Sport Analytics: Graduating from Alchemy

Abstract

Sport analytics allows sport teams and organizations to improve performance and associated business decisions. There is an increasing demand for sport analytics, in part connected to the emergence of Big Data, resulting in a new discipline in the sport industry. Business models related to sport analytics offer the opportunity to analyze the performance of athletes, teams, clubs, and sport organizations. The burgeoning yet competitive objectives based on sport analytics explain, to a degree, why it is rare to find algorithms, predictive models, and other statistical methods and analyses being carried out in the public domain. This chapter first outlines topical views of the developing field of sport analytics that suggest that its application is based on organizational self-interest, resulting in a degree of obfuscation that may limit the pursuit of knowledge. Countering these opinions, however, is evidence pointing to sport analytics becoming more mainstream and a domain of shared knowledge. The chapter provides a non-exhaustive literature review, including sections addressing statistical elements, performance optimization, theoretical frameworks, and the application of sport analytics, followed by some overall observations. Within that context, recent developments in the sport industry demonstrate that sport analytics is more than alchemy.

1. Introduction

Described as the analysis and modeling of sporting performance and achievement [1] and a data-driven approach to the business of sport [2, 3], the application of statistical methods within sport facilitates improved athletic performance for athletes and teams, ultimately leading to success for sport organizations. Globally, the sport analytics market was valued at circa USD 3 billion in 2022 and is projected to be worth USD 22 billion by 2030 [4]. With the advent of the age of Big Data [5] in sport – viz., the application of massive datasets to identify patterns – sport analytics has evolved into a legitimate division of the sport industry [1]. The application of Big Data is affecting the sport industry multi-dimensionally [6, 7] and profoundly impacting the administration of aspects of professional sport [2]. By applying comprehensive datasets and facilitating probability over prediction, sport analytics impacts on-field sporting action and, in turn, the management of sport organizations [8, 9, 10].

From the perspective of sport organizations - and their athletes and teams - sport analytics includes statistical methods and quantitative models, and visualization surrounding these methods and models to improve performance. In short, applying mathematical and statistical principles in sport can enhance athletic performance and create a competitive advantage [11]. Sport analytics is applied in a structured manner, backed by research findings, and relies upon contributions from both the computer science and sport science field, along with statistics and engineering [12]. Despite the immense growth of sport analytics in becoming a part of mainstream sport, however, some critics challenge such a general viewpoint. For example, Szymanski [13] contends that “much of the work in sport analytics seems to possess the same traits as alchemy - secretiveness and obscurantism” (p. 60), thus relegating the discipline from an integral part of the mainstream sport industry. There are references to a lack of scholarly analysis relating to the theoretical application of Big Data and associated social, legal, and ethical concerns [14], the imbalance between elite sport organizations and comparatively poorly resourced sport organizations, including women’s sport [15, 16], and the overall objectives of sport organizations turning to sport analytics to gain a competitive advantage [13, 17] and improve business performance based on corporate objectives [18, 19, 20].

With the above in mind, the objective herein is to challenge such notions and show that sport analytics is now an essential functional and operational aspect for teams, clubs, and sport organizations [10, 19]. While it is not suggested that secrecy is not uncommon, particularly concerning the pursuit of competitive advantage, there is sufficient literature to demonstrate that the increasing importance of the discipline outweighs the concerns about the nature of sport analytics or suggestions that the field is not integral to the sport industry. On a non-exhaustive basis, this chapter provides some background and a literature review that forms a platform for some general observations for future research.

2. Background

Sport analytics has a rich history of applying prognostic statistical methods based on past performance and forecasting future performance [21, 22]. Used to measure the competitive advantage of teams, coaches, and individual participants, sport analytics represents a valuable tool for informing sporting decisions [10, 23]. Historically, sport analytics was primarily characterized by the video recording of competition or training sessions and the creation of video highlights that athletes and coaches would review pre-and post-match. More recent examples include the interpretation of external load profiles [24], the analysis of acute chronic workload ratios [25], the use of GPS competition data to create and analyze training drills [26], and the application of data to analyze the impact of substitutes in team competition [27]. Within such context, sport analytics applies to team sports such as football, cricket, rugby, athletics, basketball, golf, and niche or indoor sports such as hockey, chess, table tennis, and martial arts. Sha et al. [28] point out that analytics in professional sport has seen histrionic growth in the past decade due to technological advances observed during this period.

