There is more than one way to win a game of #afl football

Note: we were originally going to call this post “there’s more than one way to skin a cat”, but is there? We don’t even know of one way to do so. Why would you skin a cat anyway? Who would do such a thing?

At HPN we are firm believers in statistics being able to tell stories, if the right things are collected and context is considered. Endless numbers without context are meaningless, like training for a marathon that will you would never run.

When analysed well, statistics can not only support a narrative but develop one of their own. One of HPN’s favourite footy related reads of recent years is James Coventry’s Time and Space, which judiciously uses football statistics to demonstrate trends in how the game changed over time, such as the reliance of the handpass and short kicks to space by Jack Oatey at Sturt to counteract Fos Williams’ approach of long, direct, physical football with Port Adelaide.

Rohan Connelly’s piece on contested possessions got a bit of discussion in the Twittersphere this week, but its core thesis was largely correct (as we understand what he was trying to say): there is no golden football statistic that wins games. Connelly leapt on the chestnut of contested possession (CP) and CP differential, something emphasised on TV but much less in the fanalytics community.

(Related: the Madness of Sport piece here is worth reading)

Let’s look at CP differential against wins, just as a teaser, with premiers highlighted. This chart goes back to 1999 when the statistic was first available, so it’s a decent sample size.

 

Contested possession is often used as a descriptor of “hardness”, along with tackles and one percenters. Likewise, there has been a long held view by many in the football world that “hard” teams do better in finals than “soft” teams, and when September (or October) comes, the men are sorted from the boys.

(Somewhat related: this piece from The Arc is pretty good)

As you can see above, this is not necessarily the case. There is a positive correlation here, but the data is noisy and the correlation is pretty weak. Teams like the current Hawthorn side can afford to be poor with respect to contested possessions if other aspects of their game outweigh any actual or perceived weakness. It’s been six seasons since the league leaders in contested possession differential won the flag, a trend that looks like continuing this year. In fact, in the 17 seasons in our sample, only four premiers had the best contested possession differential across the season.

This year, Hawthorn was the number one side in the competition for kicks per handball. Last year, Hawthorn had the highest inside 50 ratio, and still had a very high kick per handball ratio.

For fun, here’s a seemingly useless statistic (bounce differential) against win percentage:

 

As you can see, there still is a positive trend here, albeit an even weaker one than above. Only five premiers since 1999 have had negative bounce differentials, but almost no-one would point to that as a reason for their success.

Going back to Coventry’s book for a second, many of the stylistic changes in football that resulted in success were based around strategic decisions, often framed as “rules” of teams. Whilst statistics didn’t necessarily inform the changes of the time, many had a gut feel (or perhaps more) of what they were.

Ultimately, there is but one critical statistic in a game of football: scoring. That is the black and white, and it’s pretty boring. Additional statistics (as basic as they are), provide the colour and shading to our brilliant game, and provide better ways to understand it.

There may not be more than one way to skin a cat (WE DON’T ACTUALLY KNOW), but there is definitely a few ways to be a successful football team.