Analysis Champion Data articles

[Teriyakicat]

Champion Data produces a huge amount of data on every AFL player and, from that information, every team.
A lot of the data is not available to the public, though the media and the clubs are privy to the information collected by CD.

What I'd like to do here is post any articles in the media that include stats that are not readily available from the AFL website, Footywire, or AFL Tables, as well as any articles that refer to comparative data, etc., even if those stats are something that we can work out for ourselves using the data online- e.g team rankings for clearances, contested possessions, etc.

There are a few posters here who, like myself, are interested in CD stats, and we often use stats to back up or refute an argument. I thought if we could post in this thread, any articles that we found, anyone who was interested in reading the information could easily access it.

Maybe, along with the link, post the date of the article and a quick summary, so we can search for it later.

Here is how Champion Data collect and record their stats (from 2009). CD have added to their bank of data over the years by improving and adding in information that coaches like to see:



Here are a couple of websites I get my info from:
http://www.footywire.com
http://afltables.com/afl/afl_index.html

 

[Jhye Clark 13]

CD make up their stats as they go along. Take anything they say with a grain of salt.

 

[Teriyakicat]

From the CD prospectus, here's an article about the Cats' 2015 list, which is rated third overall by CD, according to their measures of how many players are elite, above average, average, etc.

Article is from Jan 1015

While there have been predictions of doom for Geelong, official analysis of the Cats' list suggests that they should remain firmly in the premiership mix this season.

In its just released 2015 prospectus, Champion Data has ranked the Cats as third best in terms of list quality, behind only last year's grand finalists, Hawthorn and Sydney.

What would surprise many in the football world is that the Cats are even ahead of Port Adelaide - an early premiership favourite of many. The Power are fifth, falling behind a Fremantle side that showed signs last year that their premiership hopes may be dimming.

The Cats have four players listed as elite for their positions - Corey Enright, Jimmy Bartel, Joel Selwood and Harry Taylor - but a major shock was that superstar power forward Tom Hawkins fell only in the above-average category.

In booting 68 goals last season, Hawkins won 63 per cent of his disposals inside 50 - the highest percentage of any player in the league - and was so strong he marked 35 per cent of his one-on-one contests, 22 per cent above the league average.

Chris Scott's men tumbled out of the finals in straight sets last season after finishing third on the ladder but the 2011 coach through the pre-season has already indicated he believes the off-season additions of Mitch Clark and Rhys Stanley and a batch of draftees has his team "optimistic without being over-confident" that there will be improvement.

"We've been able to fill a couple of needs on our list, hopefully," Scott said. "Some of those guys have got some work to do, but on balance we think our list is looking better for 2015 than it was for 2014, so we're really positive about the opportunity we have."

http://www.theage.com.au/afl/afl-ne...ership-mix-data-suggests-20150114-12nzb0.html
(Jan 15th, 2015)


The ratings take into account when a player outperforms their opposition player. The complex formula gave each side an average score with the Cats score of +0.67 behind Hawthorn’s +1.62 and Sydney’s +1.16. The top four was rounded out with Fremantle on +0.63.
http://www.geelongcats.com.au/news/2015-01-15/stats-gurus-say-cats-have-top-three-list

 

[Teriyakicat]

CD make up their stats as they go along. Take anything they say with a grain of salt.

STOP IT!!!!

 

[Teriyakicat]

At the end of 2013, CD released similar data, ranking Hawthorn 1st, with the Swans 2nd, in terms of the quality of players on their list. Looks like they got that one right.


"Champion Data's list assessment takes into account a number of different aspects.

The analysis uses player rankings points over a two-year period with 75 per cent weight given to scores from 2013 and 25 per cent from 2012.

The ranking points assigned to each action on the field are defined by what correlates to winning an AFL game. At the completion of each season the formula is reviewed to determine and identify changes in the values of each event from a win/loss point of view, with points for certain stats adjusted accordingly.

All players are measured relative to their peers, tagging them as elite, above average, average, below average or poor. For a player to be given a rating they must have played at least 10 games over the two years."

http://www.perthnow.com.au/sport/af...head-of-the-pack/story-fniv69od-1226777876749
(Dec 7th, 2013)

 

[RegHickeyStand]

I was at Glenhuntly Road at my uncle Dick's home having a cup of tea in the mid 80's, and he was in my ear about football stats. Said Kevin Sheedy was around the other day, but thought he was trying to pump him for information more than bring him on board. Dick said to me "You know Reg, you give me any stats in any competition in the country, and I will easily select the best flankers, ruckmen etc...". But until Hopkins came on board, he could get no-one interested.

