In soccer, the pass is often considered the simplest play—almost invisible because it seems so natural. Yet it is the essential tactical building block that shapes a game, organizes a press, breaks through a defensive line, or causes a defensive block to collapse. Until now, analyzing a pass meant measuring a success rate or commenting on an intention. With PassAI, an algorithm derived from recent work in explainable artificial intelligence, this perspective changes radically. For the first time, an AI model is able to classify and explain the success or failure of a pass by simultaneously drawing on tracking video and the seasonal statistics of the player making it. In other words, PassAI doesn’t just tell you what happened; it explains why it happened1. This development restores the pass to its full strategic importance by making intelligible what, until now, relied on intuition or experience.
\n\n\n\nOne of the major challenges of AI in sports lies in integrating heterogeneous data sources. The researchers behind PassAI have addressed this challenge by building a model based on two independent and complementary information streams. The first processes tracking images, i.e., the player’s position, orientation, the ball’s speed, and the density of opponents around him. The second uses the passer’s statistics, playing tendencies, success rate in similar situations, and risk-taking profile. Thanks to this structure, PassAI achieves a more nuanced understanding of the play. It does not evaluate the pass based solely on the image or solely on the data; instead, it combines both to approximate the reasoning of a tactical expert. The results are compelling: PassAI improves upon the performance of the best existing algorithms by more than 5% on a dataset comprising 6,349 passes from professional matches1.
\n\n\n\nPerhaps PassAI’s most significant innovation lies in its ability to explain its decisions. Unlike many sports algorithms that operate as black boxes, PassAI calculates the relative impact of both data sources on the classification. For example, it identifies whether a failed pass is primarily due to the spatio-temporal context visible in the video or to player-specific statistical factors. It goes even further by shedding light on factors internal to each data set, such as the proximity of three opponents, poor body positioning, or the player’s history in high-pressure situations. For a coaching staff, this level of detail changes everything. Instead of a simple “success” or “failure” label, the coach receives a detailed explanation that can inform a video session, individual feedback, or a game plan.
\n\n\n\nIn soccer, where decisions must be made in a split second, understanding the conditions that make a pass risky or optimal is a huge strategic advantage. PassAI can be used to identify team-wide trends, such as the areas where a player is most successful with his passes, the situations in which he forces an unlikely pass, or the sequences where the team would be better off switching play earlier. In professional clubs, this information can be used to:
\n\n\n\nFor tactical analysts and data scientists working within clubs, PassAI thus serves as a dual-purpose tool—both diagnostic and educational—enabling them to back up their instincts with visual and statistical data that are perfectly aligned.
\n\n\n\nPassAI isn’t just for elite players. In youth academies, this kind of model can serve as a guide to support young players’ development by objectively analyzing their decisions with unusual precision. Coaches can use real-life situations to demonstrate why a pass was too late, why a player’s body wasn’t properly positioned, or why a player failed to identify an open space. By making often implicit cognitive mechanisms explicit, PassAI could accelerate the learning process, enhance tactical understanding, and reduce variability in technical decision-making among young athletes.
\n\n\n\nLike any algorithmic tool used in professional sports, PassAI requires strict oversight. From an ethical standpoint, several questions arise.
\n\n\n\nExperts also emphasize the risk of overinterpretation: while an algorithm explains a play based on the data it has, it cannot replace a player’s feel for the game or the intuition of an experienced coach2. AI must remain a support tool, not a judge of the game or of an individual’s worth.
\n\n\n\nPassAI is part of a broader revolution—one in which every movement, every intention, and every decision in soccer becomes analyzable, explainable, and improvable. By revealing the hidden mechanisms behind ball movement, it offers clubs, analysts, and coaches a unique tool for understanding the dynamics of the game. Soccer isn’t just about spectacular plays; it’s made up of thousands of micro-decisions. And thanks to AI, these micro-decisions are finally becoming visible, understandable, and optimizable.
\n\n\n\nTo deepen your understanding of new approaches to advanced analysis in soccer, you can read our study on identifying behavioral dynamics on the field, a field that complements the analysis of passing and decision-making: AI observes what cameras don’t see: the psychological profile of players
\n\n\n\n1. Kim. (2025). PassAI, Explainable AI for Soccer Pass Success Classification.
\nhttps://www.journal-sport-ai.org \n
2. International Council for Ethics in Sports Analytics. (2024). Guidelines for Explainable AI in Tactical Decision Making.
https://www.icesportanalytics.org \n
In soccer, the pass is often seen as the simplest play—almost invisible because it seems so natural. Yet it is the essential tactical building block that shapes a game, organizes a press, and breaks through a defensive line…
\n","protected":false},"author":2,"featured_media":466201,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[146],"tags":[],"class_list":{"0":"post-466184","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-ia-sport"},"yoast_head":"\n