AFL Analytics: Part Two - The Armchair Analyst
Updated: Mar 18, 2020
In Part One of our look at the AFL Analytics landscape, we focused heavily on elite sporting teams and the media. Both have their challenges in how best to use data to tell a story, although this is slowly improving.
But another group could help to unlock a corner of analytics that we have yet to fully explore. If (and it’s a big IF) they are given enough tools to succeed, and come armed with the right attitude.
They are the Armchair Analysts – our title, not theirs – a bunch of sharp minds who live in a niche area of the football world but have started to command more attention.
Most average footy fans just want a little help tweaking their Supercoach teams or landing their office footy tipping competition. That’s as far as their stats journey goes.
Fuelled by a natural aptitude for numbers, a love of footy and an analytical mindset, these groups have made the best of a limited data offering. Imagine what they could do with a little bit more…
The best example is James Coventry’s Footballistics, a 2018 publication which quickly became the footy equivalent of Mythbusters. Several long-held theories around scoring were tried and tested using actual numbers, with some surprising results. What might have sounded like a decent opinion down at the pub was finally being debunked in a way that everyone could understand. Even those who were ten schooners deep.
Many contributors to the book were the amateur online data analysts who had been working away in the background for many years, plus a couple who had already made the leap from hobbyist to meaningful employment at Club-land. The tide was turning.
Some armchair analysts have no intention of forging a career in a footy club. Others would jump at the chance - there is no wrong or right, each to their own. Importantly, the hidden handbrake for those who would love to graduate from spare-time data tinkering to full-time club employment is attitude.
Humility is critical.
In Part One we covered those already employed by teams, but it applies even more to analysts who are trying to set up a career – if you think you know more than everyone else, you are a terrible cultural fit for a team. Not to mention that jobs in footy are generally split into two parts: one part is specialising in what you’re good at (such as analytics), and the other part is just getting shit done.
Data Tug of War: Public vs Private
The biggest problem facing the amateur analysis community is the lack of statistical information, despite AFL footy being one of the most data-rich sports on the planet.
There are a number of reasons for this which are important to understand.
Commercial realities prevent a lot of the information being made available to the masses. Money talks. Businesses can’t just give stuff away for free. This can get overlooked in the often heated “the fans deserve more” debate. Just because a small section of fans would love to be let loose on all the private numbers, doesn’t mean they should be.
Champion Data provide a comprehensive bureau service for all league stakeholders, one that stacks up against any similar service across the globe. Then the clubs and media pay good money for it (like, lots of money). Gone are the days of team statisticians sitting next door to the coaches box punching out numbers. And the standard “box-scores” in the newspaper have evolved into something far more sophisticated – first it was just just kicks, marks and handballs, now we see pressure acts, metres gained and expected scores.
All of this incurs an operational cost. The labour and investment that goes into collecting all the metrics for clubs and the media (and, by extension, the fans) would surprise many. Unlike the Amazon-backed NFL service or the NBA partnership with SAP, the AFL doesn’t have the luxury of a tech giant on board to offset the overheads. Yet.
But for the footy-mad data nerds, not even the basic data is delivered to them in the right way. Numbers published in the newspaper aren’t all that productive. In the digital age, printed figures don’t help with analysis, they just act as a reference.
Clearly frustrated, Max Barry from Squiggle AFL tried to appeal to Santa Claus in this article just to get something like the scores delivered in a suitable way. And the big fella hasn’t delivered just yet.
Barry was measured when he described the public footy data landscape to us:
“What we need is one place offering up raw numbers in an accessible format, then a bunch of places to package those stats in different ways to suit the needs of fans.”
“I’d love the AFL to make very basic stats – not the kind that Champion Data developed themselves, but simple things like scores and disposals – available to everyone. So if you want to tinker around in a spreadsheet and build a model you can. That would lead to a rush of innovation that would be good for everyone.”
In a competition renowned for pushing boundaries in the sports technology space, the lack of simple public data delivered in a usable format is really surprising. These barriers stifle progress - If these guys didn’t have to resort to writing website scrapers (which break often) and building their own databases from scratch, we could see a whole bunch of extra analysis published with that extra time. The only drawback might be a few misguided takes. They are amateur analysts, after all.
The argument isn’t to prioritise fans over other key stakeholders. But at the moment the hardcore fans aren’t given much at all compared to so many sports leagues across the world. Up until this week, the AFL Website didn’t have any statistical information. They have since launched AFL Stats Pro – with more metrics than before and a fancy new (if a little confusing) layout - but appealing to the broadest possible audience comes with significant challenges.
Arguably, there are better resources out on the internet than official channels, a sign there is significant room for improvement. Most would be familiar with AFL Tables, which has settled footy debates for decades and helped cure the appetite of stats fans and nostalgic types.
Then there is Footywire – an intuitive, old school interface neatly capturing match data (albeit in a copyright grey area) along with things like team selection, Brownlow votes and attendance figures. It would be surprising if analysts, journalists and fans all didn’t hit both of these sites multiple times a week in the peak of a season.
How does a hobbyist site with little in the way of graphic design become some kind of benchmark? Surely we can do better to move the sport forward in this space.
[EDIT: A few clever minds have pointed us towards the FitzRoy initiative, designed for easy scraping of some of the sites mentioned above. We haven't had any hands on experience using the package, hence no mention of it in the original article. But great to see some are using it effectively already for their analysis.
The point still stands though - would love for these kinds of services to originate from official channels rather than the hobbyist sites themselves, for obvious reasons]
Adversity Creates Opportunity
With the very real threat of COVID-19 impacting the season, these side data projects might be the initiatives worth focusing on. Many of us could end up with a hell of a lot of time on our hands, so let’s make it productive!
We’ve already mentioned a simple API with basic footy data would be a good start. But what else?
Could we have a Big Data Bowl type competition similar to the NFL, where the analytics community grabs a whole bunch of data and attempts to solve some complex questions about our sport? Clearly prizemoney might be limited in the current climate, but it might be a great exercise while no new games are being played.
Importantly, the NFL made a point of mentioning that 11 individuals from the 2019 Data Bowl ended up employed by teams or associated vendors. Different sport, different budgets, different attitude to “outsiders” for sure, but there isn’t much downside to throwing it out there in AFL land to see what happens.
The other window of opportunity for any amateur analyst keen to break into the industry lies in the visualisation space. With clubs now continuing their journeys into data, those who are proficient with PowerBI dashboards, R Studio, Tableau and the like are extremely employable, particularly if you also have the ability to communicate well with others and tell a story with the numbers.
But again, without enough genuinely useful figures available for public use, a lot of organic progress is halted. Squiggle AFL’s Max Barry put it simply:
“You might be a hobbyist with a great idea for a visualisation or analysis, but you can’t get the data to try it out. Which means we all miss out on a lot of innovation and are stuck with a small number of mass market products.”
An Uncertain Future
Of course, the biggest thing to halt progress will be the current health crisis. There is no doubt this will bring footy clubs back to their core: coaching, training, playing. Anything else will be seen as a luxury for a long time to come. More on this in another article to come.
So this season becomes more and more about thinking outside the square. Is it the perfect opportunity to release a whole back-catalogue of historical data to the masses? Or at the very least a small sample? The clubs live in the present not the past, while the media tend to do both, but surely the commercial sensitivities aren’t nearly as great for past seasons.
In a time of uncertainty, isolation and potential boredom, the Armchair Analysts might at least uncover something interesting while we all stay safe.