Understanding the features of player movement within team sports is a key component in athlete monitoring, performance evaluation, and training design processes (1). Determining what movement is and how to measure it effectively are fundamental requirements for developing strategies to improve sporting performance. Player movement is often defined and measured at varying levels of complexity across team sports and high-performance environments (2, 3). Developing a framework that acknowledges this complexity would allow for greater collaboration between disciplines by ensuring a shared language, potentially leading to better-applied outcomes.

Player movement can be measured across multiple scales, for example, the kinematic analysis of an anatomical structure's function through various planes of motion or positional descriptions of an individual in response to team tactical behaviour (4). In team-based invasion sports, movement is increasingly viewed through a spatiotemporal lens. Spatiotemporal data typically include the location of individuals or events with reference to space and time (5). Spatiotemporal features of movement can be described using one-dimensional derivatives of displacement, such as velocity, acceleration, and their angular properties; two-dimensional representations of manoeuvrability, such as tortuosity (5, 6); or three-dimensional depictions of limb kinematics (7), which are increasingly being used for in situ technique analysis, as well as various adjudication applications such as semi-automated offside assessment in football (8). As such, identifying a technology solution capable of measuring the intended movement feature with appropriate validity and reliability is paramount to successful evaluation.

Tracking systems such as Global Navigation Satellite Systems (GNSS), Local Positioning Systems (LPS), and Optical Tracking Systems (OTS) have provided the means to capture spatiotemporal information within indoor and outdoor team sports competitions. Such solutions provide positional data (longitude and latitude coordinates) and aggregated descriptors of movement, such as total running volume and maximal velocity. These systems often contain inertial measurement units (IMUs), which include a triaxial accelerometer, a triaxial gyroscope, and a triaxial magnetometer. IMUs have been used to assess kinematic movements such as stride length, flight time, and ground contact time, as well as acceleration metrics and the automatic detection of match phases (9, 10). A “player load” or “dynamic load” metric is an example of triaxial resultant information derived from an IMU being operationalised by a technology provider (11). Developments to these operating systems have seen capturing rates evolve from 1 to 100–300 Hz (9, 12, 13). In addition, computer vision research has brought about optical tracking algorithms that adopt deep learning models to produce ball or player-tracking solutions (14). Adoption of optical tracking solutions in sports varies due to unique sport-specific requirements. For example, greater revenue and salary cap confinements in the National Basketball Association (NBA), combined with controlled playing conditions (indoor environment, fixed camera operating systems, and fewer players to track), have granted relative access to the sport (15). Contrastingly, in Australian football, varying field sizes across the competition require significant camera infrastructure and therefore limit the viability of such a system (14).

This review aims to provide a framework for analysing players’ physical movement within team sports. To inform its development, this study evaluates the current research and identifies several open questions to guide future investigations.

To understand player movement beyond its semantic value, research has assessed the influence of various contexts. The constraint-led approach (CLA) provides an appropriate convention to categorise contexts previously used to measure sporting performance (2, 16). Constraints refer to the boundaries by which movement strategies emerge (2) and are categorised as individual, task, or environmental (17). Individual constraints refer to contextual factors that describe physical and technical attributes, historical descriptions of experience, and states of cognition (2). Task constraints describe rules and instructions and are related to the task goal (2, 18). Environmental constraints are external to the organism and refer to the physical or sociocultural conditions surrounding performance (2, 18). Table 1 provides a selection of constraints used to provide context to player movement in team sports. For example, understanding individual movement behaviour during offensive and defensive periods of play allows performance staff to make informed decisions regarding training load prescription, while understanding the influence of numerical imbalances may assist coaches in making more informed decisions when their team is influenced by injuries or player exclusions.

Research will likely continue to build on the body of work mentioned above, largely driven by rapid technology enhancements. While this alone will probably lead to incremental improvements in the understanding of player movement in team sports, access to more data also has the potential to lead to misuse and/or misinterpretation. As a result, additional attention being paid to the areas below will likely have an accelerative and perhaps more fundamental impact comparatively.

Technological improvements to commercially available tracking systems have afforded the end user the ability to answer related questions in greater detail. For example, near real-time access to limb tracking enables precise rule adjudication and technique analysis. Prior to these recent improvements, the utilisation of tracking data to date has often been limited to the use of discrete representations of movement. For example, speed is captured at 10 Hz via GNSS but is often reported as an average speed over the duration of a drill. This is likely partly due to the lack of computational power to deliver outputs in a timely manner. These representations include describing the movement as a volume of running within predetermined arbitrary velocity thresholds, counts of efforts above set thresholds, and measures of intensity derived by volume and duration metrics (107, 138, 139). Such representations of movement do not consider the emergent patterns within team sports and reduce human behaviour to isolated numerical values (140, 141). While value models and rule adjudication techniques in sports routinely use continuous movement data, their benefits are yet to be fully realised in routine practice. Discrete representations of movement are easily understood and acted upon, whereas continuous data require greater computation capacity but allow for the complexity within the sport to be accounted for. For example, previous examinations of maximal periods of movement have provided a guide to infer training intensity, whereas the automation of skeletal tracking data from competition would support the development of three-dimensional representations of movement during different components of the game (e.g., fast break, one-on-one situations, etc.). This actualises the potential for real-time coaching, omnipresent monitoring, more precise training prescription, and enhanced broadcast fan engagement features.