As the practice of sport analytics continues to develop and to attract sport scientists who increasingly rely on the predictive nature of the discipline, opinions are divided on the relevance of sport analytics [1]. While sport analytics offers an avenue to predict future performance and make informed decisions to facilitate competitive advantage predict [1, 29, 30, 31], skeptics have labeled sport analytics alchemy [13] and a discipline that comes with a degree of notoriety [32]. Amidst the apparent growth of sport analytics as a prophetic discipline in sport, opinions regarding sport analytics’ ineffectiveness are evident. For one, Szymanski [13] claims that much of the work in sport analytics is motivated by the need “to influence coaches and gain influence” (p. 61), which discourages honesty in sport, thus hindering the true spirit of sportsmanship in turn. There are opinions of those involved directly with sport, such as athletes and sports writers, who claim analytics is ruining sport [33, 34] and also an unsettling general debate about the position of sport analytics in mainstream sport [11, 32, 35, 36].

Further, despite the evident disruption of sport analytics due to technological advances, questions remain about whether sport analytics has reached a level of maturity, understanding and recognition within the sport industry [8, 37]. Cynics contend that sport analytics is yet to become a mainstream division of sport [13], that its application is inconsistent across various sports [38] and that it comes with risks and ethical concerns attached [14, 16, 39]. Some risks of Big Data in sport include athlete data and the control over their data, the accuracy of data, user privacy, and the interpretation of data, leading to concerns that the advances in sport research may be impacted by ethical considerations [39]. In some cases, concerns have been raised about mistrust amongst athletes about increasingly detailed data collection and who owns and controls the data [40]. Further, it is argued that the objective of teams to look for every competitive advantage means collecting data from as many sources as possible, which leads to ethical questions about what data may be forthcoming, including sensitive medical data relating to athletes or through hereditary issues, even their extended family [16]. Overall, scholarly discussion on literature that analyses the application of Big Data to test theory and answer research questions is limited [14], and there are suggestions that a collective fascination with sport analytics obscures the complex nature of organizational objectives to gain a competitive advantage and adds to inequalities in sport [15].

Notably, when considering the predictive nature of sport analytics, Hatfield et al. [23] adds that prognostic statistical methods are a matter of chance. The application of analytics in sport may not provide a reliable relationship between past performance and future performance or measure the competitive advantage for sport teams or participants; thus, there is a risk of ill-informed sport-related decisions. The alchemic aspect of a predictive application to sport is founded on the argument that the analytics involved in competition, team, or athlete analysis thrive on odds versus probability. From a business perspective, Szymanski [13] adds that organizational self-interest and the pursuit of competitive advantages limit the growth of knowledge, observing that “however much like alchemy, sport analytics is characterized by opacity and secrecy, and outside of baseball, evidence of success that would meet the usual scientific criteria is limited” [p. 57]. Corporate objectives for sport organizations may be founded initially on on-field success, but it is the advent of analytics linked to, for example, improved marketing functions and sponsorship processes that mean sport organizations benefit from an increasing array of commercial opportunities through data analysis [18, 19, 20, 41, 42].

Yet, in contrast, the discipline’s importance continues to grow with a rise in its use by sport organizations [17, 43]. Therefore, coupled with earlier successes, advanced automated data collection processes, and increased computational power, sport analytics has evolved into a significant division in the sport industry. From simple projections of performance and predictions for future competition, the use of sport analytics has grown over time, allowing for helpful insight into athlete and team achievement [10, 28]. Many advocates of sport analytics argue that it represents a straightforward way to improve athlete and team performance, particularly from a competitive perspective. There are examples of such applications in rugby [44], cricket [45], NFL (or American Football) [46], and soccer [7, 47, 48] where there are cases that demonstrate that sport analytics aids in the tracking live sport performance, perfecting athletic movement, and virtually eliminating injuries [49]. Further, there are instances where sport analytics is applied to an athlete’s health and the probability of an injury which enhances the general performance of a sport organization [23, 50]. Finally, there have even been suggestions in soccer that teams have won league titles with the aid of sport analytics [51].