A very deep thinker, uncle Dick. His father, my grandfather, had a gym in Geelong and actually trained the Geelong football club for a while.

http://www.tedsport.com.au/uploads/articles/AFLRecordRound82011.pdf

 

[Teriyakicat]

I was at Glenhuntly Road at my uncle Dick's home having a cup of tea in the mid 80's, and he was in my ear about football stats. Said Kevin Sheedy was around the other day, but thought he was trying to pump him for information more than bring him on board. Dick said to me "You know Reg, you give me any stats in any competition in the country, and I will easily select the best flankers, ruckmen etc...". But until Hopkins came on board, he could get no-one interested.

A very deep thinker, uncle Dick. His father, my grandfather, had a gym in Geelong and actually trained the Geelong football club for a while.

http://www.tedsport.com.au/uploads/articles/AFLRecordRound82011.pdf

I think I'd have loved to have had a chat with your uncle, RHS. I would've had loads of questions (only on the Cats players, though) to ask him!

 

[SJ]

CD make up their stats as they go along. Take anything they say with a grain of salt.

No they don't.

 

[Jhye Clark 13]

No they don't.

Yes they do.

 

[SJ]

Yes they do.

Well they don't, and you are yet to bring any evidence to the table to back up your wild assertion.

 

[Teriyakicat]

Yes they do.

Well they don't, and you are yet to bring any evidence to the table to back up your wild assertion.

errr- is this the 5 minute argument or do you wish to purchase the full half hour??

 

[corio bay]

Great thread teriyaki!

I cant contribute much at this time however, a couple of years ago i would have been able to help but alas no more.

In the HUN sports news desk they have a computer hooked up to the CD backend.
All the stats you refer to are available on this computer.
It was great, you can look up the stats and then view the actual piece of play that the stats came from.

 

[Jhye Clark 13]

Well they don't, and you are yet to bring any evidence to the table to back up your wild assertion.

Wild? What's a contested possession or even a tackle considered to be deemed worthy of a tackle? How many statisticians are producing these outcome each game?

I know the answer, let's see if you do.

You're are naive if you take SC and its outcomes in all its glory. CD are just as corrupt as the AFL.

I can't give you precise proof for reasons.

 

[AM]

Wild? What's a contested possession or even a tackle considered to be deemed worthy of a tackle? How many statisticians are producing these outcome each game?

I know the answer, let's see if you do.

You're are naive if you take SC and its outcomes in all its glory. CD are just as corrupt as the AFL.

I can't give you precise proof for reasons.

Surely you can do better than generalisations and unanswered questions though. Surely!

 

[SJ]

Wild? What's a contested possession or even a tackle considered to be deemed worthy of a tackle?

"Contested possessions are made up of:
* Hard Ball Get (win disputed ball while physically beating an opponent at ground level)
* Loose Ball Get (win other disputed ball at ground level – ie. right place, right time)
* Contested Knock-on (knocking the ball out of a pack to a teammate’s advantage – Hawthorn do this a lot)
* Gather from Hitout to Advantage (similar to looseball get, but the ball was fed from the ruckman’s tap)
* Free Kicks (excluding those paid off the ball or downfield)
* Contested Marks (all marks where an opponent had a chance to mark or spoil – not including marks on lead)"
http://www.theroar.com.au/2014/07/09/missing-tablescontested-possessions-important/

Tackle: "Using physical contact to prevent an opponent in possession of the ball from getting an effective disposal"
2015 AFL Prospectus

Of course when setting a definitions statistical categories in a complex sport like Australian football one is not going to be able to draw perfect lines around qualitative descriptions. Sure, no one would be silly enough to guarantee that every single one of the 149,010 possessions recorded in season 2014 alone is perfectly allocated. But that's not the point, is it? Are you doubting the individual numbers themselves? Do you have examples of mistakes? They always look pretty good to me...

You do know that clubs pay for their data, don't you? Champion Data would exist for very long if the clubs realised that they were "making it up".

How many statisticians are producing these outcome each game?

I know the answer, let's see if you do.

About ten per game.

You're are naive if you take SC and its outcomes in all its glory. CD are just as corrupt as the AFL.

I can't give you precise proof for reasons.

I would be surprised if anything is produced in the world in a given year with 149,010 single entries (and they are just the possessions, not the rest of them) and not have some mistakes. Measuring things doesn't have to be 100.0000000% (yes, that is not the same as 100%) perfect in order for it to be of value. Any scientist knows that.

As for the "corruption" you are talking about, I hope you've got some evidence.

