Statistics

What’s the Use of a Ruckman?

Last night had Essendon disposing of the Blues in a pretty clinical performance at the MCG. From the scoreline, or just watching general play, you would not have thought Carlton won any area of the game. Essendon had more disposals, contested possessions and inside 50s by a long way. However, Carlton had a few more clearances than the Bombers (42-40).

As you can see in the figure below, winning clearances doesn’t correlate to winning the game. In fact, teams often win by more than 10 goals while only breaking even or losing in clearances.

clear_diff_vs_margin

 

This looks a little counter-intuitive. Looking even closer, none of the big wins (>100 points) came from games where the winning team dominated in clearances. Actually, most 100+ point beltings are delivered by teams getting less clearances than their opponents! The game last night is not an outlier at all.

What does seem strange about last night’s game, though, was that Carlton dominated the hitouts. Warnock had more hitouts than Carlisle and Daniher combined, and Carlton more than doubled Essendon’s count (58-28). But the clearance count was pretty close. What is the good of a ruckman if hitouts don’t lead to clearances?

The same thing happened on Friday, where Hawthorn easily disposed of Fremantle by 58 points. Sandilands dominated in the ruck against Hale and Ceglar, bettering their combined number of hitouts. Yet Hawthorn won the clearances 41-36. And on Saturday night, Dean Cox, Nic Naitanui and Callum Sinclair dominated Rhys Stanley and Tom Hickey in the ruck as West Coast won the hitouts 60-29, yet St. Kilda won the clearances 43-36.

GWS v Melbourne proved the exception, as GWS dominated hitouts 79-37 and clearances 61-36 (and won the game). However, the only stat Melbourne won was free kicks, and were probably saved from another humiliation by the weather.

The figure below shows number of hitouts plotted against number of clearances for the past 3 years.

hitouts_vs_clearances

 

There is nothing unusual about these numbers (except that the GWS game must be some sort of record!). It just so happens that hitouts don’t correlate at all with clearances! Sometimes teams get a lot of hitouts and a lot of clearances, but just as often teams will get just as many hitouts and very few clearances. Notice the fewest clearances of any team in the sample – 12, by Fremantle when they beat Brisbane in 2012 – came from a game where Freo had many more hitouts than average (60).

And now, as you expect, we can see that the hitout differential between teams doesn’t correlate at all with margin. Here are a few of this round’s games plotted against past data.

hitouts_vs_margin

 

So what is the purpose of a ruckman? I would hesitate to say that their role is completely meaningless. Hitouts to advantage would be an interesting statistic to investigate. It could just be that in the data available, a hitout is awarded to one ruckman or the other although most of them are largely ineffective. Really good hitouts are in there, but the signal is drowned out by most of the other hitouts.

So perhaps just having a tall player there to make it difficult enough for the opposing ruckman to get a clear and accurate hitout is enough to nullify their effect. So that David Hale, who gives up 10cm to Aaron Sandilands, could create enough of a contest that Sandilands would win the hitout, but not as easily as otherwise. Then the rest is up to Hawthorn’s superior midfield.

Another possibility is that a small advantage in the quality of rovers is much more important in winning a clearance than the quality of the hitout.

Either that, or ruckmen just don’t know what they’re doing.

 

Advertisements

The Strange Case of Melbourne vs West Coast

The West Coast Eagles flexed their wing-muscles again this weekend, beating up a bottom-four team for the second week in a row as Melbourne got well and truly towelled up by the visitors. What was surprising was to see that Melbourne had more disposals, contested possessions and tackles than the Eagles, yet still managed to lose by 93 points. Commentators are always quick to say that a key to winning is to “win the contested ball” and “maintain possession”. But if you were to look at just the stats (excluding the score), you would be excused for thinking the contest was much closer than the score line says.

Historically, number of disposals correlates reasonably well with final margin (with some scatter). Unsurprisingly, teams that possess the ball more typically outscore their opponents.

disp_diff_vs_margin

Melbourne (red) do not seem to deserve to have lost this game by so much!

Melbourne had more disposals than West Coast. So it seems like this game was an outlier. Did Melbourne get a lot of soft possessions in their back line? Well Melbourne only had 14 fewer inside fifties. As you might expect, inside 50s are also a very good indicator of margin.

i50_diff_vs_margin

This game still seems to have been an extreme outlier

 

But you can see that this still does not explain why Melbourne was beaten by so much! Having a few less inside fifties per quarter puts you at a disadvantage, but can’t explain a 93 point margin! The result makes more sense if we look at how a team’s score depends on forward-line efficiency. We would expect that, everything else being roughly equal, a team’s score is a linear function with scoring efficiency. And that is what we see:

goalsperi50_vs_score

Melbourne (red) were extremely inefficient in their forward 50 on the weekend.

The upper bound to a team’s score is dictated by their forward-line efficiency (which is fairly linear in the above plot). That is to say, at a fixed inside-50-to-goal conversion rate, there is a maximum score you can get because there are only so many times you can get the ball up there in a match. And the better this conversion rate is, the higher the score. The area underneath this limit is filled in by teams that function at reasonable efficiency, but just can’t get the ball into their forward line enough (i.e. are outplayed elsewhere on the ground). But Melbourne didn’t fall into that category on the weekend. The red point (Melbourne) is not sitting in the average-goals-per-inside-50-but-low-score region (where forward lines function okay, but just don’t get it enough), they are in the terribly-low-goals-per-inside-50-and-low-score  region! Melbourne didn’t get smashed because their forward line was starved, they lost because their forwards couldn’t turn an inside 50 into a goal.

The numbers show that Melbourne’s game doesn’t fall into a category where West Coast were simply far too good and beat Melbourne all across the ground – Melbourne matched it with the Eagles in many areas, including disposals, tackles and clearances, and weren’t smashed in inside 50s. But still lost by 93 points. The reason that Melbourne didn’t walk away with a more-respectable 5/6 goal loss is that they had a shocking conversion rate in their forward line. I suppose the silver lining is that this probably bodes well for Melbourne once they can get Mitch Clarke, Chris Dawes and Jesse Hogan back in the forward line.