User:Teratix/Australian rules football analytics

In sports, analytics is the practice of collecting and using data to gain insights into a player or team's performance. Analysts of Australian rules football, dubbed "the most data-rich sport on Earth" in James Coventry's Footballistics, have gathered simple statistics on the sport since the 1930s. The collection and analysis of this data has become more sophisticated over time, and analytics is now commonly employed by club personnel, media commentators, and fans.

Australian rules football features two teams of 18 players each, who compete for the highest score by kicking an oval ball through their team's goal posts to score points. Players can kick, handball (punch), and run with the ball in order to move it across the oval-shaped field. They can be tackled by their opponents at any time, except when they have marked the ball (caught the untouched ball on the full from a kick), received a free kick from the umpire, or during a stoppage of play.

Analytics is predominately used in the premier Australian Football League (AFL), which has employed Champion Data as its official statistics provider since 1999. Many AFL clubs employ analysts to seek a competitive edge over their rivals, whether on a consultative, part-time or full-time basis. Recruiters also use analytics when scouting potential draftees in elite junior competitions, such as the AFL Under 18 Championships and the NAB League. Fans can access analytics through sports commentary, media reports, statistics websites and apps, and fantasy football. However, Champion Data's advanced metrics cannot be viewed without paying a large fee, making it impractical for fans outside clubs and media organisations to access them.

Simple statistics – scores, kicks, marks, free kicks and successful passes – were published for certain Victorian Football League matches in newspapers such as The Sporting Globe as early as the 1930s. However, comprehensive, weekly statistics for every player and match would not appear in print until the publication of Harry Beitzel's Footy Week from 1965. New statistics organisations such as Ray Young's APB Sports and Gundars Mantinieks' Prowess Sports were established in the 1980s and began to employ computers, rather than pencil and paper, to assist in recording, storing and transmitting statistics. In 1999 the AFL engaged Champion Data as its official provider of statistics; the company had been founded three years earlier by ex-Carlton player Ted Hopkins.

The publication of Michael Lewis's Moneyball in 2003, examining the successful analytics-based management of a Major League Baseball team, provoked inquiry into whether the approach could be replicated successfully in the AFL. In response to coaches' desires for more sophisticated measures of performance, Champion Data began to provide more complex metrics which incorporated players' position on the ground and the influence of pressure, aided by the increasing use of wearable GPS tracking units to follow players' movements and speeds.

1930–1964: Early statistics
Statistics were infrequently recorded during this early period. In the 1930s, Dave Stewart, ex-Richmond vice-president, published tables of VFL players' statistics in The Sporting Globe for one match per week. Stewart noted players' kicks (including a separate metric for successful kicks to teammates), goals, behinds, shots for goal, marks and free kicks (both received and conceded). Ben Collins of the AFL Record, writing in 2016, noted Stewart's tables displayed a "surprising level of detail", although they were not as exhaustive as modern-day statistics. In the 1950s, newspapers including The Sporting Globe (from 1953) and The Age (from 1958) published basic statistics for certain VFL finals matches.

1965–1995: Footy Week and APB Sports
In 1965, Harry Beitzel founded Footy Week, which became the first publication to report individual player statistics from every VFL match in every round. Beitzel, a radio broadcaster and former VFL umpire, was motivated to publish statistics due to fans' interest – Collins, writing in 2015, described the magazine as "cutting-edge", and Ted Hopkins, founder of Champion Data, dubbed

1996–present: Champion Data, Moneyball and the ascent of analytics
In 2003, the American author Michael Lewis published Moneyball: The Art of Winning an Unfair Game, which became popular reading among AFL coaches. The book examined Billy Beane's analytics-based approach to managing the Oakland Athletics, a Major League Baseball team, in an attempt to achieve on-field success without the large budgets enjoyed by wealthier sides such as the New York Yankees. Glenn Luff, a senior Champion Data statistician, cited the influence of Moneyball as the reason why the company, hitherto "purely data collectors", became "analysts".

Clubs

Collection of statistics
Champion Data, as the AFL's official statistics provider, has a team of ten people recording and transmitting metrics live from every AFL match. Five members work from the match venue; the other five work from Champion Data's headquarters, known as the "bunker", in Southbank, Melbourne.

Personnel at the venue include the main caller, who follows play through binoculars and provides a running description of players' actions; a "support/IT" assistant, who listens to umpires' calls and "helps out the media"; an assistant who tracks players' match-ups – their position and direct opponent; and an interchange assistant, who records players' movements on and off the interchange benches. Staff in the Southbank "bunker" include the back caller, who checks the main caller's description for errors; the "keyboarder", who inputs simple statistics through a keyboard (entering around 5000 per match); the pressure caller and pressure capture, who record advanced aspects of play, including how much physical pressure players face, which foot they kick with, and where they attempt to kick; and the graphical operator, who marks the location of every possession on a virtual field.

