Not all kicks are the same. But for much of footy’s history, they have been treated as such, aside from the totals in the goal and behind columns.
A kick has been a kick, whether it be a drop, flat or screw punt, stab pass, checkside, banana, torpedo, drop or place kick.
That all changed in the 1990s, with one man deciding that the quality of kicks was worth investigating. That man was Ted Hopkins, a hero for Carlton in the 1970 grand final and founder of Champion Data.
Hopkins surmised that putting more descriptive information around elements of the game might help give us a better understanding.
The information revolution around football has progressed with a fair deal of pace, but many of the terms that we still use now were developed by Hopkins in the late 1990s and early 2000s. One of the main ones is effective — and ineffective — kicks.
Commentators each week talk about good players having a lot of effective kicks, with the term believed to align with the better ball users in the game.
One data scientist thinks he might have unlocked the next evolution in defining what a good kick might be. In the process, he has uncovered some uncomfortable truths about the official “effective” kick.
Ready to roost it
It is a sight you have likely seen hundreds of times before. Fremantle have the ball hemmed up in the back half of the ground, close to their boundary line.
Hayden Young sums up the situation from behind the mark, looking for options up the field.
Loading YouTube content
According to Liam Crowhurst, a data scientist, football analyst and co-founder of the Useless AFL Stats online community, this is one of the most common kicks in football.
Footy staffers call it the “line kick”, or more commonly “the roost”.
“A kick down the line about 45 metres in length is effective 90 per cent of the time, but is only retained approximately 30 per cent of the time,” Crowhurst said.
This hits at the tension in the term “effective kick”.
As per Champion Data and the AFL, an effective kick is where the ball finds a teammate, or is kicked to a contest at least 40 metres away.
That latter part is an arbitrary line that was added to recognise an old truth, that a long bomb to a contest is good enough.
Unfortunately, in modern football, that rule of thumb definition of effectiveness seems to be hiding the true value of many long kicks.
Crowhurst has inferred the likely outcomes of kicks given their starting location, target location, and whether the kicks occur in general play or after a free kick or mark.
To do this, Crowhurst analysed more than 170,000 kicks taken in the past three years, using machine-learning techniques.
“By taking an outcome-based approach to kicks, rather than an antiquated definition, we can model how likely a kick is retained by a teammate or ends in a contest,” Crowhurst said.
Crowhurst has also worked out the likelihood the kick will end up being contested, and how often it could be directly turned over.
That kick we highlighted above — from Young to traffic and eventually Melbourne’s hands — is expected to be retained 30 per cent of the time, turned over 25 per cent and results in a contest the remaining 45 per cent of the time.
The roost is also almost always not the first choice of kick, or the second or even third.
Before he kicks the ball, Young’s eyes dart anywhere but long. An example involving James Aish from the same match demonstrates the thought process even more clearly.
Loading YouTube content
It is often the option of last resort: to limit damage caused by an effective defence when other options are closed.
“Kicking efficiency is based on a metric from a bygone era, and if the AFL is to evolve into the modern era, we need to get more nuanced in the way we describe the game,” Crowhurst said.
A by-product of the definition of effective kick having to travel 40 metres before getting to a contest is that there is a spike in the expected number of effective kicks that travel just over 40 metres.
There is a divergence between “effectiveness”, as per the official measure, and retention. Instead of measuring the game from a neutral point of view, the concept of “effectiveness” may be directly influencing it.
Some teams, like Fremantle and Adelaide, sit near the top of the league for kick effectiveness, but retain the ball at far lower rates than this suggests.
Crowhurst has also derived another further measure: retention rating. It shows how well teams retain the ball, accounting for the difficulty of the situation.
“Retention rating normalises the difficulty of kicks and allows us to see which teams are hitting their targets,” Crowhurst said
“If teams are taking on more difficult kicks, are they biting off more than they can chew?”
But retention alone does not tell the whole story.
Let us review Young’s kick above. Crowhurst has worked out that kick is expected to be retained 30 per cent of the time, but how often will it lead to anything of substance?
“While the kick down the line is a regular feature of the game, it generates a scoring shot about 10 per cent of the time,” Crowhurst said.
Holding onto the ball is only one part of the equation in football. Doing something meaningful with it is something else entirely.
Crowhurst has looked into how “threatening” a kick is, and which players use their kicks to turn ordinary situations into damaging scoring opportunities within the same chain.
“By looking at each kick within the chain of possession, we can model the probability that a kick leads to a scoring opportunity,” Crowhurst said.
Generally speaking, the further up the ground the ball is, the more likely it is expected to lead to a score.
Kick threat can show the intent of sides and how effective teams are at pulling off what could be described in a quantifiable sense as a valuable kick.
For example, while the Crows are not retaining the ball as well as expected from their kicks, in offence they are being much more threatening than their field position and kick choices would suggest.
Kick threat highlights the better ball users in the game, such as Geelong’s Gryan Miers.
Loading YouTube content
Despite generally taking tough kicks, his ability to use the ball by foot has consistently led Geelong to generate scoring opportunities.
“Players like Amon, Dawson, Houston and Wilkie have positive retention and threat ratings,” Crowhurst said.
“Their kicks outperform the AFL average, their teammates are more likely to retain possession and convert it into a scoring shot later in the chain.”
Some players with a low kick efficiency actually perform well when adjusting for the difficulty of their chances.
“Matt Rowell’s kicks have a kicking efficiency of 54 per cent, one of the lowest of the top 50 most kicks in 2023,” Crowhurst said.
“When you take into account the difficulty of the kicks, he outperforms the AFL average for threat generation.”
Certain kicks are also more damaging than others. They are the kicks that have a high expected retention rate and a high threat rating.
Generally, these kicks travel across the field more than straight down it, across shorter distances rather than longer. They often shift defences and open up angles of attack.
The clock stops ticking
Already, some within the broader media landscape question the value of the effective kick.
Loading Twitter content
Hopkins himself questioned the lack of progress of quantitative measures beyond his groundbreaking early work.
“By the time Angelika [Oehme] and I sold the business to new private owners in 2007 and I continued working under contract for the company, I noted the appetite for innovation waned and the stats product itself was becoming more routine and embedded in the official lexicon,” Hopkins wrote on the Australian Football website.
“My observation, during this period, was the game itself had changed dramatically from the mid-2000 period and onwards and was hurtling into uncharted waters compared to the fundamentals of yesteryear.”
As the game continues to evolve, the need for the terms used to describe the game accurately grows. This needs to be done in a way that doesn’t alienate the audience.
At the same time, tensions still remain high between the heaviest data users and the official stats provider.
Some clubs are growing increasingly frustrated by elements of services provided by Champion Data, raising those concerns in a meeting with the AFL recently.
There is no doubt that the work of Champion Data has progressed the understanding of the game, but the general feeling exists that more — and better — access to data could be provided.
Crowhurst’s work is an example of a new way of looking at the game and it may start to propel footy to a new analytical basis.
Source: AFL NEWS ABC