Champion Data Statistics - Does Anyone Really Know?

[Fadge]

I do not like Supercoach because of the mystery around exactly how Champion Data statistics are calculated.

Looking at the Supercoach website, there is a general summary with an interesting disclaimer that goes like this - 'Due to the large number of factors affecting scoring, we cannot publish the full system'

www.championdata.com.au provides a one page pdf with a few more hints, but with a similar disclaimer - 'Given the unique intellectual property contained in the rankings formula and the degree of underlying computer logic applied, Champion Data does not publish the full rankings formula'

My occasional watching of the quarter by quarter scores provided last year led me to the belief that each of the quarters are weighted based on how influential they were to the result of the match. In the event of a close match, the scores accrued in the final quarter have more weighting that the scores accrued in each of the first three quarters. Conversely, in the event of a blow-out in the first quarter, the scores obtained in that first quarter are given more weighting.

Using this logic, if Travis Cloke were to kick 5 goals in the first quarter and the Pies were 7 goals up against St. Kilda, he might be 80 Supercoach points. Now if he goes off injured and misses the remainder of the match, and the Pies cruise to a 10 goal victory, my theory suggests that Travis would end up with more than 80 points as a higher weighting would be given to the first quarter scores.

Has anyone else seen this pattern, and/or does anyone actually know how the mystery that is Champion Data really works?

 

[Hodgepodge]

Yes.

 

[Rohan25]

Yeah, you're pretty much right with the weighting thing. A goal to win the match in the dying seconds may be worth 30 odd points on its own.

 

[100action]

you might need to consult KS on that one.

 

[Hodgepodge]

Yeah, you're pretty much right with the weighting thing. A goal to win the match in the dying seconds may be worth 30 odd points on its own.

No, I don't think it is.

A goal is counted points for being an effective kick(4 points) and for the score(8 points)

 

[divvydan]

No, I don't think it is.

A goal is counted points for being an effective kick(4 points) and for the score(8 points)

Rohan is correct. There is significant weighting to actions like kicking goals that win matches, or give a team the lead in the dying minutes of a match, can be as much as 4x that standard score.

Edit: Example from the prospectus. Leo Barry's mark in the grandfinal would normally be worth 10 pts but because of the situation, he was awarded 39 pts for that mark.

 

[Hodgepodge]

Rohan is correct. There is significant weighting to actions like kicking goals that win matches, or give a team the lead in the dying minutes of a match, can be as much as 4x that standard score.

Edit: Example from the prospectus. Leo Barry's mark in the grandfinal would normally be worth 10 pts but because of the situation, he was awarded 39 pts for that mark.

I don't know about that.

 

[Catsace]

I can vouch that in the Carlton/Eagles game in Perth in 2006 (the week after the infamous Geelong comeback game) i had Andrew Embley, who by midway through the last quarter had about 90-odd points. A couple of late marks and goals to him that put the Eagles in front for the narrow victory saw his score leap to 177, which i was very happy with.

But some of the scores are a bit hard to fathom.

 

[bedford]

Champion Data is the greatest waste of space in footy, but it keeps a dud like Mark Stevens in a job

 

[the_interloper]

Champion Data is the greatest waste of space in footy, but it keeps a dud like Mark Stevens in a job

He'd get paid **** all for his job, seriously, average journo in a shit paying job fair dinkum!

 

[Crow-mo]

Champion Data is the greatest waste of space in footy, but it keeps a dud like Mark Stevens in a job

you should offer your services to the league.

 

[Crow-mo]

I do not like Supercoach because of the mystery around exactly how Champion Data statistics are calculated.

Looking at the Supercoach website, there is a general summary with an interesting disclaimer that goes like this - 'Due to the large number of factors affecting scoring, we cannot publish the full system'

www.championdata.com.au provides a one page pdf with a few more hints, but with a similar disclaimer - 'Given the unique intellectual property contained in the rankings formula and the degree of underlying computer logic applied, Champion Data does not publish the full rankings formula'

My occasional watching of the quarter by quarter scores provided last year led me to the belief that each of the quarters are weighted based on how influential they were to the result of the match. In the event of a close match, the scores accrued in the final quarter have more weighting that the scores accrued in each of the first three quarters. Conversely, in the event of a blow-out in the first quarter, the scores obtained in that first quarter are given more weighting.

Using this logic, if Travis Cloke were to kick 5 goals in the first quarter and the Pies were 7 goals up against St. Kilda, he might be 80 Supercoach points. Now if he goes off injured and misses the remainder of the match, and the Pies cruise to a 10 goal victory, my theory suggests that Travis would end up with more than 80 points as a higher weighting would be given to the first quarter scores.