Numerous platforms, including websites, television channels, and other digital avenues, are dedicated solely to sport analytics [45]. Increasingly, sport analytics is a more structured field, and research findings support the increasing levels of application of complex data within sport organizations. According to Steinberg [52], referring to sport in North America, “today, every major professional sport team either has an analytics department or an analytics expert on staff” [para. 3]. Further, data-driven decision-making means sport analytics is now a significant focus for the entrepreneurial space in the sport industry [53, 54]. Whether sport analytics is a science or a pseudoscience, however, the ultimate standard of the discipline is the capacity to predict with significant statistical accuracy [1, 29, 30, 31]. Accordingly, it is necessary to delve into a discussion supported by literature outlining the position of sport analytics in mainstream sport and whether it is now a science rather than alchemy.

3. Literature review

This review explores examples of literature on sport analytics to demonstrate if the discipline has become mainstream while simultaneously opposing the view that it is alchemy. In particular, predictive statistical methods, or forecasting [55], are evaluated to find a reliable relationship between past performance and future performance and to see if that renders sport analytics as science or alchemy. The review also examines the functions of sport analytics and its role in aiding the competitive advantage of sport organizations and the relationship with interconnected organizations, employees, and fans, that collectively help drive performance and profitability [56]. The key sections address statistical elements, performance optimization, theoretical frameworks, and the application of sport analytics.

3.1 Statistical elements

Studies confirm that sport analytics is a progressive discipline with immense growth potential [17]. The field continues to explore new ways to incorporate statistical methods for improving predictive accuracy [10]. Kapadia et al. [45] recognize the practice noting that “sport analytics is a promising research field which involves deriving valuable information about the game, based on past games played, or even games in progress” [p. 2]. Such a position implies that using statistical methods to evaluate and improve an athlete’s or team’s performance to assist stakeholders in managing an athlete’s health (e.g., the probability of injury) gives sport analytics significant legitimacy. Sport analytics derives significantly from mathematical and scientific methods [48]. The idea that it is alchemy – “characterized as the fruitless attempt to turn straw into gold” [13] – is far from reality. Groll et al. [57] proposed that the scientific application in the analytical process should qualify the discipline as a science. Furthermore, within the discipline of sport analytics and the development of technology, the domain provides evidence of engagement between various people ranging from analysts, presenters, sport betting sites, and companies together with their employees and fans [17, 56, 58].

Some studies observe that sport enthusiasts, participants, and fans use information derived from sport analytics to better understand a team’s performance with sufficient accuracy [59, 60]. With technology taking center stage, it is evident that sport analytics incorporates multiple technological applications to improve its statistical precision [6, 58]. According to Cortsen and Rascher [18], predictive analytics has been used successfully in many other sectors of the economy to foster commercial progress while, at the same time, improving business performance [3]. Such situations affirm the position of predictive analytics, thus, the potential for sport analytics to make accurate predictions.

Applying analytical methods to Big Data and combining them with predictive algorithms yields superior performance in sport and other industries [61]. For example, Szymanski [13] points out that sport analytics is characterized by Big Data, advanced statistical methods, and complexity, illustrating how sport analytics combines with technology to improve accuracy, efficiency, and interactivity. Goes et al. [48] contend that sports analytics involves using Big Data that requires complex statistical synthesis to harness its potential. Szymanski [13] notes that Big Data is revolutionizing several aspects of sport, based in part on increasing levels of accessibility with “the development of open-source software platforms such as Python and R, together with the increased availability of sport data online” [p. 60]. This revolution includes various applications, from genomics research to business [62, 63]. In business, companies can track a customer’s purchases, forecast the product they are likely to purchase and then present them with the opportunity to make such a purchase [20]. Accordingly, this development indicates that using Big Data in professional sport offers the potential for a competitive advantage that will appeal to every sport organization.

