Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering

Hasan Arkada\c{s}; Oktay Karaku\c{s}

REVIEW 3 major objections 5 minor 27 references

Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read An unsupervised clustering rule identifies line-breaking passes in the 2022 World Cup and quantifies their tactical worth.

desk verdict A plausible but unvalidated unsupervised detector for line-breaking passes; missing parameters and no benchmark make the rankings illustrative rather than established. read the letter →

arxiv 2506.06666 v1 pith:FUR2YIQJ submitted 2025-06-07 cs.LG stat.ML

Oktay Karakuş , Hasan Arkadaş This is my paper

classification cs.LGstat.ML

keywords line-breakingpassesfootballanalyticstrackingdataagglomerativeclusteringunsuperviseddetectionspacebuild-upratiotacticalmetrics2022FIFAWorldCup

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a pass can be classified as line-breaking without any labelled examples: at the moment of the pass, cluster the defending team's players by their lateral position into a small number of vertical bands, and call the pass line-breaking if its path crosses one of those bands while moving forward and bypassing at least two opponents. Applied to all 64 matches of the 2022 FIFA World Cup, the rule finds 7,477 line-breaking passes among 21,349 passes, about 117 per match. The paper adds three metrics—the Space Build-Up Ratio (SBR) and two chain variants, LBPCh1 and LBPCh2—to separate raw break volume from real spatial gain and from chances created. A reader would care because this offers an explainable and reproducible way to scout and compare teams and players on vertical penetration without relying on proprietary labels, and the paper's rankings tie frequent line-breaking to deep tournament runs.

What carries the argument

The load-bearing object is the rule in Eq. (1): opponent players at pass time are grouped into $k$ vertical clusters by x-coordinate; each cluster is represented by a vertical segment at its centroid $x_j$ spanning the y-range of its players; a pass counts as line-breaking if the pass segment intersects that vertical segment after crossing $x_j$, plus two filters requiring the pass to be forward and to bypass at least two opponents. Everything downstream—the 7,477 LBPs, the team and player rankings, the chain metrics—depends on this geometric definition. Two secondary mechanisms carry the evaluation: SBR($p_i$) = $d_r^2/d_p^2 - 1$, comparing the receiver's nearest-opponent space to the passer's, and the chain definitions LBPCh1 (a single LBP followed by a shot or assist in the same possession) and LBPCh2 (two consecutive LBPs then a shot, goal, or assist).

What would settle it

Take the 7,477 passes flagged by Eq. (1) plus a random sample of passes it rejected, have independent analysts label line-breaking versus not, and measure agreement; the claim stands only if agreement is high and if varying $k$ from 3 to 5 does not materially change which teams lead the LBP rankings.

Watch

Extended reading notes

Core claim

On its own terms, the central discovery is that a defensive line can be read as a cluster signature rather than a formation label. At each pass, the opponent's players are agglomeratively clustered by x-coordinate into $k$ vertical segments, and a pass is line-breaking exactly when it crosses a cluster's centroid within that cluster's y-span (Eq. 1), subject to a forward filter and a bypass-two-opponents filter. Across 21,349 passes this labels 7,477 LBPs, roughly 117 per match, with possession-heavy teams such as Croatia, Argentina, France, and Spain leading the rankings and three of the four semi-finalists among the top teams by total LBP count. The paper further reports that SBR separates volume from value—for instance, Andersen has many LBPs but low cumulative SBR—and that only 13 two-LBP chains (LBPCh2) occurred in the whole tournament, one of them a Morocco goal.

Load-bearing premise

The whole method rests on the assumption that grouping defenders into a fixed number of vertical columns and calling a pass line-breaking when it crosses one of those columns matches what a coach means by breaking a line; the paper never fixes $k$, the distance threshold, or the forward requirement, and does not check this geometric proxy against human labels.

Editorial extensions

If this is right

LBP volume can be read as a team-level measure of vertical intent; in this dataset Croatia, Argentina, France, and Spain lead, and three of the four semi-finalists rank among the top teams.
SBR should be reported alongside LBP count: a high-volume breaker can still be low-value, as the paper shows with Andersen, so scouting conclusions change when space gain is included.
LBPCh1 identifies the players and teams that convert a structural break into an immediate shot; Theo Hernández's four such events make him the clear outlier in the tournament.
LBPCh2 is a rare-event signal of coordinated progression—13 sequences in the whole tournament, mostly with xG below 0.10—so it measures tactical structure rather than finishing quality.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