 

[Lana]

Measuring things doesn't have to be 100.0000000% (yes, that is not the same as 100%) perfect in order for it to be of value. Any scientist knows that.

This deserves more than a like.

For the football stats, as long as there isn't any purposeful manipulation and the methods of collection(and definition criteria) are the same over all games, there isn't much more that you can ask for.

I personally don't take much stock in there ranking points but the individual stats would be reliable enough.

 

[SJ]

This deserves more than a like.

For the football stats, as long as there isn't any purposeful manipulation and the methods of collection(and definition criteria) are the same over all games, there isn't much more that you can ask for.

I personally don't take much stock in there ranking points but the individual stats would be reliable enough.

AFL Player Ratings are the new breed and are far superior to the old Ranking Points (SuperCoach). Have a read up on the approach, I reckon it's brilliant.

 

[Teriyakicat]

Great thread teriyaki!

I cant contribute much at this time however, a couple of years ago i would have been able to help but alas no more.

In the HUN sports news desk they have a computer hooked up to the CD backend.
All the stats you refer to are available on this computer.
It was great, you can look up the stats and then view the actual piece of play that the stats came from.

If you read any stats-related articles that mention our boys, corio bay, just post the links here. Anything from the Hun, the Aged, the AFL site- whatever you find interesting. That's all I was hoping to do- share my interest in stats with other interested posters. No inside knowledge necessary. I don't read a lot of articles these days but, sometimes when I'm looking for an article I remember from ages ago, I find interesting little titbits of other info but the thread I'm in isn't relevant, so I don't know where to post the other articles. Hopefully this thread will be where these articles will find a home

 

[Lana]

AFL Player Ratings are the new breed and are far superior to the old Ranking Points (SuperCoach). Have a read up on the approach, I reckon it's brilliant.

I'm was under the impression that they took the same criteria into consideration but had a different weighting to the different acts.
If I'm not mistaken the weighting of the acts is based on the probability of that act leading to a score. Where as SC scores the acts have an arbitrary valuation.

But it still undervalues the performance of KPD, and IMO defensive acts in general.

I also don't like how they do the total rankings(goes back to 30 then from 31-40 with decreasing value) but I can't think of a better system other than average per game. They also don't provide an easy way to look at past performances which is why no one uses it or talks abut them.

 

[Teriyakicat]

This deserves more than a like.

For the football stats, as long as there isn't any purposeful manipulation and the methods of collection(and definition criteria) are the same over all games, there isn't much more that you can ask for.

I personally don't take much stock in there ranking points but the individual stats would be reliable enough.

If the data is collected by people and recorded by another group of people who have all had the same training- even allowing for differences in interpretation which would (hopefully) even itself out over the course of the year- couldn't you assume that the relative rankings of players would be about right? It's only a number but, forgetting the actual mathematics of it all, if everything was applied equally (weightings, etc.) across all players, shouldn't the relative numbers be a true indication of where each player sits within the various positions on the field? i.e. you wouldn't dream of comparing the numbers of, say, Gary Ablett with Tom Lonergan, for instance. But you can compare him with Joel Selwood. Or you can go back to the raw data and compare them yourself, if you don't agree with the (unspecified) weightings of the stats that CD have chosen.

 

[SJ]

I'm was under the impression that they took the same criteria into consideration but had a different weighting to the different acts.
If I'm not mistaken the weighting of the acts is based on the probability of that act leading to a score. Where as SC scores the acts have an arbitrary valuation.

But it still undervalues the performance of KPD, and IMO defensive acts in general.

I also don't like how they do the total rankings(goes back to 30 then from 31-40 with decreasing value) but I can't think of a better system other than average per game. They also don't provide an easy way to look at past performances which is why no one uses it or talks abut them.

Yes, you're spot on. Defensive actions are very hard to measure.

But I think it's a step in the right direction.

 

[Teriyakicat]

I'm was under the impression that they took the same criteria into consideration but had a different weighting to the different acts.
If I'm not mistaken the weighting of the acts is based on the probability of that act leading to a score. Where as SC scores the acts have an arbitrary valuation.

But it still undervalues the performance of KPD, and IMO defensive acts in general.

I also don't like how they do the total rankings(goes back to 30 then from 31-40 with decreasing value) but I can't think of a better system other than average per game. They also don't provide an easy way to look at past performances which is why no one uses it or talks abut them.

Isn't SC the stats that use the difference in the score at that particular time of the game to weight the effect of a stat? ie. a player gets more for an act if the scores are close but that same act is weighted less if the game is a blowout?

