💻 computer science

Structural Pass Analysis in Football: Learning Pass Archetypes and Tactical Impact from Spatio-Temporal Tracking Data

This paper introduces a structural framework for analyzing football passes using spatio-temporal tracking data to derive metrics that quantify defensive disruption, enabling the identification of distinct pass archetypes and the evaluation of their tactical impact on territorial progression.

Original authors: Oktay Karakuş, Hasan Arkadaş

Published 2026-04-01

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Oktay Karakuş, Hasan Arkadaş

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine watching a football match. Usually, when we talk about a "good pass," we think about the result: Did it lead to a goal? Did it result in an assist? Or did the team keep possession?

This paper asks a different question: What does the pass do to the other team's defense, even if it doesn't immediately lead to a shot?

Think of the defending team not as a group of individuals, but as a living, breathing net or a rubber sheet stretched across the field. When a team attacks, they are trying to poke holes in that net, stretch it until it snaps, or pull it so far to one side that a huge gap opens up.

Here is a simple breakdown of how the authors analyzed this, using everyday analogies:

1. The Problem with Old Ways

Previous methods were like judging a chef only by whether the customer ate the meal. If the chef made a beautiful, complex sauce that didn't get eaten, it was considered "bad."
In football, if a pass didn't lead to a shot, it was often seen as low-value. But the authors realized that some passes are like pulling a rug out from under someone. They might not knock the person down immediately, but they make them stumble, lose their balance, and leave them vulnerable for the next move.

2. The New Tool: "Structural Impact"

The authors created a new way to measure passes based on how they mess up the opponent's "rubber sheet" (the defense). They invented three specific rulers to measure this:

The "Line-Bypass" Score (Cutting the Cake):
Imagine the defense is a cake with layers of frosting (defensive lines). A "Line-Bypass" pass is like a knife slicing straight through all the layers to get to the bottom. It counts how many defenders the ball jumped over.
The "Space Gain" Metric (Finding the Empty Seat):
Imagine the defense is a crowded room. If you pass the ball to a spot where there are no people, you've gained "space." This metric measures if the ball moved from a crowded, dangerous area to a wide-open, safe area.
The "Structural Disruption" Index (The Jenga Tower):
Imagine the defense is a Jenga tower. A good pass doesn't just move the ball; it makes the whole tower wobble. This measures how much the pass forces the defenders to scramble, stretch, or shift their positions, making the whole structure unstable.

3. The "Tactical Impact Value" (TIV)

They combined these three rulers into one big score called Tactical Impact Value (TIV).

Low TIV: A "circulatory" pass. Like passing a ball back and forth in the corner of a room. It keeps the ball safe, but it doesn't really change anything about the opponent's shape.
High TIV: A "destabilizing" or "line-breaking" pass. This is the pass that makes the defense panic, stretch out, or leave a gap.

4. The Four Types of Passes (The Archetypes)

By looking at thousands of passes from the 2022 World Cup, they found that passes naturally fall into four "personality types":

The Circulator: The "Safe Keeper." Keeps the ball moving but doesn't take risks. Good for resting, bad for attacking.
The Destabilizer: The "Tease." It doesn't break the line, but it makes the defenders shift slightly, creating a tiny imbalance.
The Line-Breaker: The "Spear." It punches straight through the defense, bypassing multiple players.
The Space-Expander: The "Door Opener." It moves the ball into a huge, empty area that didn't exist a second ago.

5. What Did They Discover?

The Best Passes Aren't Always the Ones That Score: The passes with the highest TIV (the ones that messed up the defense the most) were the ones most likely to get the ball into the "Final Third" (the dangerous area near the goal). They are the setup for the goal, even if they aren't the goal itself.
The Unsung Heroes: When they looked at who made the most "High TIV" passes, they weren't just finding the star strikers. They found Center Backs (defenders). Players like John Stones (England) and Rúben Dias (Portugal) were the master architects who started the chaos from deep in their own half.
Team Personalities: Just like people, teams have different "passing styles."
- Some teams (like Argentina) were great at Line-Breaking (slicing through the middle).
- Others (like Morocco) were great at Space-Expanding (using the wide wings to stretch the defense).
The Goalkeeper Connection: Surprisingly, some of the most impactful passes were actually to the goalkeeper! Passing back to the keeper (like Süle to Neuer in Germany) might look like "giving up," but it often forced the opponent's defense to shift, creating a new gap for the next move. It's like a chess player moving a piece back to reset the board for a better attack.

The Big Takeaway

This paper teaches us that in football, chaos is good. A pass that doesn't score a goal but makes the opponent's defense look like a tangled knot is actually a huge success.

By measuring how a pass "breaks" the opponent's structure, we can finally see the true value of the players who build the attack, not just the ones who finish it. It's like realizing that the person who cracks the safe open is just as important as the person who grabs the money inside.

1. Problem Statement

Traditional football analytics heavily rely on outcome-based metrics (e.g., Expected Goals, Expected Threat, VAEP) to evaluate player actions. While effective at quantifying the probability of scoring, these models often fail to capture the structural impact of a pass on the opponent's defensive organization.

The Gap: Many passes are tactically valuable not because they immediately lead to a shot, but because they reshape the defensive shape, create imbalances, or open passing lanes for subsequent actions. Existing frameworks rarely quantify how a pass alters the spatial configuration of defenders.
Objective: To develop a framework that explicitly models the interaction between a pass and the defensive structure, quantifying how passes deform, penetrate, or manipulate the opponent's organization.