An implication the paper leaves implicit: the same cluster-intersection rule could be run in reverse to score defensive resilience, counting how often a team's vertical bands are crossed, which would give a purely tracking-based defensive metric.
A testable extension: vary $k$ (e.g., 3, 4, 5) and the bypass threshold on the same data and check whether team rankings by LBP count are stable; the paper does not report this sensitivity analysis.
The chain notion could generalize to $n$ consecutive LBPs, allowing one to trace long multi-pass vertical buildups; the paper stops at two, so any n-chain behaviour is an open question.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick take: this is a useful, readable paper on a genuine gap, but the detection rule at its core is under-specified and unvalidated, so the headline numbers are hard to trust.

What's genuinely new: the combination of clustering opponent shape into vertical bands and calling a pass line-breaking when it crosses those bands is a reasonable unsupervised alternative to commercial 360 products. The SBR metric, d_r^2/d_p^2 - 1, is simple, interpretable, and measures how much space the pass buys. The chain definitions (LBPCh1, LBPCh2) are clearly laid out. Using the public PFF World Cup data is a plus, and the methodology section is straightforward enough to re-implement from the text—except for the missing parameters.

The soft spots are real and not minor. Equation (1) depends on k, the number of clusters, which is never reported. The 'bypass at least two opponents' and 'forward pass' filters are also left vague. Without those values, the detector is not fully defined, and the downstream SBR/chain metrics inherit that ambiguity. More importantly, there is no validation against anything: no human annotation, no comparison to StatsBomb 360 or Michalczyk's line-crossing heuristic. The only supporting evidence is the semi-finalist correlation, a four-team anecdote from one tournament that would probably survive under many vertical-pass definitions. The paper does not claim validation, but the abstract says the framework 'identifies' line-breaking passes, so the burden is on the authors to show the geometric proxy is right.

That said, the core idea is plausible, and the presentation is honest about being unsupervised. This is not a paper with a hidden circular step or a fabricated dataset; it is a promising framework that needs reproducibility and validation. The 'future work' mentions releasing code, but no code is attached yet, which limits what a referee can check.

Who should read it: football analysts and sports analytics researchers looking for a transparent way to measure vertical penetration. I would send it to peer review—it deserves a real referee—but I would require the authors to report all parameters, run a sensitivity analysis, and benchmark against at least one existing definition before publication.

Recommendation: engage with it, but treat the current numbers as illustrative, not established.

Referee Report

3 major / 5 minor

Summary. The paper proposes an unsupervised, clustering-based framework to detect line-breaking passes (LBPs) from synchronised event and tracking data. Opponent players are grouped into a fixed number k of vertical clusters based on x-coordinates, and a pass is labelled as line-breaking if it crosses a cluster centroid segment and satisfies two additional filters (bypassing at least two opponents and being forward). The authors introduce three metrics built on this detector: LBP volume, SBR (space build-up ratio), and two chain-based variants LBPCh1 and LBPCh2. They apply the framework to the 2022 FIFA World Cup dataset and report team and player rankings, claiming that three of the four semi-finalists rank among the top teams in total LBP count. The paper frames the contribution as an explainable, reproducible, and scalable alternative to proprietary LBP definitions.

Significance. If the detection rule were validated, the proposed framework would be a useful interpretable baseline for line-breaking pass analysis, and the SBR and chain metrics add vocabulary for discussing spatial and tactical impact. The use of a publicly available dataset and the intention to release code are strengths that support reproducibility. However, the significance is currently conditional: the core detector is under-specified and unvalidated against any reference definition, so the descriptive findings in Section 5 rest on an unestablished foundation. The work is best seen as a promising methodological proposal that needs a validation step before the substantive tactical claims can be assessed.

major comments (3)

