Player Vectors - Characterizing Soccer Players' Playing Style from Match Event Streams

Oct 22, 2026

Upload by Eliana

Transfer fees for soccer players keep rising, creating pressure for clubs to understand what types of players they have and which profiles are available on the market. Existing playing-style analyses often rely on subjective opinions from experts, which can be error-prone. This paper presents an objective, data-driven characterization using a fixed-length “player vector,” interpretable by both humans and machine learning systems. Validity is shown by retrieving identities from anonymized event streams and by demonstrating scouting and development monitoring use cases.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Introduction

Event Stream Data

Data and Challenges

Player Vectors: Characterizing Soccer Players'Playing Style from Match Event Streams Tom Decroos and Jesse Davis KU Leuven, Department of Computer Science Abstract. Transfer fees for soccer players are at an all-time high. To make the most of their budget, soccer clubs need to understand the typeof players they have and the type of players that are on the market. Current insights in the playing style of players are mostly based on the opinions of human soccer experts such as trainers and scouts. Unfortunately, their opinions are inherently subjective and thus prone to faults. In this paper, we characterize the playing style of a player in amore rigorous, objective and data-driven manner. We characterize the playing style of a player using a so-called `player vector' that can be interpreted both by human experts and machine learning systems. We demonstrate the validity of our approach by retrieving player identities from anonymized event stream data and present a number of use cases related to scouting and monitoring player development in top European competitions. 1 Introduction Data analysis is becoming increasingly important in many professional sports [19] . Sports clubs are analyzing huge amounts of data in order to gain a competitive advantage over their opposition. Soccer has been a relative late comer to this trend. The classic statistics about a soccer match (e.g., that appear in boxscores or are often reported on television) tend to be raw counts or fractions, such as the ball possession percentage, number of shots on target or pass success percentage. While interesting, these statistics do not give a complete picture of the match. Moreover they can sometimes obscure important information. For example, the raw number of shots a player took does not tell us about the relative di􀀎culty or quality of the attempts. Recently, research has focused on approaches that allow for a deeper and more insightful analysis of soccer players [4] . One well-known example is the expected goals statistic [9], which is now discussed on the popular soccer talk show Match of the Day on BBC One. A reoccurring concept when discussing soccer is a player's style of play. While both Messi and Ronaldo are great players, each one approaches the game in adi􀀋erent way. Fans often form preferences for players based on his perceived style. From a practical point of view, characterizing playing style is important for professional clubs for three following reasons: Scouting Soccer clubs can search the market more intelligently if they know the type of player they are looking for and how well prospective targets match 2 Tom Decroos and Jesse Davis that type. Transfers are expensive, and clubs are always looking for bargainsand ways to mitigate risks in player recruitment. Monitoring player development The coach can inspect the playing style of a player in a human-interpretable player vector. If the player vector matches the expectations of the coach, then the coach can monitor that this player vector remains stable and unchanged. If the player vector does not match the expectations of the coach, then he can give his player some pointers and afterwards monitor how well the player is implementing the advice. Match preparation Understanding the playing style of your opponent cano􀀋er certain tactical advantages. The defenders of a team will wish to know what type of attackers they are up against. Similarly, the attackers will be interested in the playing style of the defenders they need to score against. In this paper, we attempt to characterize a player's playing style in an objective and data-driven manner based on analyzing event stream or play-by-play match data. While playing style is a somewhat subjective concept, our working de􀀌nition is that a playing style manifests itself as where on the pitch a player tends to perform speci􀀌c actions with the ball. Our goal is to summarize this playing style in a 􀀌xed-length player vector

FAQs

Why do soccer clubs need an objective way to characterize playing style?

Clubs must manage expensive transfers and reduce recruitment risk by understanding player profiles. The paper also frames scouting, monitoring development, and match preparation as practical reasons to characterize style reliably.

What is a “player vector” in this work?

A player vector is a fixed-length representation derived from match event stream data. It summarizes where on the pitch a player tends to perform specific ball-related actions.

How are player vectors constructed from event stream data?

The method overlays a grid on the pitch and counts action frequency per location, then applies non-negative matrix factorization to reduce dimensionality. Compressed vectors for multiple action types are concatenated to form the final player vector.

Oct 22, 2026

Upload by Eliana