Another potential accelerator in this area relates to the implementation of theoretical frameworks to underpin player movement analysis. The adoption of an overarching, organisation-wide theoretical framework has been suggested as a method of unifying action across disciplines (16). Such a framework would offer an approach to combining operational methodologies and techniques alongside theory and a set training philosophy (2, 142, 143). It could also serve to coordinate and unify activity, redirect resource allocations, stimulate coherent interdisciplinary communication, and promote a more holistic understanding of sports performance (5, 144). The collaboration of subdiscipline specialists allows for analysis to be communicated through consistent language, which could influence the transfer of knowledge and ultimately improve performance (41). Within sports science, the integration of a framework allows scientific methodologies to reflect the broader training philosophy of organisations. Within the strength and conditioning discipline, it underpins exercise selection and progression. Within the coaching discipline, it underpins training design and performance evaluation, while in the medical discipline, it underpins rehabilitation design and physical screening.

Historically rooted in the physics literature, complex systems theory has recently been adopted by sports practitioners to underpin the analysis of movement (62, 121, 145, 146). This theory offers a framework through which a system’s stability, variability, and transition between fluid states can be described (147). Its adoption would be advantageous to sports performance, as this process-oriented philosophy embraces the complex constructs that operate within the team-sporting performance (147). Although all sports performance staff may benefit from its adoption, analysts and coaches are examples of two beneficiaries of complex systems research. The analysis of collective movement behaviour has enabled analysts to describe opposition ball movement methods and train accordingly, while the documentation of constraints on performance has permitted coaches to manipulate constraints in training to promote learning outcomes (148, 149). Complex systems theory describes athletes and teams as a complex adaptive system (CAS) whose behavioural properties continuously evolve in response to changing environments (12) and are unable to be understood outside of its context (150). In team-based sports, the theory recognises that individuals and teams regulate tactical behaviour in response to changing match constraints. It considers that control of tactical behaviour is distributed among different players and that individuals have the capacity to learn (4, 41, 151). Previous investigations have utilised a CAS approach to conceptualise team synergy in various sporting environments (13, 62, 121). Recent efforts to enhance interpretability have manifested in updated definitions and visual depictions of the 16 defining features of CAS theory (152). Complex systems theory provides a framework in which data analysis techniques can reference theory in the analysis of movement features. Table 2 provides an outline of the CAS features, their definitions, an analytical method that aligns with each feature, examples from the literature, and their proposed influence on practice.

More detailed data require not only greater computational but also analytical capacity. This permits moving beyond viewing sporting contexts in isolation and provides the ability to establish causal relationships between constraint groups or movement features (153). Such methodology offers a one-dimensional representation of movement while emphasising outcome behaviour and neglecting motives and causes (153). Recent investigations have adopted multivariate or time series approaches that address the temporal features and relatedness within constraints (18, 154–156). This allows the analytical process to be aligned to a feature within a CAS.

The proposed framework below (Table 3) can assist practitioners and researchers in extracting the most from player movement data while developing operational standards that align with a set training philosophy. An example using a complex systems theory to evaluate the change in running intensity throughout a match has been included as a case study.

Assuming progress in each of the above-mentioned areas, both practice and research are well-positioned to find solutions to further pervasive and impactful questions relating to player movement in team sports. A summary of some of these questions, many of which have already been posed in the literature, aimed at stimulating future investigations is provided as follows: •

OPEN QUESTION 1: Existing models for evaluating the difficulty of skilled actions have been made assessing the two-dimensional distance between ball carrier and receiver/target. Can a model be developed that accounts for three-dimensional space, the effect of ball spin and the role movement has on skill execution? (12) Such information would provide practitioners with a method of evaluating the complexities of skilled behaviour that could influence the training design and analysis processes.

As long as technology continues to develop, the collection and analysis of spatiotemporal data will remain a common practice in high-performance sporting environments. Given this persistent nature, it is imperative that the analysis and description of player movement reflect not only the underlying physical quality but also the broader department philosophy rather than being reactive to resource availability. The merger of a set training philosophy with appropriate data analytical tools allows for the assessment of the complex features of movement in team sports. The stated open questions illustrate nascent areas of sports practice in which benefits could be actualised through the unification of theory and analytics. Such practice describes a methodology where data analysts operate within the confinements of a broader department-shared philosophy when analysing human physical movement. Ultimately, this promotion of interdisciplinary collaboration has the potential to achieve better-applied outcomes for players and organisations.