[§4.1, Eq. (1)] The detection rule is not fully specified. Eq. (1) defines only the cluster-intersection condition, but the accompanying text adds two supporting filters (bypass at least two opponents, and forward pass) that are not formalized; there is no threshold for 'lateral proximity' in the bypass filter, no definition of forward direction (e.g., angle relative to goal line), no value or selection criterion for the number of clusters k, and no specification of the agglomerative clustering linkage or distance metric. Because k controls how many vertical bands are fitted to the opponent's shape, applying a single fixed k across all 64 matches is an untested prior; formations vary in the number of defensive and midfield lines. Since every downstream metric (LBP volume, SBR, LBPCh1, LBPCh2) conditions on this detector, the central claim that Eq. (1) correctly identifies line-breaking passes is not yet established. Please provide explicit parameter values, justify their choice, and include a sensitivity analysis over k and the bypass threshold.
[§5 and §4.1] There is no external validation of the detector against any reference definition of line-breaking passes. The only supporting evidence offered is the observation in §5 that 'three of the four semi-finalists rank among the top teams in total LBP count,' which is a four-team anecdotal correlation from a single tournament and would plausibly survive under many alternative vertical-pass detectors. The paper does not compare against StatsBomb 360, Michalczyk's line-intersection heuristic, packing counts, or human annotations, nor does it report precision or recall on any labeled sample. Please add a validation subsection that quantifies agreement with an existing definition or a manually annotated set of passes, or at least a qualitative side-by-side comparison on a sample of passes.
[§5.1 and §5.2] The variable 'pass verticality' is used as a key axis in Figures 3 and 4 and as the basis for the claim that 'all top 50 players exhibit verticality values above 0.6, underscoring the strong association between LBPs and vertical tactical build-up,' but the manuscript never defines how pass verticality is computed. Without a definition, the reader cannot assess whether this association is informative or tautological (e.g., if verticality is defined as forward progress beyond some threshold, the correlation with line-breaking may be built in). Please define the verticality measure explicitly, including any angle or coordinate-based formula.

minor comments (5)

[§5.2, Figure 4 caption] The caption of Figure 4 reads 'Team (Top) and Player (Bottom)', but the text in §5.2 refers to 'Fig. 4-Bottom' for the team level and 'Fig. 4-Top' for the player level; this labelling conflict should be fixed.
[§5, Figure 2] The legend 'Lines Crossed' in Figure 2 is ambiguous: the x-axis is 'Total Number of LBPs' and the y-axis is 'Percentage of LBPs to Total Passes', while the text says bubble size reflects the total number of defensive lines broken. Please clarify what the legend refers to.
[Abstract and §5.2] There are typographical artifacts in the rendering, such as 'FIF A' in the abstract and dataset description, and 'LBP V olume' in §4.1; these should be corrected.
[§6 Conclusion] The conclusion states that the authors aim to release code and curated metrics, but no repository or URL is provided in the manuscript; the availability statement should be placed in a clearly marked section.
[§5.2] The discussion of 'the only player in the dataset with four LBPCh1 events' (Theo Hernández) is based on four instances; the text could more explicitly acknowledge the small-sample nature of this finding without changing the descriptive intent.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the detection rule, metrics, and rankings are self-contained definitions and empirical summaries with no fitted-input predictions.

full rationale

The paper's central derivation is a fixed geometric rule: Eq. (1) defines a line-breaking pass by whether the pass crosses the x-centroid of an agglomerative cluster and intersects its vertical span, with two static supporting filters (bypass at least two opponents, forward pass). No parameter is fitted to the outcomes it later reports; SBR, LBPCh1, and LBPCh2 are all definitions applied to the detector's output, and the team/player rankings are direct descriptive statistics. There is no prediction loop, no fitted parameter renamed as a finding, and no uniqueness theorem invoked. The only self-citations (refs. [12] and [19] to the first author's prior xG work) are cited as general xG contributions in the introduction and are not load-bearing for the LBP detection or metric claims. The forward-pass filter makes the LBP definition partly vertical by construction, but the reported verticality magnitudes (e.g., > 0.6 for the top 50 players) are empirical and not entailed by the definition, so this does not constitute a circular reduction. The paper is under-specified regarding k and thresholds, but under-specification is a correctness risk, not circularity. The derivation chain is therefore self-contained, and no circular step rises to the level requiring a formal flag.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central method rests on several unreported parameter choices (k, bypass tolerance, forward definition) and on the domain assumption that lateral x-clusters capture true defensive lines. These are the main sources of uncertainty.

free parameters (4)

k (number of vertical clusters)
The number of opponent clusters is used in Eq. (1) but never specified or justified; results likely depend on it.
bypass threshold (minimum bypassed opponents) = 2
The filter 'must bypass at least two opponents' uses an unreported distance tolerance for 'lateral proximity to the pass vector'.
forward direction threshold
The forward-pass filter is stated but no quantitative definition is given.
agglomerative clustering linkage and distance metric
The clustering method is named but no linkage criterion, distance measure, or cluster validation is described.

