Which is why I keep getting into lengthy posts trying to explain that to people.
Especially people who try and justify an opinion with dodgy stats.
Opinions are fine.
Misusing stats is annoying.
2 Likes
Thatâs why they donât publish it. Pay up!
The AFL paid up for live VFLW stats this weekend, at last.
There are lies.
There are damned lies.
Then there are stats.
(apologies to Samuel Clemens.)
Why are you trying to redefine kicking efficiency? Champion data tells you what it means and how they measure it. If you want a different stat made available like kicking to advantage percent or kick retention etc then ask for that stat, kicking efficiency is what it is. The data only looks skewed if you dont understand it.
1 Like
He should know - he had a lot of effective disposals.
Handy points not so handy
1 Like
Nothing wrong with the assumptions that the stats are based on (although obviously the interpretation of those in game by the CD stats monkeys is open to interpretation). The problem is that people interpret âkicking efficiencyâ to mean something that it doesnât.
For your example of effective/neutral/ineffective, that even ignores context and meaning of a kick. If Harry Jones outmarks 4 opponents in a 1 on 4 a kick becomes effective, whereas a brilliant kick to the advantages side of Hooker that becomes and intercept mark because Hooker gets illegally held out of the contest becomes innefective.
The problem is that CD stats that are presented for mass consumption are overly simplistic, but also very unclear. And their âspecial sauceâ stats that they sell to the clubs arenât available for public consumption.
In an ideal world there would be a massive community of amateur stats takers who were taking next gen stats and providing them to the rest of the stat loving community. Unfortunately there arenât the numbers of people, and itâs actually really difficult to generate interesting observations from the TV coverage. You just canât see enough of the field at any time to understand the context of a play to the extent that youâd need to in order to develop next gen stats.
1 Like
I thought youâd understood but clearly you donât get the premise of statistical models.
And it still sounds like youâre employed by Champion Data.
Perhaps you might like to cast your eyes over this article;
Umm, I think you made Tylerâs point. The interpretation and understanding is more important than the numbers. I think youâd probably both agree with that. You are arguing that because kicking efficiency doesnât identify the best kicks itâs a stuffed stat. Tyler is arguing kicking efficiency isnât trying to measure the best kick, itâs trying to measure the highest âkicking efficiencyâ, which is defined as it is.
In simple terms, stats mean nothing if you donât understand what is being measured, and trying to apply those stats to show something is meaningless if you canât interpret what the stat actually means
2 Likes
And further to that, to go back toy point. CD provides rubbish stats for public consumption, and then ex-footballers with very little analytical expertise or training are translating these for public consumption.
No wonder AFL is amongst the worst sports in the world for analysis.
2 Likes
Logic would suggest youâre right but you forget theyâre not operating in a free market. They are merely another entity designed to rinse cash out of AFL ânot for profitâ coffers and into the hands of the individuals who control the game. I imagine they get paid very very well by the comp, the media outlets and the clubs.
4 Likes
No, Iâm arguing that the parameters used to determine kicking efficiency by Champion Data need to be changed.
Thatâs because of your own understanding of âefficiencyâ.
Of it was called KTLS Percentage would you care?
1 Like
Kicking needs a large expansion of data collected to be of value.
You need to know the scenario at both ends - at the disposal side (pressure rating of the kick, distance, speed of kicker, field position, from stoppage, from centre clearance, from mark, from kick out, from loose ball) then look at the impact at the other end (mark, to advantage, contest vs free player, does it form part of a scoring chain, distance, field position etc.)
Only then can you start really understanding the most important aspect of the game. Itâs too much to collect in game, so it would have to be done in post processing of stats.
The other problem you will run into is sample size, with so many categories you are unlikely to have any in game trend, and it would be more useful as a historical view of the player based on a larger sample size. It may have some use team wise week to week.
Iâll have one last go at this.
What is your understanding of efficiency?
Is it the same as that used by Champion Data?
If so, are you happy to concede that Champion Data is infallible when determining statistical models?
