Can statistics predict the AFL Grand Final winner? The last 10 premiers tick these metrics but only one current finalist does too

Over a quarter of Australia’s population will watch this year’s grand finals. Over four million will watch the AFL Grand Final at the MCG on Saturday 24 September, slightly fewer will tune into watch the NRL premiership decider the following weekend.

And among the plethora of fans whose teams are still in contention, they’d surely love some Nostradamus-like ability to know whether their side will take home the silverware.

Plenty of methods exist to try and predict football winners – from the simple act of assessing a form guide, to complicated ranking formulas, to the late Paul the Octopus.

Right now, AFL bookmakers are tipping Geelong – the top team in the home and away season – ahead of Sydney, Collingwood and Brisbane. In the NRL, Penrith is firm favourite (and by some margin) among bookies to defend last year’s title.

When it comes to the stats that really matter when it comes to picking a premier, we went to those who deal with data every day.

Premiers tend to tick four boxes

Since 1996, Champion Data has been responsible for recording every kick, handball and mark in the the AFL (since then its client base has expanded to Super Netball, horseracing, NRL and the semi-professional Aussie Rules leagues).  

And in the vast library of data recorded every week for six months of the year, there are four metrics that narrow down who might win the flag.

The first two are defensive statistics.

“When you’re looking back at what the premiers have done over the last ten year period, basically, without a strong defensive profile, you haven’t been able to succeed, or give yourself a chance at winning a premiership at the end of the year,” says Champion Data’s lead AFL analyst, Daniel Hoyne.

Defensively, premiers tend to rank in the top six for the ability to prevent opposition scores.

That probably doesn’t come as much of a surprise. ‘Points Against’ is one of three ranking metrics used to calculate the AFL ladder (the others being ‘Points For’ and competition points for winning or drawing a game) – it’s pretty important at determining your ladder position.

This year, the best six teams for preventing opposition scores were Melbourne, Fremantle, Geelong, Sydney, Port Adelaide and Carlton. Among this group, only Geelong and Sydney remain in contention – each playing a home preliminary final this week.

A more sophisticated stat than a raw ‘points conceded’ metric, is ‘score conceded from a turnover.’

Hoyne says each of the last 10 premiers have finished in the upper third of the competition for this metric as well. At the end of the home-and-away season, Geelong, Fremantle, Melbourne, Richmond and Port Adelaide were the top-ranked.

So based on Champion Data’s analysis that the premier usually comes from the top third of these metrics, Geelong fans will be sleeping easy.

Why are turnovers so important?

“We always refer to the turnover number as the most important aspect in terms of how your game is actually going, because 60% of your overall score comes from turnovers,” Hoyne says.

So what about attacking stats?

Unsurprisingly, keeping the ball in the forward half of the field is very useful in every football code where goal kicking is the sole scoring mechanism (as opposed to rugby codes and American football which include territory-gain ball placement). Specific to Aussie Rules, having the ball inside the 50-metre arc is also crucial, as this is where most goals are scored.

According to Hoyne, each of the last 10 premiers also finished in the top six for two attacking differentials: inside 50 and time in the forward half.

Geelong tops the league for inside 50 differentials, and is second for time in forward half.

So the Cats, which chalked up thirteen consecutive wins in 2022 and finished two games clear atop the ladder, are, to borrow a phrase you might hear in the outer, looking pretty bloody good.

How does Champion Data know all this?

At every game of the season, statisticians sit in the ground’s media centre and a ‘bunker’ at the venue, inputting observations into Champion Data’s systems.

Leading the team in the stats box is a main ‘caller’ who narrates – statistically speaking – the live game: reciting players and their action to a surrounding team of note takers who convert these verbal observations into computational inputs.

This system is designed to be as efficient as possible, giving keyboard operators limited, logical pathways to convert witnessed events on the field into a numerical statistic.

In a possession-based game like Aussie Rules, a player has very few things they can do, statistically speaking.

“If you followed the sequences of events in a game, as one event happens, only certain things can logically happen after that,” says Champion Data director, Tim Kelsey.

For instance, a player who obtains possession of the ball can only legally kick or handpass it to dispose. Therefore those are the only options that would be presented to the operator.

There are more complex statistics, describing where the ball is positioned (forward and defensive half statistics), to how teams interact with the ball at particular events like stoppages.

There are staff who log interchange moves, those who track player matchups, one will plot possession locations on a computerised playing field, others are responsible for adding additional tags and time codes to the data that is placed into the system.

Atop this, GPS units fitted to players’ uniforms constantly transmit data back to the stats teams.

All this information in compiled in close to real time, and distributed on-demand to coaches boxes, media agencies and other clients.

It’s used by club analysts and coaches to plot their next moves on the field, broadcasters increasingly use the data to add an important additional layer of information to the consumer experience.

There’s up to a dozen Champion Data staff at each AFL game collecting, recording and interpreting data, not to mention the in-house teams working to make sense of the information provided.

