Written by Zac Pross, Data Scientist at Fusion Sport
The Australian Sports Tech Cup Hackathon
Imagine you are an analyst for a newly formed AFL franchise called the Fusion Sport Fury. You’ve just been emailed several years’ worth of AFL data. One dataset contains AFL Draft Combine Testing Results while another dataset contains Champion Data’s AFL Match Ratings, where the same players in the Combine dataset are in the Ratings dataset.
The head scout, head sport scientist and head coach are expecting a presentation on one of the following problems:
“Can you tell us whether AFL Combine results are good predictors of performance in the AFL, as measured using Champion Data’s ratings data?”
“Can you design a series of dashboards that allows the head coach to dissect the data and find a new midfielder?”
You have 24 hours to prepare. Good luck!
That was the hypothetical situation presented to 50 participants (split into teams) at the inaugural Sports Tech Cup Hackathon sponsored by Fusion Sport and held at the La Trobe University City Campus on November 24th and 25th of last year.
Spending their weekend celebrating a common love of data science, data visualisation and the AFL, the teams worked non-stop (give or take a few pizza breaks) until they presented to the judging panel late on the second day.
Three things we took away from the first Australian Sports Tech Cup
1. Combine Performance Probably Doesn’t Predict Performance in the AFL
Most teams who attempted the first problem relayed a message that Combine results are just ‘one piece of the puzzle’. There is more to being an elite footballer than strength, speed and isolated skill metrics.
For example, the AFL Combine does not measure game awareness, coachability or mental resilience, which are clearly important traits found in successful elite athletes.
Recognising that it was a difficult problem and understanding that a null result is still a useful result, the judging panel were looking for approaches that laid bare their assumptions while ‘telling a story’ that a non-analyst could get excited about.
2. Concision Over Clutter
Many teams chose to tackle the second problem – to design a series of dashboards that would allow a coach to find a new midfielder. There are of course different types of midfielders in Australian Football; the aim of this task was to produce a dashboard that could allow coaches to slice, dice and interrogate the data – i.e. to find a player for any position!
Several of the dashboards showed an acute understanding of the types of decision-making tools applied practitioners need day-to-day. Our one criticism was that some of the solutions had too many buttons, too many sliders, too many checkboxes… too many elements in general.
Of course, balancing usability with interactivity and customisability is exactly the ‘art’ of dashboarding. Our advice to anyone who uses dashboards, especially those who enjoy building their own dashboards in Smartabase using the new Dashboard Builder, is that simplicity is almost always best!
3. Analytics is a Team Sport
Data science is interdisciplinary and requires varied skill sets. Cleaning and wrangling data requires persistence and grit. Modelling demands creativity and scientific rigour. Getting people excited about the results needs story-telling savvy.
And yet, if your team doesn’t have domain expertise…
What we found is that the best presentations were delivered by well-rounded groups who had at least one data viz expert, one person conversant in statistical modelling & machine learning and at least one domain expert (read: footy tragic).
…and the winner is
As in all elite sport, there could only be one winner of the inaugural Sports Tech Cup Hackathon.
For their intuitive and visually arresting dashboard design, a deep understanding of the stats and an ability to communicate with purpose, we awarded $1000 to the Mildura Redbacks. Congratulations on a fantastic effort from our winners, and to all our participants.
Join us at the next Sports Tech Cup
Hackathons are not just about nurturing innovative technical solutions, they’re also about learning, connecting and having fun – and we felt the event nailed all four points.
Thank you to the Australian Sports Technologies Network, La Trobe University, Champion Data and Launch Vic for a fantastic weekend.
Look out for the next Sports Tech Cup Hackathon that will be held in Melbourne on the 16th-17th of March.