There are examples of sport organizations creating brand value and obtaining a competitive advantage in sponsorship and purchase predictions [41] and clubs and teams that gain a competitive advantage in the field of play [1]. Coupled with the recent advances in computational power, Big Data has paved the way for sport analytics to become mainstream. This development partly relates to the fact that sport organizations are moving beyond low-level intervention-based observational data [64] to predict performance as increasingly complex data becomes available, permitting more refined analysis [7]. For example, in statistical analyses of soccer, game logs – files of soccer matches – are “obtained through next-generation tracking technologies and physiological training data collected through novel miniature sensor technologies have become available for research” [7]. By encoding game logs, there is the potential to predict when an opponent will perform strategic actions like pass prediction [65].

Mumcu and Fried [66] point out that sport administrators can use sport analytics datasets to identify patterns suggestive of an athlete’s or team’s competitive advantage. For instance, some analysts have demonstrated that athletes consistently display the same reactions in competitions, meaning sport organizations can gauge an athlete’s resilience in certain circumstances [67]. By providing interactive prediction through situational analysis [28], sport analytics offers an off-field element linked to commerce and organizational profitability [56]. For example, some athletes are valued for endurance, some for speed, and others for style, skill, and how they blend in a team setup. Hence, sport analytics goes beyond predicting outcomes, as Szymanski [13] suggests, because it also includes using various metrics to evaluate talent, build teams and accelerate engagement with fans, which has a significant commercial element [17, 56, 68]. In other words, the statistical elements associated with this technique apply to both on-field and off-field simulations.

Kostakis et al. [69] refer to the application of scientific methods in data mining and the use of algorithms that makes sport analytics a mainstream sport division. Numerical algorithms and machine learning reduce the margin of error, thus increasing the accuracy and efficiency of sport analytics [45, 70]. According to Singh [17], such technology supports the business aspect of sport and allows organizations to “capture and collect data on games, bidding, bookmaker odds, playing styles, scores, and many other sport attributes” [p. 64]. Thus, using mathematical methodologies coupled with different prognostic statistical methods provides reliable relationships between past and future performance, further qualifying sport analytics as a science. While Szymanski [13] argues that “much of the work in sport analytics seems to possess the same traits as alchemy” [p 60], the continuous application of sport analytics and its ability to derive accurate outcomes overtime outweighs the general view that it is secretive and obscurant.

Finally, sport analytics has developed to become an integral part of the mainstream sport, evidenced in part by what has been referred to as the ‘Moneyball moment’ [13, 71], a story from American baseball involving the novel application of data analytics to gain a competitive advantage. Developments since the Moneyball moment have resulted in various sports applying algorithms to guide business decisions for athlete assessment [72] and team success [51]. These advances include the entrepreneurial application of sport analytics, witnessing millions worldwide participating in statistics-based imaginary sport leagues such as Fantasy Football [68]. Thus an approach to the business and marketing of sport, based on the application of data, is potentially of immense value [3, 17], and in this context, the disciplines’ validity warrants acknowledgement.

3.2 Performance optimization

It is essential to note that sport is a highly competitive and performance-oriented engagement. According to Hatfield et al. [23], sport analytics’ essence is to optimize performance, as participation in any sport equally needs a performance-based strategy. Hence, sport analytics provide relevant statistical measurements to improve a team’s performance and examine sport participants’ athletic capacity [73], the probability of incurring an injury [23, 49], and its impact on the team; sport analytics makes the approach sport management intelligent [13]. Since sport is a viable business and one of the leading occupations, sport analytics’ role enables teams to perform at the peak from a managerial perspective [3].

In advanced sport management, sport analytics involves profound elements of human kinetics, performance psychology, cognitive-motor, exercise psychology, sport analysis, and performance history. An important aspect of sport analytics is psychophysiological monitoring, which is used to obtain “a better understanding of the processes underlying athletic performance and to improve it” [74]. Therefore, sport analytics cynics may have more to lose from failing to recognize the essential analytics function in planning for better team and individual athlete performance [73]. That being the case, what becomes paramount is sport analytics’ capacity to predict with substantial accuracy [1, 29, 30, 31].

