AFL Sabermetrics

[West123]

Flowing on from the "Moneyball" thread, I'm just opening up a discussion for sabermetrics which is basically advanced statistics that would be relevant to AFL and judging differences in players ability. Basically, if you scrapped disposal stats what would you use?

Obviously goal assists has been a buzz term in the last decade or two and pressure efforts have come in recently and contested marks inside 50 will always be a massive indicator.

One I'd be interested to know is disposal efficiency on inside 50's.

What other stats are teams using or that you could think of that would be useful outside of the traditional stats?

 

[Prem87]

Supercoach stats.

 

[Cynic]

I'd like to see more defensive stats

Backs get ridiculed for the number of goals kicked on them but rarely praised for the number of goals they prevent...I guess it should be scoring shots allowed rather than goals however.

 

[legend166]

Uncontested marks allowed in the forward 50 is a good indicator of a crappy defence. I want to see that.

 

[The Falcon Strike]

How many times a player looks at his teammate and then points randomly in some direction.

 

[Falchoon]

Effective clearances v clearances

 

[baja]

The 1%er stat is a new one, I still dont fully understand how its officially scored although I know a 1%er when I see one if that makes any sense..

 

[krayfesh]

The 'find a team mate when kicking' stat.

Shots at goal. Pressure acts. Tackles. Effective clearances.

 

[doppleganger]

High quality disposals that hurt the opposition.

 

[Patch Adams]

a weighted disposal efficiency which gives a disposal eff% which weighs uncontested possessions more...

and an AFL equivalent of VORP, as the closest thing we have now is Champion Data points which is heavily biased towards midfielders

 

[Black Diamond]

Maybe consider that baseball (along with cricket) while a team sport is highly individualised. Nobody but the pitcher affects the trajectory. Nobody but the batter controls that response. The coordination of fieldsmen as a team work together, but still are highly individualised. I'd speculate this would allow for a set template of stats for baseball recruitment references.

In contrast is footy, soccer, and other 3 dimensional sports. Here I'd suggest a particular recruiter would have to be far more in tune with any given game plan and have a particular skill for analysing the present circumstance, combining genuine game analysis with raw stats to reach an outcome. For example, a player who takes 5 contested marks out of 5 attempts may well have done better than another taking 10 contested marks from 30 attempts.

 

[ilikepotatoes]

How many times a player looks at his teammate and then points randomly in some direction.

want mj

 

[frantelle]

For key position players:

% of marks/disposals taken when the target going forward

 

[divvydan]

Really the only thing that could be advantageous to what is currently looked at now is disposal under varying forms of pressure; i.e. Low, Medium, High, Severe. With the forward press becoming a dominant tactic, disposal under pressure and by extension decision making under pressure will, imo, replace disposal efficiency as the primary method of evaluating a player's usage of the ball.

Done correctly, this would also take into account the pressure on the various options a player had (i.e. loose teammate, 50/50's, etc) so as not to unreasonably disadvantage players with less options than one with more.

The main difficulty with measuring this is the difference pace and style of under age and state level football compared to AFL level.

 

[Donkson]

AFL isn't anywhere near as mathematically/statistically explainable as baseball is.

 

[Slattery_20]

Maybe consider that baseball (along with cricket) while a team sport is highly individualised. Nobody but the pitcher affects the trajectory. Nobody but the batter controls that response. The coordination of fieldsmen as a team work together, but still are highly individualised. I'd speculate this would allow for a set template of stats for baseball recruitment references.

There's a lot of stats in gridiron, too.
And (from the limited I know of that baffling game) I think there's different classifications of turnovers/dropped passes - dropped pass, fumble, etc which I think relate to whether the passer made a mistake, the defender affected the catch, or the catcher made a mistake.
I think.

One are where the yanks are waaaay ahead of us.

In basketball there are stats like +/- (score differential while you're on), %age of rebounds (what proportion you're getting of the total rebounds while you're on), effective shooting %age (shooting percentage adjusted for difficulty of the shot)
I think similar could be worked out for AFL - perhaps %age of centre clearances, +/- clearances while you're in there etc, +/- score, effective kicking efficiency (ie 20m pass in the backline vs hurried kick from the middle).

 

[ladder_goat]

I remember reading something a while ago that some stats guys did try this. What they came up with was that two of the most important attributes were running bounces and long kicks to advantage.

Of course i may have been dreaming.

 

[West123]

I think similar could be worked out for AFL - perhaps %age of centre clearances, +/- clearances while you're in there etc, +/- score, effective kicking efficiency (ie 20m pass in the backline vs hurried kick from the middle).

I've seen these kind of numbers. The AFL.com site did publish a top 20 clearance group at some point in the season.

 

[calyam]

I think similar could be worked out for AFL - perhaps %age of centre clearances, +/- clearances while you're in there etc, +/- score, effective kicking efficiency (ie 20m pass in the backline vs hurried kick from the middle).

I like the idea of this. I imagine this is the sort of data that clubs have internally but which we never see.

AFL is not a great game for statistics since it's fundamentally messy and there is always a lot going on. But there's plenty of scope for better and more relevant statistics. Borrowing some concepts from other team sports such as basketball might be a valuable way to begin to derive new statistical measures.

 

[manicmagpie]

I remember reading something a while ago that some stats guys did try this. What they came up with was that two of the most important attributes were running bounces and long kicks to advantage.

Of course i may have been dreaming.

You're not dreaming. I too remember this and seem to recall that it was an American stat company that was involved in Moneyball theory.

I am sure that an effective long kick was what they determined to be the single biggest positive influence on the outcome of a game.

 

[Gladbaqs]

There is one stupendously simple stat not recorded at the moment to my knowledge.

Shots on goal

The closest we have is "scoring shots" which is inaccurate because it includes rushed behinds and omits misses including those that dropped short.

 

[SJ]

Most of these would already be measured by Champion Data. It's just that the public doesn't have access to them.

 

[Belnakor]

i'm still mystified why the public doesn't have access to champion stats, yet they'll license out to the hun.

we don't even have simple stats like go in front goals etc

 

[stefoid]

Disposal under pressure for mine.

 

[calyam]

i'm still mystified why the public doesn't have access to champion stats, yet they'll license out to the hun.

we don't even have simple stats like go in front goals etc

Probably because they want to make money ...