A Western Bulldogs backed hackathon has resulted in a proof-of-concept model that could help teams find gaps in their opponents’ defenses.
The AFL team made training-match data available to the public as part of the hackathon event – the first to be hosted by an Australian sports club – resulting in the winning ‘dominant regions’ model.
A player’s dominant region is the area of the pitch they can reach before any other player.
“We’ve traditionally used player tracking data more for looking at how far guys have run, and how fast, rather than patterns of play. [Dominant regions modelling] is something that’s certainly not used currently in the AFL,” explains Dr Sam Robertson, Western Bulldogs Football Club’s senior sport scientist.
“The idea being that if a ball was kicked there, to point x,y on the field, who would reach that ball first. Teams defend so well in every team sport now at the elite level, so how do we move the ball to space, to create a scoring opportunity?”
The success of the event, ran in partnership with the City of Ballarat last weekend, has inspired the club to put on more hackathons which are expected to focus on player match readiness and injuries data. The club is also in talks with a team from a different code to organise a large-scale innovation event in Melbourne later this year.
The challenge was one of two at the hackathon event, the other being put forward by the City of Ballarat.
Ballarat will host its first AFL game this year at the recently redeveloped Eureka Stadium, and the winning council hack was a visitor guide tool that highlighted the alternative entertainment that could be found in the city on match-day. Another entry ran analysis on residents who have pet bulldogs and the names they gave them.
The winning overall team will be heading to a Western Bulldogs training session in coming weeks before spending the afternoon with the club’s in-house data analysts.
No hacks, no glory
Sports teams collect huge amounts of data on players and matches, but the ability to gather information is not always matched by the capacity or capability to analyse it. Over the last year more forward thinking clubs have been trialling public hackathons as a solution.
English Premier League soccer club Manchester City held its inaugural hackathon in July last year during which it made exclusive player tracking data available. It repeated the idea last month with a focus on fan engagement.
In September, the NBA in the US invited student statisticians, developers and engineers to take part in its first hackathon. The winning team analysed the difference in team-centric versus individual-based playing styles shown by teams between the playoffs and regular season. They found that teams tend to play more selfishly in the playoffs, to the detriment of their offense.
Football Federation Australia, the local governing body for soccer, is hosting a hackathon later this month in Sydney with three challenges relating to women's football.
By making data available to the public, organisations can benefit from expertise they don’t have in house. While the Western Bulldogs has an impressive data analytics function it doesn’t, for example, hold spatio-temporal expertise like many in the winning hack team.
“We do have our own data analytics guys, but the wealth of data we get now in Australian football is so diverse as well as large,” says Robertson.
As well as solving the challenge set, hackathons have wider benefits to organisers, Robertson adds.
“There’s engagement with our fans and the broader sports analytics community, because there is quite a large amateur community out there and they’re quite involved and active online. It’s an opportunity to engage them. And a little bit of talent identification – who are the good guys out there worth us having a look at. And all the new ideas.”