47
D. Upamecano
Upamecano is the useful sanity check. He is not rating as a final-third creator. He rates because France repeatedly used him to advance possession securely: 661.5 territory gain per 90, 9.75 final-third entries per 90, and 86.6% success.
- TV p90
- 20.75
- Gain p90
- 661.5
- Success
- 86.6%
What the 2022 sample shows
The top of the table immediately shows why the diagnostic is useful. Griezmann, Modrić, Messi, De Paul and Ziyech all score highly, but not for the same reason. Griezmann and Messi show strong box-zone value. Modrić and De Paul are heavier territory and final-third movers. Mbappé and Gakpo show how direct danger can compensate for lower involvement volume.
This is also why the model should be read as a profile, not a trophy. A defender such as Upamecano can rate well through secure progression and final-third entries while offering very little box-zone work. That does not make him a creator in the same way Griezmann is a creator. It tells us his tournament possessions repeatedly changed field position.
The Croatia names tell a similar story. Modrić, Juranović, Kovačić, Perišić, Brozović and Gvardiol all appear in the top twenty. That is not an accident. Croatia's tournament was built on keeping the ball alive and moving it through pressure. Territory Progression catches that structure because it does not only care whether the move ended in a shot.
Argentina's midfield shows a different pattern. Messi is still Messi, but De Paul and Enzo Fernández give the possession platform around him. De Paul rates through relentless involvement and final-third movement. Enzo rates through secure central progression. The number does not replace watching the match; it tells you where to look when you rewatch it.
Top 20 table
Same progression diagnostic output as Impact Lab: World Cup 2022, minimum 450 minutes.
| Rank | Player | Team | Pos | Min | Index p90 | Gain p90 | F3 p90 | Box p90 | Success |
|---|---|---|---|---|---|---|---|---|---|
| 1 | A. Griezmann | France | Attacking Midfielder | 537 | 32.45 | 359.6 | 6.70 | 9.89 | 77.9% |
| 2 | L. Modrić | Croatia | Midfielder | 656 | 32.24 | 521.9 | 9.47 | 7.13 | 79.8% |
| 3 | L. Messi | Argentina | Striker | 690 | 27.22 | 361.8 | 6.65 | 7.30 | 76.5% |
| 4 | R. De Paul | Argentina | Midfielder | 602 | 26.38 | 468.9 | 8.67 | 4.93 | 81.3% |
| 5 | H. Ziyech | Morocco | Striker | 638 | 25.28 | 326.2 | 5.64 | 7.19 | 66.7% |
| 6 | J. Juranović | Croatia | Defender | 600 | 24.56 | 429.1 | 6.30 | 5.85 | 78.3% |
| 7 | E. Fernández | Argentina | Midfielder | 563 | 23.68 | 446.3 | 9.27 | 3.36 | 82.1% |
| 8 | K. Mbappé | France | Attacking Midfielder | 597 | 22.76 | 181.6 | 3.32 | 8.14 | 69.8% |
| 9 | C. Gakpo | Netherlands | Striker | 453 | 22.02 | 189.1 | 2.38 | 8.34 | 65.9% |
| 10 | M. Kovačić | Croatia | Midfielder | 638 | 21.19 | 301.1 | 8.89 | 3.10 | 84.9% |
| 11 | D. Upamecano | France | Defender | 480 | 20.75 | 661.5 | 9.75 | 0.56 | 86.6% |
| 12 | A. Rabiot | France | Defensive Midfielder | 483 | 20.01 | 339.4 | 8.20 | 2.80 | 81.0% |
| 13 | A. Tchouaméni | France | Defensive Midfielder | 623 | 19.81 | 430.5 | 8.96 | 1.73 | 86.2% |
| 14 | I. Perišić | Croatia | Striker | 670 | 19.76 | 255.7 | 3.22 | 6.31 | 68.8% |
| 15 | D. Dumfries | Netherlands | Midfielder | 480 | 19.41 | 281.3 | 5.06 | 4.88 | 66.3% |
| 16 | T. Hernández | France | Defender | 508 | 19.33 | 379.8 | 5.85 | 3.72 | 81.0% |
| 17 | N. Aké | Netherlands | Defender | 479 | 17.43 | 533.7 | 7.52 | 0.94 | 85.5% |
| 18 | M. Brozović | Croatia | Midfielder | 554 | 16.95 | 414.3 | 7.31 | 1.30 | 84.8% |
| 19 | J. Gvardiol | Croatia | Defender | 690 | 16.73 | 588.0 | 6.52 | 0.91 | 86.6% |
| 20 | H. Lloris | France | Goalkeeper | 570 | 16.39 | 985.6 | 4.74 | 0.16 | 67.3% |
How to read the number
Start with the rank, then immediately read sideways. If the player is high on the index because of box-zone entries, you are looking at danger access. If the player is high because of territory gain and final-third entries, you are probably looking at build-up progression. If the success rate is low, the value may be more volatile. If actions per 90 is very high, the player may be carrying possession structure as much as producing decisive moments.
This is why I do not want the metric presented as a FIFA-style rating. A single score makes the page easy to scan, but the real analysis starts when the score is broken into ingredients. Griezmann, Modrić and Upamecano can all be valuable territory players while doing completely different jobs.
What it adds to player comparison
On a player profile, this progression diagnostic should sit beside xG, xA, shot creation, carrying, passing and defensive work. It gives the comparison page a way to talk about a player before the final action. For a casual fan, it answers a simple question: who gets the team into better areas? For an analyst, it opens the next set of questions: how, from where, how often, and with what security?
That is the real product use case. The site should not just say “good passer” or “creative player”. It should be able to say: this player changes territory through central carries, this one through early final-third entries, this one through secure defender progression, and this one through direct box-zone access.
That is a better way to compare footballers because football roles are not equal. A winger, an eight, a full-back and a centre-back can all move the team forward, but they do it from different starting points and against different defensive shapes. The metric is not the whole answer. It is a cleaner first question.
The point
The diagnostic is useful because it respects the part of football that is easy to miss: the action before the action. It rewards players who make the next pass, carry or decision happen in a better place. That does not always show up as xG. It does not always become an assist. Sometimes it just turns a harmless possession into a problem the opponent has to solve.