If the answer to the last two questions is yes, then fine, all bow to the master stats behemoth.
For those that arenât as convinced of CDâs infallibility, and might dare to question them.
Some footy stats are objective. Marks & goals for example.
Some are subjective.
Kicking Efficiency being one. In order for CD to produce stats on this measure that are comparable they have created an arbitrary set of parameters to determine if a kick is âefficientâ or âinefficientâ. Otherwise it would be recorded differently by every different collator.
So far so good.
Thatâs the statistical model.
However it leaves the question of that initial set of parameters.
Eg. If the ball is kicked further than 30 metres to a contest it is deemed efficient
Why 30 metres?
Why not 25, or 40?
Is that distance significant in any objective way? No.
But again, Iâve got no issues with them developing such a set of parameters. All statistical (and scientific) modelling requires them.
However I reserve the right to question the validity of the model. Again, basic stats & science. Itâs how these things improve.
Why would I want to question the subjective rules underlying the model?
Because thereâs an obligation for those doing the modelling to compare their model - in this case Kicking Efficiency - with objective, real world outcomes.
And if there are significant discrepancies, then the model, or at least some of the assumptions underpinning it, needs to be improved.
Whew.
A defence of the scientific method before my second coffee
Iâll shut up and exit this Thread.
3 Likes
Donât exit the thread, discussion is good.
I understand many different meanings of âefficiencyâ. When it comes to fuel efficiency, or aerodynamic efficiency, or efficiency with language I generally understand the meaning based on the context in which the word is used.
When it comes to âkicking efficiencyâ I understand it to be completely made up to try and convey something about the quality of someoneâs kicks.
The definition tells me something, based on some arbitrary set of rules. Is that set of rules good? Not really, itâs just a set of rules.
If your argument is âkicking efficiency is a fairly meaningless stat that doesnât offer much insightâ, sure I agree.
If your argument is âCDs stats donât offer much insight into how or why football games are won or lostâ Iâd agree with you, at least in terms of the stats that are available for public consumption. I have no idea if they have other, better stats.
If your argument is âCDs definition of what makes a kick good is wrongâ Iâd be somewhat on the fence. Theyâve at least tried to create a definition that differentiates between the contexts in which a kick is made. Is it overly simplistic, yes. Does it offer insight? Not much. Iâd think that CD could greatly improve their modelling of good and bad kicks, how they measure the components and how they report the outcomes. And the general football public would be confused. But thatâs ok.
If your argument is âHow CD measures of kicking efficiency is wrongâ, I would disagree. They define it, they tell you how itâs measured, they put the numbers together. In terms of what they define it to be, kicking efficiency is what it is. How you choose to interpret the reported number is up to you. Want to throw it out as a stat that means nothing? Sure, that is reasonable. The measure that is reported means nothing other than what it is defined to mean.
Hopefully we can come up with the WSTMKQ soon. (The Windsock-Tyler Measure of Kicking Quality)
It will definitely be letter than Kicking Efficiency.
2 Likes
Yes, yes, yes, but what happens if you throw a car out of a plane?
I think the contested possession stat is rubbery too fwiw.
I think tackles could be split into effective and ineffective too.
Desperately need a ball retention stat.
Iâd also like a count of successful evasions
What ever happened to hard ball gets (stupid name, good stat)
First possession also use to be reported separately to clearances.
Itâs difficult to really assess the value of CD because only a fraction of it is published
1 Like
It depends, is Gil driving?
1 Like
Anyway, my take on CD publically available stats and the use of them in broadcast and media in general is that the analysis presented is very very shallow and offers very little to help the average fan to understand what is happening in the game and why it is happening beyond what they can see with their own eyes.
It could and should be better, but the CD monopoly has created a situation where real insight is hidden behind âcommercial secrecyâ and the stuff available to the public is designed purely to appease the publicâs desire for numbers, without actually providing any understanding.
2 Likes
There is - effective tackles are called âtacklesâ.
Ineffective tackles only get counted as a pressure act
1 Like