Has anyone else seen this pattern, and/or does anyone actually know how the mystery that is Champion Data really works?

the champion data rankings are proprietary information, any reason you feel you should be told of the exact constituents?

whilst you're there, I suggest applying a freedom of information request to Coca Cola and McDonalds

 

[EaglesFlyingHigh]

Yeah, you're pretty much right with the weighting thing. A goal to win the match in the dying seconds may be worth 30 odd points on its own.

What a crap system

 

[Powerade 08]

What a crap system

Why?

Isn't the aim of supercoach points to measure the effect a player has on the result of a game?

 

[Stoddy]

Yeah complicated system but doesnt make it a bad one. I'm no statistician but i've been told that every game is allocated the same amount of points and bell curves etc are used to standardise the values of the various stats used by CD in there scoring system. So for example a game (or quarter for that matter) that a lot of goals are kicked the goals are worth less than a game(quarter) that fewer goals are scored...dunno if that answers your qst but thats all i know.

 

[Vacuous Space]

Have they changed the weighting of the system this year or is it the same? It does seem odd that Luke Miles and Travis Cloke ended our game with the same score (79). Miles had less possessions, less contested, less marks, a lower efficiency, didn't kick any goals, and had one more clanger. Miles had one more tackle. Riewoldt kicked 4 goals and ended up on a lower score (63). Leigh Brown also had four goals (81).

I'm not criticising, I'm just wondering.

 

[Damon_3388]

Have they changed the weighting of the system this year or is it the same? It does seem odd that Luke Miles and Travis Cloke ended our game with the same score (79). Miles had less possessions, less contested, less marks, a lower efficiency, didn't kick any goals, and had one more clanger. Miles had one more tackle. Riewoldt kicked 4 goals and ended up on a lower score (63). Leigh Brown also had four goals (81).

I'm not criticising, I'm just wondering.

That's pretty curious, considering Luke Miles was barely sighted during the game, and Cloke was instrumental in our comeback late in the 3rd quarter.

AFAIK, the Champion Data Player Ranking does place importance on when and where on the field players do things, and gives heavier weighting to forward-moving play and contested possessions as well.

 

[Fadge]

the champion data rankings are proprietary information, any reason you feel you should be told of the exact constituents?

whilst you're there, I suggest applying a freedom of information request to Coca Cola and McDonalds

Not sure I said I had a god-given right to have access to the intellectual property, was just interested to know if anyone knew more about it than I do (which isn't much).

When I participate in a competition, I just like to know how it is scored. Maybe that's just me - it obviously doesn't interest you. Sounds as though you're happy living your life like a mushroom.

 

[Fadge]

Yes.

No, I don't think it is.

A goal is counted points for being an effective kick(4 points) and for the score(8 points)

I don't know about that.

Do you think, or do you know Hodgey?

Thanks for your valuable contributions nonetheless.

 

[Crow-mo]

Not sure I said I had a god-given right to have access to the intellectual property, was just interested to know if anyone knew more about it than I do (which isn't much).

there is a reason why the exact formula isn't known.
its no different to Baseball Prospectus' PECOTA or Hollingers NBA Player Efficiency Ratings (PER).

in each case, broad factors are understood by the exact algorithms are not.

When I participate in a competition, I just like to know how it is scored. Maybe that's just me - it obviously doesn't interest you. Sounds as though you're happy living your life like a mushroom.

so the scoring system isn't described in the rules?

 

[anjing merah]

There appears to be a noticeable difference in scores posted thus far with most scores being much lower. For example, even the better scores seem to just break the hundred mark. The exception was Nic Natanui who scored 177 from memory. Has there been a change to the overall system which in turn would influence the type of players selected.

 

[Fadge]

so the scoring system isn't described in the rules?

To the extent that I would like? No.

The Champion Data scoring system is like the umpires telling you at the end of a match who won, and how much by, due to the fact that different points are awarded for level of difficulty of each of the goals kicked.

 

[Fadge]

There appears to be a noticeable difference in scores posted thus far with most scores being much lower. For example, even the better scores seem to just break the hundred mark. The exception was Nic Natanui who scored 177 from memory. Has there been a change to the overall system which in turn would influence the type of players selected.

Might it have anything to do with the shorter quarters for the NAB Cup, and therefore fewer total points to be allocated?

 

[BlueVeins]

Scores are lower because there are more players.

 

[damochandler]

i don't know how it works, but i wouldn't rely on it every year to pick a brownlow winner.