assumptions (4)

domain assumption Defensive lines correspond to clusters of players with similar lateral (x) coordinates.
Section 4.1 models opponent shape by vertical segmentation into k clusters based only on x-coordinates.
domain assumption The pass path is a straight line from passer to receiver.
Equation (1) checks whether the pass path intersects a vertical segment.
domain assumption PFF FC tracking and event data are synchronized and sufficiently accurate.
Section 3 asserts synchronization and 29.97 Hz tracking without error analysis.
domain assumption Possession phases can be identified consistently from the event data.
LBPCh1 and LBPCh2 rely on 'same possession phase' but no possession model is described.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering." pith.science (2026). https://pith.science/paper/FUR2YIQJ

@misc{pith2026250606666,
  author       = {Pith},
  title        = {Pith review of: Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FUR2YIQJ}},
  note         = {Machine review of arXiv:2506.06666}
}

abstract

Line-breaking passes (LBPs) are crucial tactical actions in football, allowing teams to penetrate defensive lines and access high-value spaces. In this study, we present an unsupervised, clustering-based framework for detecting and analysing LBPs using synchronised event and tracking data from elite matches. Our approach models opponent team shape through vertical spatial segmentation and identifies passes that disrupt defensive lines within open play. Beyond detection, we introduce several tactical metrics, including the space build-up ratio (SBR) and two chain-based variants, LBPCh$^1$ and LBPCh$^2$, which quantify the effectiveness of LBPs in generating immediate or sustained attacking threats. We evaluate these metrics across teams and players in the 2022 FIFA World Cup, revealing stylistic differences in vertical progression and structural disruption. The proposed methodology is explainable, scalable, and directly applicable to modern performance analysis and scouting workflows.

Figures

Figures reproduced from arXiv: 2506.06666 by the authors.

**Figure 1.** Illustration of an LBP and an SBR. and progressive play. Across all 64 matches and 21,349 passes in the dataset, the proposed model identified 7,477 LBPs, averaging around 117 per match [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗

**Figure 2.** Most frequent line-breakers as detected by our clustering-based model. Top: Team-level; Bottom: Playerlevel. Each bubble represents an entity about LBPs where the bubble size reflects the total number of defensive lines broken [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Space progression effectiveness of the top 50 players. The x-axis shows the percentage of LBPs with positive SBR, whilst the y-axis represents the cumulative SBR. Bubble size indicates average pass distance, and colour encodes pass verticality. consistently generate space and accumulate high total gains. In contrast, Amrabat shows a high success rate but modest cumulative values despite topping the verticality scale… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Direct Line-Breaks to Chances (LBPCh1 ). Team (Top) and Player (Bottom) Level. Bars (i) represent the cumulative SBR for LBPs that directly led to a shot, (ii) are coloured by average pass verticality, and (iii) are ordered by total LBPCh1 [PITH_FULL_IMAGE:figures/fu…

**Figure 5.** Figure 5: Summary of all detected LBPCh2 sequences. Each row represents an LBPCh2 . The table includes the initiating, connecting, and finishing players, match details, final outcomes (with xG) and chain cumulative SBR values. 6 Conclusion In this study, we proposed an unsupervi…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

27 extracted references · 26 canonical work pages

[1]

https://dataglossary.wyscout.com/progressive_pass/ (2021), accessed: 2025-05-30 Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering 11

Progressive pass. https://dataglossary.wyscout.com/progressive_pass/ (2021), accessed: 2025-05-30 Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering 11

work page 2021
[2]

Data Mining and Knowledge Discovery 31, 1793–1839 (2017)

Andrienko, G., Andrienko, N., Budziak, G., Dykes, J., Fuchs, G., Von Landesberger, T., Weber, H.: Visual analysis of pressure in football. Data Mining and Knowledge Discovery 31, 1793–1839 (2017)

work page 2017
[3]

In: Proceedings of the 2014 IEEE International Conference on Data Mining

Bialkowski, A., Lucey, P., Carr, P., Yue, Y., Sridharan, S., Matthews, I.: Large-scale analysis of soccer matches using spatiotemporal tracking data. In: Proceedings of the 2014 IEEE International Conference on Data Mining. pp. 725–730. IEEE (2014)

work page 2014
[4]

Master Thesis, Erasmus University Rotterdam (2017), https://thesis.eur.nl/pub/41346/Bransen.pdf

Bransen, L.: Valuing passes in football using ball event data. Master Thesis, Erasmus University Rotterdam (2017), https://thesis.eur.nl/pub/41346/Bransen.pdf

work page 2017
[5]

Journal of Quantitative Analysis in Sports 15(2), 97–116 (2019)