 

[Lana]

If the data is collected by people and recorded by another group of people who have all had the same training- even allowing for differences in interpretation which would (hopefully) even itself out over the course of the year- couldn't you assume that the relative rankings of players would be about right? It's only a number but, forgetting the actual mathematics of it all, if everything was applied equally (weightings, etc.) across all players, shouldn't the relative numbers be a true indication of where each player sits within the various positions on the field? i.e. you wouldn't dream of comparing the numbers of, say, Gary Ablett with Tom Lonergan, for instance. But you can compare him with Joel Selwood. Or you can go back to the raw data and compare them yourself, if you don't agree with the (unspecified) weightings of the stats that CD have chosen.

Isn't SC the stats that use the difference in the score at that particular time of the game to weight the effect of a stat? ie. a player gets more for an act if the scores are close but that same act is weighted less if the game is a blowout?

I can't agree with there relative valuation of different acts since I don't have access to there very complicated formula/models. But what I do know is there ranking system can have some funny results which make me take it less seriously.

For example the 2011 grandfinal
http://www.footywire.com/afl/footy/ft_match_statistics?mid=5342

They gave Leon Davis 119 points,
even though his direct opponent scored the first two goals of the game and he was pretty meh,
His ranking is more than anyone on our team other than Selwood,

More than Hawkins (who changed the game after half time) 117
Bartel 95 (3 goals and the norm smith)

Even though we won quite comfortably in the end, 6 out of the top 8 ranked players played for Collingwood and as a team they still ranked more points then Geelong.

 

[Teriyakicat]

I can't agree with there relative valuation of different acts since I don't have access to there very complicated formula/models. But what I do know is there ranking system can have some funny results which make me take it less seriously.

For example the 2011 grandfinal
http://www.footywire.com/afl/footy/ft_match_statistics?mid=5342

They gave Leon Davis 119 points,
even though his direct opponent scored the first two goals of the game and he was pretty meh,
His ranking is more than anyone on our team other than Selwood,

More than Hawkins (who changed the game after half time) 117
Bartel 95 (3 goals and the norm smith)

Even though we won quite comfortably in the end, 6 out of the top 8 ranked players played for Collingwood and as a team they still ranked more points then Geelong.

Ah, I can understand why you're gunshy now.

Unfortunately things like direct opponents running off a player are always going to be hard to track, when you consider the speed at which the game flows and the fact that players can switch opponents back and forth - if it was otherwise, we could subtract points from a player quite easily.

The Pies/Cats GF rankings don't make sense to me but maybe things are weighted differently than we think, then there's the factor that is applied to all actions when the game is tight and the smaller factor that is applied when the scoreline is bigger. Tackles and marks are valued more highly than kicks and marks and zero points are given for ineffective disposal, probably also subtracted for clangers (which includes frees against) I thought there was some sort of explanation that went around at the time, about how a mark or tackle or something was worth 4 points and a kick was worth 2, a handpass worth 1 point, etc.?

Just had a look around and found this article (the first post in the link) slightly helpful- hopefully it might shed some light on the issue:

http://www.fantasyfootyclub.com/showthread.php?2637-Supercoach-Scoring-System

 

[Lana]

Ah, I can understand why you're gunshy now.

Unfortunately things like direct opponents running off a player are always going to be hard to track, when you consider the speed at which the game flows and the fact that players can switch opponents back and forth - if it was otherwise, we could subtract points from a player quite easily.

The Pies/Cats GF rankings don't make sense to me but maybe things are weighted differently than we think, then there's the factor that is applied to all actions when the game is tight and the smaller factor that is applied when the scoreline is bigger. Tackles and marks are valued more highly than kicks and marks and zero points are given for ineffective disposal, probably also subtracted for clangers (which includes frees against) I thought there was some sort of explanation that went around at the time, about how a mark or tackle or something was worth 4 points and a kick was worth 2, a handpass worth 1 point, etc.?

Just had a look around and found this article (the first post in the link) slightly helpful- hopefully it might shed some light on the issue:

http://www.fantasyfootyclub.com/showthread.php?2637-Supercoach-Scoring-System

Ignoring the relative action importance weighting(which is a big issue on itself when you see it in action, if a team is three goals down and goes on to win, the first goal is just as important than the goal that puts them in front), each act, handball,kick mark on the lead etc is subscribed a certain number of points (I making up the numbers but the principal is the same) why is a kick worth 3 points and a hand ball 2 or a tackle 4 (I got these values from DT, the SC valuation has a lot more categories) there is still no basis for giving these acts this relative valuation. That's what the AFL player ratings is trying to do.