As an organizational objective, superior performance in sport is characterized by better decision-making informed by consistent practice, having the right resources, and the best coaching. While some instances may require coaches, athletes, or teams to explore their intuition, it does not mean, as Szymanski [13] suggests, “that almost all-important insights from science can be explained intuitively” [p. 60]. For example, Szymanski [13] highlights the importance of Big Data in business decisions, including predicting consumer behavior. Hence, all-important insights heavily rely on “the science of recognizing patterns on a grand scale” [13]. As such, all-important decisions today rely on data patterns rather than intuition.

In elite sport, high-performance sport managers regularly explore and analyze data patterns in the past to improve future athlete performance [73]. Doing so demonstrates that prioritizing statistics as part of professional decision-making is gaining widespread acceptance [52]. Various sport stakeholders use sport analytics for multiple purposes, such as improving the business’ bottom line – “the pursuit of profit” [13] – but coaches, athletes, or teams need the information to optimize performance. For instance, the data derived from the analysis can create profiles for athletes and their teams which may act as a point of reference for past, present, and future performance [17]. Thus, any professional sport team, coach, or athlete that does not use the insights derived from sport analytics fails to exploit the competitive advantage effectively.

Apart from the athlete and team, sport analytics evaluation includes creating a team’s strategy and managing competition. Part of this involves the financial aspect of sportsmanship, driving customer engagement, and augmenting back-office intelligence. Statistical modeling has become an integral aspect of business [9], spreading into other essential business functions such as marketing, administration, member retention, and expanding partnerships [42, 66]. For instance, since most sport now depends on sport analytics insights, sport organizations can attach the correct value to a team, the athletes, or even leagues or tournaments [73]. The same information is used in ticket pricing, identifying, and signing new athletes, and attracting partners and opportunities for investment. Consequently, the position and function of sport analytics depend on various stakeholders’ ability to put the information gained into meaningful use with sufficient accuracy [1]. An example of the meaningful use of sport analytics occurs when different sport platforms such as websites, television programs, and digital avenues adopt the science to advance commercial aspects of sport.

Coupled with the growth of the computational capacity of the most advanced computer systems, the development of open-source software platforms and the availability of sport data, sport analytics is increasingly part of the contemporary mainstream sport culture [75]. At the organizational level, some activities related to sport analytics include recording matches and training sessions, analyzing each athlete’s movement in the team context, delivering feedback, and updating databases for further analyses [76]. Such processes are typical of any business seeking to optimize its market performance, usually coined market research and analysis. Therefore, sport now involves using analytical tools to reduce risks and identify the best in a pool of talent, thus ensuring that every decision offers value to a sport organization [10, 77, 78].

Finally, sport organizations are increasingly aware of integrity issues relating to sport analytics combined with their business objectives. Critics dismiss sport analytics labelling as a practice motivated by the need to influence coaches and gain influence over competitions and teams, discouraging sporting integrity and thus hindering the true spirit of sportsmanship. While some ethical concerns are tied to sport analytics [16, 39], there is the need to appreciate it as an insightful professional tool rather than dismissing it based on its negative aspects, a characteristic synonymous with most innovations. Alamar [79] notes that “as data becomes more accessible, decision-makers have found clearer insight into their organizations and the nature of the decisions they face through the use of metrics that did not exist even a few years ago” (p. 65). Within that context, data collected through the sport analytics process is subject to legislation, including, for example, the Privacy Act 1988 (Privacy Act) in Australia and international legislation such as SOC 2 (Service Organization Control Type 2) cybersecurity compliance frameworks designed to ensure that third-party service providers (e.g., sport data analytics companies) store and process sensitive data in a secure manner [80].