By fans
A dedicated community of fans maintain blogs and websites discussing analytics in the AFL. Some have created models to predict the outcomes of AFL matches. These models were highlighted during the suspension of the 2020 AFL season due to the impact of the COVID-19 pandemic; Max Barry, an Australian author, organised a simulated season on his AFL analytics website Squiggle with the aid of several other statistics enthusiasts, to compensate for the lack of actual AFL matches.

Other bloggers have created new metrics for Australian rules football – an example is Robert Younger's xSco metric, which measures a team's expected score based on the quality of its shots, taking into account type, angle, and distance from goal. The metric allows teams to be judged on the quality of chances players create, rather than merely those which successfully impact the scoreboard. Younger, who publishes the analytics blog Figuring Footy, adapted the concept from soccer and ice hockey; the equivalent metric in these sports is expected goals, or xG. Port Adelaide later hired Younger as a full-time analyst in the club's football department.

Analysts outside of clubs and sports media have been hindered by restrictions on access to Champion Data's advanced statistics. Jack Howes, writing for The Guardian, described the company's undisclosed fee as "exorbitant", which "lock[s] prospective analysts out of the loop", and unfavourably contrasted public AFL statistics with those available for sports such as American football, baseball and basketball. Craig Little, also of The Guardian, acknowledged the difficulties for external analysts caused by the dearth of "interesting, worthwhile numbers", but also pointed out Champion Data's for-profit nature, dubbing openness a "lousy business model" in the small Australian market.

Player ratings
Player ratings measure individual performances in a single quantitative metric. Developing such ratings is particularly difficult in Australian rules football, because players adopt a wide variety of roles and positions, their actions seldom directly lead to quantifiable results (in contrast to other team sports such as baseball), and interactions between players are often complex. Nevertheless, several such systems have been proposed for the sport.

One popular system is the AFL Player Rankings, developed by Ted Hopkins of Champion Data and Stephen Clarke from the Swinburne University of Technology, which identifies the most important player actions (such as kicks, handballs and marks), and assigns them a certain number of points. The points a player accrues over the course of a match are combined to determine their final score. A version of the system, revised in 2002, became the foundation of AFL SuperCoach, an online fantasy football competition. About 50 variables are considered when determining players' scores.

In 2013, the AFL released an official alternative known as the AFL Player Ratings, based on a system created by Karl Jackson, chief statistician at Champion Data. Players gain or lose points based on their impact on a passage of play – however, how much pressure they experienced and their location on the ground are also taken into account, making the system among the most sophisticated employed in any team sport. However, it can only be used for AFL matches because the data required to calculate a rating is so complex, which limits how useful the system is for recruiters examining lower-level games.

Unofficial ratings systems have also been produced – for example, Cody Atkinson and Sean Lawson, who maintain the AFL statistics blog Hurling People Footy, published a system known as Player Approximate Value (PAV) in 2017. Unlike Champion Data's ratings systems, which reflect the quality of an individual's performance, PAV measures players' contributions to their team's overall performance. Separate ratings are calculated for players' contributions to defence, midfield and the forward line, which are combined to create a PAV.

Champion Data's ratings systems are relatively limited in their application to historical matches – AFL Player Ratings can only be determined for matches from 2010 onwards, because pressure-related statistics had not previously been collected, while AFL Player Rankings cannot be calculated for games prior to 2002. In contrast, PAVs have been calculated for matches as long ago as 1988.

GPS analytics


AFL players' location, speed and movements are frequently monitored by GPS tracking units, worn in a pouch between players' shoulders during training sessions and matches. The data collected can be employed by clubs to track players' performance, potentially indicating whether they are overworked or underperforming; and inform match tactics, such as positional changes and substitutions.

Players were first permitted to wear tracking units during AFL matches in 2005; for a total of ten games, teams were allowed to track five players. By 2009, this number had increased to ten players.

The Conversation AFLPA ZDNet Herald Sun ESPN ESPN 2 The Age news.com.au Fox

News articles/guides
Vic Library The Guardian The Guardian 2 The Roar (Expert) AFL.com.au 1 AFL.com.au 2 AFL.com.au TAC Cup AFL.com.au timeline source Champion Data official definitions ESPN The West Fox Sports AFLW stats (AFL Record round 16 and 19, 2011, p. 78) The West 2 SMH (AFL Record July 2-5 2015)

Journal articles
(double-check for reliability) International Journal of Sport Finance Journal of Science and Medicine in Sport International Journal of Sports Science & Coaching PLOS ONE