Bransen, L., Van Haaren, J., van de Velden, M.: Measuring soccer players’ contributions to chance creation by valuing their passes. Journal of Quantitative Analysis in Sports 15(2), 97–116 (2019)

work page 2019
[6]

In: KDD (2019)

Decroos, T., Bransen, L., Van Haaren, J., Davis, J.: Actions speak louder than goals: Valuing player actions in soccer. In: KDD (2019)

work page 2019
[7]

In: Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence

Decroos, T., Bransen, L., Van Haaren, J., Davis, J.: Vaep: an objective approach to valuing on-the-ball actions in soccer. In: Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence. pp. 4696–4700 (2021)

work page 2021
[8]

https://www.driblab.com/blog/ arrigo-line-breaking-actions-and-ball-into-space (2025), accessed: 2025-05-30

Driblab: Arrigo: Line-breaking actions and ball into space. https://www.driblab.com/blog/ arrigo-line-breaking-actions-and-ball-into-space (2025), accessed: 2025-05-30

work page 2025

Show all 27 references

[9]

In: 3rd Workshop on Machine Learning and Data Mining for Sports Analytics (MLSA 2016)

Eggels, H., Van Elk, R., Pechenizkiy, M.: Explaining soccer match outcomes with goal scoring opportunities predictive analytics. In: 3rd Workshop on Machine Learning and Data Mining for Sports Analytics (MLSA 2016). CEUR-WS. org (2016)

2016
[10]

https://www.cs

El Kadi, T.: Machine learning models for detecting line-breaking passes in football matches. https://www.cs. vu.nl/~wanf/theses/el_kadi-bscthesis.pdf (2024), accessed: 2025-05-30

2024
[11]

In: KDD Workshop on Large-Scale Sports Analytics (2014)

Gyarmati, L., Kwak, H., Rodriguez, P.: Qpass: a merit-based evaluation of soccer passes. In: KDD Workshop on Large-Scale Sports Analytics (2014)

2014
[12]

Franklin Open 4, 100034 (2023)

Hewitt, J.H., Karaku¸ s, O.: A machine learning approach for player and position adjusted expected goals in football (soccer). Franklin Open 4, 100034 (2023)

2023
[13]

https://expandyourgame

Ireland, J.: What the heck are packing points in soccer? only the best statistic ever. https://expandyourgame. com/using-packing-points-soccer-game-statistic/ (2021), accessed: 2025-05-30

2021
[14]

Lucey, P., Bialkowski, A., Monfort, M., Carr, P., Matthews, I.: Quality vs Quantity: Improved shot prediction in soccer using strategic features from spatiotemporal data (2015)

2015
[15]

Sports Medicine 47(1), 1–10 (2017)

Memmert, D., Lemmink, K., Sampaio, J.: Current approaches to tactical performance analyses in soccer using position data. Sports Medicine 47(1), 1–10 (2017)

2017
[16]

Michalczyk, K.: How impactful are line-breaking passes? https://www.statsperform.com/resource/ how-impactful-are-line-breaking-passes/ (2020), accessed: 2025-05-30

2020
[17]

arXiv preprint arXiv:1109.2378 (2011)

M¨ ullner, D.: Modern hierarchical, agglomerative clustering algorithms. arXiv preprint arXiv:1109.2378 (2011)

2011 arXiv
[18]

Scientific Reports 9(1), 13172 (2019)

Narizuka, T., Yamazaki, Y.: Clustering algorithm for formations in football games. Scientific Reports 9(1), 13172 (2019)

2019
[19]

Frontiers in Sports and Active Living 6, 1348983 (2024)

Scholtes, A., Karaku¸ s, O.: Bayes-xG: player and position correction on expected goals (xG) using Bayesian hierarchical approach. Frontiers in Sports and Active Living 6, 1348983 (2024)

2024
[20]

Singh, K.: Introducing expected threat (xt) (2018, Personal Blog - Accessed:July 2023), https://karun.in/blog/ expected-threat.html

2018
[21]

https://www.statsperform.com/ opta-vision/ (2023), accessed: 2025-05-30

Stats Perform: Generative AI Powered Football Insights: At Scale. https://www.statsperform.com/ opta-vision/ (2023), accessed: 2025-05-30

2023
[22]

https://statsbomb.com/articles/soccer/ statsbomb-obv-model/ (2023), accessed: 2025-05-30

StatsBomb: OBV: On-Ball Value Model Documentation. https://statsbomb.com/articles/soccer/ statsbomb-obv-model/ (2023), accessed: 2025-05-30