3.3 Theoretical frameworks

Any form of statistical synthesis requires the qualitative aspects of analysis to foster understanding and comprehensive application [81]. The qualitative aspects of sport analytics explore theoretical frameworks to make sense of the complexity involved in Big Data analysis to derive value [1]. Thus, theory becomes an essential part of scientific investigation, which helps solve complex questions regarding sport analytics’ history and whether it has a place in mainstream sport. There are suggestions that “research in sport analytics avoids theorizing” [13], yet studies discussed herein are evidence of the research efforts associated with sport analytics. Hayduk [3] argues that sport analytics continues to attract sport enthusiasts and stakeholders who rely on the analytical nature of the practice based on existing theoretical concepts from different disciplines.

Goes et al. [48] contend that “contributions from the domain of sport science and the domain of computer science are typically characterized by distinctly different research paradigms” (p. 2). Specifically, sports science often involves deductive reasoning to form hypotheses and experimental design studies. For example, teams are complex dynamical systems, and as such, a hypothesis concerning group or individual behavior is illuminated based on theoretical perspectives such as deductive reasoning [82, 83]. On the other hand, the computer science domain utilizes a very different research paradigm than the sport science domain. Gudmundsson and Horton [84] maintain that computer science involves the theoretical underpinnings of information retrieval and advanced analysis, thus realizing high levels of complex and large data representations. Therefore, combining the domains points to science in practice and counters any notion of a lack of theoretical application in sport analytics. Regardless of the theoretical frameworks used to justify sport analytics, the ultimate benchmark of sport analytics is the capacity to predict [1, 29, 30, 31].

The predictive analysis involved in sport analytics is characteristic of ‘complex systems theory’ – which exploits scientific methodologies such as computational modeling and simulation to understand or predict social issues [85]. Typically, a complex system is a structure made up of multiple components that how they interact [86]. On that basis, McCarthy et al. [87] suggested that it can be challenging to understand sport analytics’ complex nature. Accordingly, it is conceivable that an aversion to the mathematical application may lead to the rejection of statistical methods used in sport analytics based on the issue of chance. Such dismissal rests on the fact that chance may not provide a reliable relationship between past and future performance [88, 89]. Further, chance may not measure the competitive advantage of sport teams or participants in a complex system. Such systems are fundamentally challenging to model owing to multiple interdependencies, competitive relationships, and other types of interactions between some of their parts or between a given system and the environment in which it exists [90].

Exploring the discipline’s theoretical aspects and statistical elements is essential to generate a balanced understanding of the application of sport analytics [91]. Some skepticism about crucial components of sport analytics arises from the still-developing links with performance history, cognitive motor skills, performance psychology, and human kinetics [12, 92, 93]. Further, the attribution of doubt is limited to understanding exercise psychology, the ability to analyze Big Data, and the complexities of sport analysis in a statistical manner [94]. Since aspects of the fundamentals of sport analytics may appear unrelated, the interconnections that are difficult to trace have become one of the grounds for discharging sport analytics as a legitimate area of the sport industry [13]. Therefore, the discussion about the position of sport analytics in mainstream sport and whether it is a science or alchemy should consider the link between the theoretical and statistical aspects of predictive analytics [13]. There is a need to improve the illustrative relationship between the qualitative vis-a-vis the quantitative elements of sport analytics to foster a balanced understanding of its application as a tangible division of sport [95, 96].

3.4 Application of sport analytics

To ascertain the position of sport analytics as a legitimate branch of sport, it is crucial to assess multiple sources of information ranging from analytics provided by in-house departments of sporting divisions to evidence from sport analytics companies and organizations that sell sport data. Further, there is a need to arrive at evidence-based conclusions against the ongoing discussions regarding the position of sport analytics as a mainstream sporting practice and whether it is a science or alchemy. Companies such as WyScout, Opta, FiveThirtyEight, Sportradar, and Diamond Kinetics sell sport data based on statistics relating to playing styles, bookmaker odds, competitions, and results [17, 55]. Several platforms, such as websites, television channels, and digital avenues, are dedicated solely to analyzing competitions and teams or athlete performance [45]. According to companies such as WyScout, the information derived from sport analytics readily applies to the actual management of teams and clubs with high accuracy.