In the nail-biting qualifying final between Geelong and Collingwood, there were 421 kicks, 152 marks, 303 handpasses, 155 tackles, 27 free kicks, and 83 stoppage clearances – that’s over one thousand basic events taking place in a game, even before we get to more complex metrics.

Someone needs to see them, and someone needs to record them.

It’s an intense, fast-paced job, and while enthusiasts for the game and the art of stats collection have always been part of the team (among the most experienced Champion Data staff are those from the bygone era of ‘pen and paper’ recording) there is an emerging breed of talent that is finding a home among the numbers people of the AFL.

“We get a lot of success out of gamers,” Kelsey says.

“The actual capturing of games and, in particular, some of the stuff that goes on in a game actually require a reasonable amount of intense ‘button clicking’.

“To do ‘kicking pressure’ capture [a statistic that registers kicks taken under pressure from an opponent], we actually use a gaming console, so it’s generally people who are heavily educated in computer use and have an interest in sport.”

Grand finals rarely dish up the weird; it’s a different story for prelims

Champion Data are in the business of procuring game statistics, and analysts extrapolate these into meaningful interpretations.

When it comes to the stats that win a grand final, the last 10 premiers won those four key home-and-away metrics against their grand final opponent.

Obviously a team has to concede the least number of points to win the premiership decider. But premiers conceded the least number of points from turnovers, and also spent more time with the ball at the attacking end of the ground.

“We haven’t seen a premier, on grand final day, lose the territory battle. We haven’t seen them, for the last 10 years, have the ball camped in their back half for more than the opposition and be able to win,” says Hoyne.

“We haven’t seen a premier concede higher scores from turnovers.”

Anyone from a senior coach to a newcomer to the game would expect a grand final winner to tick these metrics off.

Where quirks begin to emerge is in the preliminary final – the game that gets you into the big dance.

Hoyne cites three recent games where the winning team produced noteworthy aberrations on the stats ledger to win the game.

In 2018, Collingwood was able to win its way through to the grand final with a sublime stoppage performance against Richmond.

That, Hoyne says, was an occurrence rarely seen in a final.

“We saw Collingwood completely destroy Richmond from stoppages,” Hoyne says.

“Richmond still defended the turnover petty well against Collingwood that night, but were absolutely destroyed at stoppages.”

Goals from stoppages account, on average, for around 35% of a team’s final score. That night, around two thirds of the Magpies’ score came from stoppage wins.

Collingwood was on the receiving end of that unusual statistic the following year at the hands of the GWS Giants. GWS was the top team for scoring from stoppages that season, but the following week, Richmond beat them at their own game to clinch the title. That adds weight to Hoyne’s stat that stoppage scoring is a luxury that accounts for just a third of a team’s score.

The third example of preliminary finals dishing up unusual results is the Western Bulldogs’ away-game drubbing of Port Adelaide in 2021. There, the Doggies were able to dominate the home side in contested possessions – those earned when the ball is in dispute.

“That was more that was more a contested possession mauling.”

“We very, very, very rarely see that in a prelim or grand final. I think it was one of the worst ground-ball-gets differentials that we’ve ever recorded in a final.”

For context, Port Adelaide had more possessions than the Bulldogs overall, but thirty fewer contested possessions. At the first break, the Bulldogs had achieved what broadcaster Fox Footy described as an “utterly ridiculous” differential of 23 contested possessions.

Still, none of the teams that produced those unusual statistical milestones in a preliminary final claimed victory in the game that mattered the following week.

The other stats that matter

Punters and pundits will look for any edge they can to correctly predict a winner. Unsurprisingly, betting companies employ their own expert statisticians and data scientists to give the most accurate odds.

With the knowledge that Geelong is the only team still in contention which meets Champion Data’s ‘premiership metrics’ there are many more, simple stats some tipsters will look to.

In simple win-loss terms, Geelong has the best record of the remaining contenders: It only lost to Sydney this year. Brisbane has only lost to the Cats, the Swans have only lost to Brisbane.

Collingwood hasn’t beaten any of these three sides, and is one of the league’s lowest-ranked clubs in terms of differentials for contested possessions, clearances, and disposals.

Perhaps the fact Collingwood has won nine games by margins of under 10 points is proof that they do well in tight matches – the type of game you’d expect in a do-or-die final.

Or as Hoyne emphasises, all the data in the world can’t deliver certainty when it comes to what one group of humans will do with an inflated leather spheroid, against another.

“At the end of the day, regardless of the result, there’s always a story to tell from a data perspective. There’s always a reason as to why a team won or lost,” Hoyne says.

“It might not have gone as predicted, and a lot of the time it doesn’t go as predicted.

“But we never use the word ‘can’t’.

“We’ll never say someone can’t win because they haven’t been top six all season for points conceded – well they can, they can, but it’s unlikely or they’ll be bucking the trend.”

So it looks like the only way you’ll know for certain who wins the 2022 grand final is to skip into the future like Marty McFly and buy a guide to all the winners you’ll go back and bet on.

Marty mcfly

Good luck with your tips this weekend.

Please login to favourite this article.