Kicking Accuracy : Teams, Rounds and Correlations

/

Accurate kicking, obviously, contributes to a team's success. But, just how much does it contribute, what are some of the sources of its variability, and just how predictable is it? 

In today's blog I want to explore four questions:

  • What's more important, generating Scoring Shots or converting more of those Shot opportunities into Goals rather than Behinds (ie being more accurate or having a higher conversion rate)?
  • Are home teams more accurate, on average, than away teams in the Home and Away season? Does this differ by team?
  • Is there a correlation between a team's accuracy on a given day and its accuracy during earlier portions of the season?
  • Does average accuracy by home and away teams vary across the season?

To explore these questions I'll be drawing on the combined scoring data for the period from 2006 to 2014, which spans 1,743 games.

SCORING SHOTS OR ACCURACY?

Is a team more likely to win if it generates more Scoring Shots or if it converts those Shots at a rate greater than its opponent? The data on this issue is unequivocal: teams that produce more Scoring Shots than their opponents win more often than teams that convert those Shots at a higher rate.

In quantitative terms, the team generating the greater number of Scoring Shots wins about 85% of contests, while the team that is more accurate wins about 66% of the time. Note that "accuracy" (or "conversion") is here defined as Goals / (Goals + Behinds).

Thinking about the logic of the game's scoring helps to understand why this might be the case. Consider an average team that kicks 14.12 (96). Its conversion rate is about 54%. A team registering just one Scoring Shot fewer (ie 25) would need to kick 15.10 (100) to beat them, which is a conversion rate of 60% or about 10% higher. Kicking, instead, 14.11 (95), which still represents a higher conversion rate (viz 56%) would be insufficient for victory for the team registering just a single Scoring Shot fewer. So, in this case, a 4% drop in Scoring Shot production (from 25 to 24) requires a 10% increase in conversion rate to offset it. The additional leverage of Scoring Shot production over conversion is therefore apparent.

Across the seasons, there's been a little variability in the winning rate of more accurate teams, falling as low as 62% in 2010 and peaking at 69% in 2008. Less variability has been displayed in the winning rate for teams with greater accuracy, however, the low being 84% in 2010 and the high 89% in 2011.

HOME TEAM AND AWAY TEAM ACCURACY

My ingoing hypothesis was that home teams would have slightly better accuracy than away teams. Looking only at games from the regular home and away season shows that this is not at all the case, home teams and away teams both converting about 53% of all Scoring Shots.

On a team-by-team basis we do see some variability in conversion rates when playing at home compared to playing away, but only for Essendon, Port Adelaide and St Kilda are these differences statistically significant. The effect sizes are about 2.5% to 3% points. Essendon and St Kilda both convert at statistically higher rates at home than away, but Port Adelaide does the opposite. That said, Port Adelaide still only has the 4th best record for conversion when playing away.

Across the 18 teams we see greater variability in conversion rates, with Hawthorn recording the best conversion rate at home (55.3%) and Gold Coast the worst (50.2%), and the Western Bulldogs recording the best conversion rate away (54.9%) and Gold Coast the worst (49.9%).

Overall, across home and away games, Hawthorn has the best conversion rate (54.9%) and Gold Coast has the worst (50.1%).

CORRELATIONS IN ACCURACY

Given the differences in conversion rates at the team level, you might expect there to be a correlation between a team's conversion rate in the current game and its conversion rate across recent games. In other words, you might expect that knowledge of a team's recent levels of accuracy would help you predict its accuracy on a given day.

I certainly would have expected this to be the case before I ran the number. But it's not true at all, as the table at left reveals. There's virtually no correlation at all between a team's conversion rate in a particular game and its historical accuracy.

For example, the correlation between the conversion rate for home teams and the conversion rate of those same teams, at home or away, over the three most recent games in the current season (or over two or one game if the season is only two or three weeks old), is just +0.04. For away teams, it's even lower at +0.01.

Taking in larger swathes of recent history in the hope of removing one source of unwanted variability does little to help. Even if we use a team's entire year-to-date record, we're still unable to lift the correlations above +0.04. For all practical purposes that number might as well be zero.

ACCURACY ACROSS DIFFERENT PORTIONS OF THE SEASON

Lastly, let's review home and away team accuracy at different points in the season, splitting the home and away portion of the season into rough quarters, and treating Finals as a distinct, fifth portion.

Home teams are less accurate than away teams in both quarters from the first half of the season, though in neither instance are the differences statistically significant. In the second half of the season, the opposite is true, with home teams the more accurate, though again the differences fail to clear the significance hurdle.

There's some evidence that teams, perhaps, are slightly less accurate across the first six rounds of the home and away season compared with the remainder, though the effect size is only about 1% point. Evidence that teams are less accurate in Finals compared to the Home and Away season is only marginally more compelling, the effect size being about 1.5% points.

SUMMARY

Evidence from the last nine completed seasons suggests that:

  • Scoring Shot production is more important than accuracy in driving team victories
  • Home team and away teams tend to have similar levels of accuracy across the Home and Away season taken as a whole
  • Only a few teams have statistically significant differences in their accuracy levels when playing at home compared to when playing away
  • Accuracy levels vary to a larger extent between teams than they do for any given team when playing at home versus playing away
  • There is virtually no correlation between a team's single-game accuracy and its historical same-season accuracy, no matter how small or large a portion of history we take
  • Accuracy might be slightly lower in the early parts of the Home and Away season compared to the remainder, but the effect size is small. Accuracy might also be lower in Finals compared to the Home and Away season, though again the effect size is relatively small.