Organizations like WyScout sell sport data and provide information used in several fields by various sport stakeholders, including analysts, presenters, and fans. Some companies provide real-time information about competitions and teams to manage and commercialize fan expectations and experiences before and during competition [56]. Such predictions are possible due to the data supplied through aspects of sport analytics [97]. Data that companies provide include seasonal datasets and historical data. Within that context, betting on the performance of individuals or teams, or the outcome of competitions, is evidence of the commercial application of sport analytics. Indeed, sport betting is an emerging trend in sport analytics [98] and demonstrates that data-driven decision-making is not only a reserve for high-profile sport management as sport fans benefit from applying sport analytics to inform betting strategies [53].

More evidence about the application of sport analytics in sport management derives from companies that provide analytics data and platforms to help organizations improve performance. Professional sport purchases such data and platforms because they offer valuable insights into, or management of, coaching processes and teams and athletes. Some of the most recognized companies in this category include Covatic, Edj Analytics, AISpotter, Fusion, Catapult, EDGE10, Sportradar, MacVar, and StatsBomb. The information derived from such companies is based on technology that synthesizes information – a data-driven ecosystem [99] – about athletes and competitions. Capturing athletes’ past and present performance is critical for planning future competitions. Armed with such information, it becomes possible to attach value to sport teams, athletes, or events. For example, some athletes are valued for endurance, some for speed, style, skill, and how they blend in a team setup.

Further, technology provided by analytics companies has been assimilated into sport analytics to improve accuracy, efficiency, interactivity, and team performance [100]. For instance, a company such MacVar exploits simulated fields that clone positions of athletes wearing smart jerseys in a match with finite levels of accuracy to analyze athlete movement [101]. Athlete profiles and team performances have become leading sources of information for coaches and sport managers. Various sport stakeholders use sport analytics for various purposes, but coaches, athletes, or teams need the information to optimize performance. Sport analytics provides details such as the position of a ball, work rate, fatigue, the time of a specific action, and the results of such action. Software firms such as Coach Paint, Dartfish, and Sportscode, coupled with third-party data and statistics companies such as Opta, have made it possible to collect, retrieve, code, capture and analyze essential data points efficiently and effectively for real-life applications. Applying statistics as part of professional decision-making is gaining wide acceptance as a mainstream activity [102], no less so in sport due to sport analytics’ numerical value [6].

From a sport organization perspective, many teams also have internal analytics departments that provide data to inform decision-making. In-house analytics departments offer greater precision than data derived from sport analytics companies [103]. Amidst considerable reliance on sport analytics supported by Big Data, some teams find it helpful to set up internal sport analytics departments that enable them to own such vital information. The internal departments use technologically supported models and software to break down complex data matrices [1]. In-house data analysis is often more precise and tailored – the information is more contextual [104]. Football clubs, Australian Football League clubs, and American baseball teams value the sport analytics resource and continuously invest in internal analytics departments. This development has created roles such as tactical analyst, technical scout, scouting analyst, training analyst, goalkeeper analyst, and research analyst, individuals increasingly viewed as sport scientists [105]. Sport organizations are investing significant resources towards building these departments that strictly focus on robust metric and statistic systems, and analysts are increasingly valuable recruits for such structures [106]. Therefore, sport analytics is an increasingly crucial scientific venture supporting team and club performance within that context.

Finally, the application of sport data serves other purposes as data analytics companies seek to monetize data for commercial reasons. Some examples of data produced relate to digital media consumption needs, sport writers, teams, and leagues [100]. Sport organizations can use data that tracks online sentiment and behavior [54], which gives rise to a new and advanced approach to customer relationship management [107]. Further, from the perspective of selling to fans away from competition, data available through sport analytics enables the creation of gaming models prevalent in games such as Fantasy Football [100]. The variety of applications demonstrates multiple creative ways in which sport analytics is applied [100]. Analytics companies recognize sport as a serious business that can inform strategic management and intelligent planning from a data perspective [3]. As most sport teams and clubs continue to put information derived from sport analytics into meaningful use, sport analytics’ position as a mainstream factor becomes more challenging to deny.