2. Methodology

The authors propose a structural framework utilizing synchronized spatio-temporal tracking data (29.97 Hz) and event data from the 2022 FIFA World Cup (64 matches, 41,078 successful open-play passes).

A. Structural Metrics

The core of the methodology involves three complementary metrics that quantify the effect of a pass ( $p_i$ ) on the set of defending players ( $D_i$ ):

Line Bypass Score (LBS):
- Definition: Counts the number of defenders positioned vertically between the passer and the receiver who are effectively bypassed by the pass.
- Mechanism: Measures vertical penetration and the ability to break defensive lines.
- Formula: $LBS(p_i) = \sum_{j=1}^{n} b_j$ , where $b_j=1$ if the defender's y-coordinate is between the passer and receiver.
Space Gain Metric (SGM):
- Definition: Calculates the difference in available attacking space between the receiver's location ( $x_r$ ) and the passer's location ( $x_s$ ).
- Mechanism: Uses a density function based on the proximity of defenders to determine if the pass moves the ball into a region of lower defensive pressure.
- Formula: $SGM(p_i) = S(x_r) - S(x_s)$ , where $S(x) = 1 / \rho(x)$ and $\rho(x)$ is the defensive density.
Structural Disruption Index (SDI):
- Definition: Measures the change in the distance between the ball and the defensive centroid ( $c$ ) before and after the pass.
- Mechanism: Quantifies the degree to which the pass stretches or deforms the defensive compactness.
- Formula: $SDI(p_i) = \|x_r - c\| - \|x_s - c\|$ .

B. Tactical Impact Value (TIV)

To create a unified measure, the three metrics are standardized (z-score) and combined into a weighted composite score:
$TIV(p_i) = w_1 \tilde{LBS} + w_2 \tilde{SGM} + w_3 \tilde{SDI}$
In the study, equal weights ( $w_1=w_2=w_3=1/3$ ) are used, assuming equal contribution from each structural mechanism.

C. Unsupervised Clustering

The normalized structural metrics serve as feature vectors for K-means clustering ( $K=4$ ). This unsupervised approach discovers interpretable pass archetypes without pre-defined labels:

Circulatory: Low structural impact; maintains possession.
Destabilising: Introduces positional imbalances; high SDI.
Line-breaking: Penetrates defensive lines; high LBS.
Space-expanding: Moves ball into open space; high SGM.

3. Key Contributions

Structural Framework: Introduced a novel framework that shifts focus from outcome (scoring probability) to process (defensive reorganization), explicitly modeling the interaction between passes and defensive structures.
Taxonomy of Pass Archetypes: Derived an interpretable classification of passes (Circulatory, Destabilising, Line-breaking, Space-expanding) directly from data using unsupervised learning.
Tactical Impact Value (TIV): Proposed a composite metric that quantifies the structural influence of a pass, demonstrating its strong correlation with territorial progression.
Multi-Level Analysis: Applied the framework across three levels:
- Action Level: Identifying specific pass types.
- Team Level: Mapping distinct tactical styles and spatial progression patterns.
- Player Level: Characterizing individual roles (e.g., build-up defenders) and identifying high-impact passer-receiver partnerships.

4. Key Results

Archetype Distribution: Circulatory passes are the most frequent (34.4%), followed by Destabilising (31.3%). Line-breaking (18.3%) and Space-expanding (16.0%) are less frequent but carry higher structural weight.
Correlation with Progression: There is a strong monotonic relationship between TIV and territorial progression.
- Passes in the highest TIV quantile have a >12% probability of entering the final third (vs. ~4.5% for low TIV).
- However, the correlation with immediate finishing (shots/goals) is weaker, suggesting TIV primarily facilitates the creation of dangerous situations rather than the final execution.
Team Styles: The framework revealed distinct tactical fingerprints. For example:
- Argentina/France: High structural activity in central and wide advanced zones.
- Morocco: Asymmetric, relying on wide corridors.
- Spain/Japan: Strong structural impact in wide attacking zones despite possession dominance.
Player Roles: The analysis identified that central defenders (e.g., John Stones, Nicolás Otamendi, Rúben Dias) are the primary drivers of structural progression, often initiating line-breaking passes.
Passer-Receiver Dynamics: The study highlighted that specific partnerships amplify structural impact. Notably, passes from defenders to goalkeepers (e.g., Süle to Neuer) showed high $\Delta TIV$ (increase in impact), as these "reset" passes reorganize the attack and stretch the opponent's defense, creating new lanes for subsequent progression.

5. Significance and Implications

Complementary Perspective: The framework complements existing Expected Threat (xT) and VAEP models by providing a "structural" lens. It explains why a pass is valuable even if it doesn't immediately increase scoring probability.
Tactical Interpretability: By mapping passes to archetypes, the model bridges the gap between raw data and coaching concepts (e.g., "breaking lines," "stretching the block").
Practical Applications:
- Scouting: Identifying players who excel at structural progression (e.g., ball-playing center-backs).
- Opposition Analysis: Visualizing how specific teams manipulate space and where they are vulnerable to structural disruption.
- Performance Evaluation: Assessing the quality of build-up play beyond simple possession retention.
Future Directions: The authors suggest integrating sequential models to track structural advantages over full possession sequences and using Graph Neural Networks (GNNs) to learn structural representations directly from raw tracking data.

In conclusion, this paper demonstrates that modeling the structural interaction between attackers and defenders reveals interpretable tactical patterns that outcome-based metrics miss, offering a robust new tool for football analytics.