2023
[23]

Journal of the Royal Statistical Society: Series A (Statistics in Society) 179(2), 513–533 (2016)

Szczepa´ nski, L., McHale, I.: Beyond completion rate: Evaluating the passing ability of footballers. Journal of the Royal Statistical Society: Series A (Statistics in Society) 179(2), 513–533 (2016)

2016
[24]

Medium blog post (2024), https://half-space.medium.com/space-creation-4e928f989f84 12 Oktay Karaku¸ s and Hasan Arkada¸ s

Tayyab: Space creation — using tracking data to quantify line breaking passes & space creation. Medium blog post (2024), https://half-space.medium.com/space-creation-4e928f989f84 12 Oktay Karaku¸ s and Hasan Arkada¸ s

2024
[25]

In: Proceedings of the AAAI-20 Workshop on Artifical Intelligence in Team Sports

Van Roy, M., Robberechts, P., Decroos, T., Davis, J.: Valuing on-the-ball actions in soccer: a critical comparison of xt and vaep. In: Proceedings of the AAAI-20 Workshop on Artifical Intelligence in Team Sports. AI in Team Sports Organising Committee (2020)

2020
[26]

https://statsbomb.com/articles/soccer/ statsbomb-360-exploring-line-breaking-passes/ (2022), accessed: 2025-05-30

Yorke, J.: Statsbomb 360: Exploring line-breaking passes. https://statsbomb.com/articles/soccer/ statsbomb-360-exploring-line-breaking-passes/ (2022), accessed: 2025-05-30

2022
[27]

Academic Journal of Clinical Research and Reports 1(1), 1–14 (2025)

Zare Zardiny, A., Bahramian, Z.: A Two-Stage Spatio-Geometrical Clustering of Football Team Shape for Post- Match Review. Academic Journal of Clinical Research and Reports 1(1), 1–14 (2025)

2025

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

https://dataglossary.wyscout.com/progressive_pass/ (2021), accessed: 2025-05-30 Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering 11

Progressive pass. https://dataglossary.wyscout.com/progressive_pass/ (2021), accessed: 2025-05-30 Through the Gaps: Uncovering Tactical Line-Breaking Passes with Clustering 11

work page 2021

[2] [2]

Data Mining and Knowledge Discovery 31, 1793–1839 (2017)

Andrienko, G., Andrienko, N., Budziak, G., Dykes, J., Fuchs, G., Von Landesberger, T., Weber, H.: Visual analysis of pressure in football. Data Mining and Knowledge Discovery 31, 1793–1839 (2017)

work page 2017

[3] [3]

In: Proceedings of the 2014 IEEE International Conference on Data Mining

Bialkowski, A., Lucey, P., Carr, P., Yue, Y., Sridharan, S., Matthews, I.: Large-scale analysis of soccer matches using spatiotemporal tracking data. In: Proceedings of the 2014 IEEE International Conference on Data Mining. pp. 725–730. IEEE (2014)

work page 2014

[4] [4]

Master Thesis, Erasmus University Rotterdam (2017), https://thesis.eur.nl/pub/41346/Bransen.pdf

Bransen, L.: Valuing passes in football using ball event data. Master Thesis, Erasmus University Rotterdam (2017), https://thesis.eur.nl/pub/41346/Bransen.pdf

work page 2017

[5] [5]

Journal of Quantitative Analysis in Sports 15(2), 97–116 (2019)

Bransen, L., Van Haaren, J., van de Velden, M.: Measuring soccer players’ contributions to chance creation by valuing their passes. Journal of Quantitative Analysis in Sports 15(2), 97–116 (2019)

work page 2019

[6] [6]

In: KDD (2019)

Decroos, T., Bransen, L., Van Haaren, J., Davis, J.: Actions speak louder than goals: Valuing player actions in soccer. In: KDD (2019)

work page 2019

[7] [7]

In: Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence

Decroos, T., Bransen, L., Van Haaren, J., Davis, J.: Vaep: an objective approach to valuing on-the-ball actions in soccer. In: Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence. pp. 4696–4700 (2021)

work page 2021

[8] [8]

https://www.driblab.com/blog/ arrigo-line-breaking-actions-and-ball-into-space (2025), accessed: 2025-05-30

Driblab: Arrigo: Line-breaking actions and ball into space. https://www.driblab.com/blog/ arrigo-line-breaking-actions-and-ball-into-space (2025), accessed: 2025-05-30

work page 2025