4. Observations

Various authors opine that sport analytics is yet to reach a level of maturity, understanding and recognition within the sport industry [8, 37], is yet to become mainstream, appears secretive and obscure, and is characterized as alchemy [13]. Other authors suggest the relevance and value of sport analytics is questionable [1], and there is unsettling debate about the position of sport analytics in mainstream sport [11, 32, 35, 36]. There are suggestions that the application of sport analytics is inconsistent across various sports [38], and it comes with risks and ethical concerns attached [14, 16, 39], and that the development of the field adds to inequalities in sport [15].

Nevertheless, based on the recent advances in sport and computer science, Big Data and the capacity to analyze data, there is robust evidence that sport analytics is meaningful, impactful, and widely applied to great effect in the sport industry. Further, numerous developments in science and technology related to sport analytics assist with organizational decisions, from a team or athlete performance basis to strategic planning and management. Such developments are based on sophisticated data collection methods, storage, analysis, and the meaningful presentation of these data. With the growing impact of Big Data in sport, analytical skills have improved and become paramount to and synonymous with the sport industry, especially considering the large amount of data handled, disseminated, and analyzed to generate valuable insights. Evidence that significant insights have developed to improve team, athlete, and club performance is portrayed in several examples featured herein.

While there are situations where statistical methods and algorithms used to forecast performance might be closely guarded details for a sport organization, it is essential to remember that the discipline was developed specifically to help such organizations drive team efficiencies in addition to revenue and profit generation [108]. Where profit – the bottom-line – is involved, it is impossible to perform these analyses in public, let alone share the models and algorithms used in the analyses. In so doing, there would be a threat to the competitive edge of analytics firms, clubs, and sport organizations. The increased exposure and use of sport analytics in sport, however, indicates how data analytics has been transformed into a mainstream component of the sport industry. Analytical tools and the availability of Big Data have improved sport organizations’ ability to evaluate team performance and identify both team and individual strengths and weaknesses. An examination of the role of data analytics in sport points clearly to the discipline being a science, and, in this respect, the argument that sport analytics is notorious or alchemy is flawed.

5. Future direction

Whilst not obfuscation, protecting one’s competitive edge is logical. In assessing the development of sport analytics, however, there is a sufficiently robust platform that demonstrates those involved in the discipline are directly or indirectly adding to knowledge. Although sport analytics research is in a nascent stage [17], an increasing body of research is available to assess how data-driven analytical methods influence sport [2, 14, 109]. Quantifying such a situation – and challenging any notion that sport analytics lags in the shadows of alchemy – is an area for future research. Such research might include gaining insight into the burgeoning data analytics organizations providing data and platforms for professional sport teams, the success of sport teams as they rely increasingly on sport analytics, and the impact of sport analytics on individual athletic performance. Although there will always be organizations that seek to limit knowledge sharing to protect their competitive advantage, the cumulative impact of sport analytics in mainstream sport, as evidenced herein, means future research has abundant data to apply to the science of sport analytics. Indeed, analysis of large datasets will provide a more comprehensive understanding of ever-changing phenomena [14], increase knowledge in sport science, inform competitive strategies, and facilitate innovative research [110].

6. Conclusion

In conclusion, the increasing demand for sport analytics is primarily because of the available data (i.e., Big Data), which has increased exponentially over the last decade. The radical advances in information technology increased computing power, the amount and frequency of data collected, and reduced storage space costs are more far-reaching today compared to some years hence. Sport analytics has become a new discipline in the sport industry, enabling a new level of efficiency and competitive advantage [52]. Most business models related to this field aim to improve the performance of individual athletes, teams, clubs, and sport organizations while at the same time helping achieve corporate objectives, particularly the bottom line. Therefore, the profitability aspect, like in any other business, makes it challenging to find sport analytics companies or sport organizations benefiting from the data output of said companies, readily sharing their algorithms and predictive models. While this may limit the pursuit of knowledge in one aspect of sport, it does not demonstrably impact the potential for datasets to be used in a more open, beneficial way for sport science. More importantly, it does not suggest that alchemic traits of notoriety, opacity, and secrecy characterize sport analytics.