Our Age and Games Demographic

Someone needs to do up some bell curves I reckon. How many of our players are within one standard deviation of the league average? How many are more than 3 above? Since when is disregarding outliers in statistical analysis considered cherry-picking?

Quoted Post

Quoted Post
Interesting, considering our list is skewed a bit by fletch. Wouldn't be surprised if without fletch were closer to port.
Always hated this argument. If Port took out their oldest guy, they'd probably be a bit lower down as well.

He still plays. He’s still 1 of our 22 most weeks. So why shouldn’t he count? That’s called cherry picking.

I’ll run with ‘it’s a stats thing’

Mean, median and mode are all simply measures of central tendency. Which can be heavily skewed by outlier one off result. Hence house values are measured in median rather than mean.

It’s interesting to see the average, but generally a poor statistical representation in this case. Hence my comment.

Someone needs to do up some bell curves I reckon. How many of our players are within one standard deviation of the league average? How many are more than 3 above? Since when is disregarding outliers in statistical analysis considered cherry-picking?
When you only disregard outliers at one or two clubs and not the rest. IMHO.

If anyone wants to run actual proper stats, and set a threshold for outliers for all clubs - then go for it. But don’t only do it for us and North and ignore everyone else.

Quoted Post

Quoted Post
Quoted Post
Interesting, considering our list is skewed a bit by fletch. Wouldn't be surprised if without fletch were closer to port.
Always hated this argument. If Port took out their oldest guy, they'd probably be a bit lower down as well.

He still plays. He’s still 1 of our 22 most weeks. So why shouldn’t he count? That’s called cherry picking.

I’ll run with ‘it’s a stats thing’

Mean, median and mode are all simply measures of central tendency. Which can be heavily skewed by outlier one off result. Hence house values are measured in median rather than mean.

It’s interesting to see the average, but generally a poor statistical representation in this case. Hence my comment.

When about 98% of ~800 results are between 18 and 31 - and only 2 are over 35 - it's clearly a very very different set of numbers to the housing market, where you could have the top price could be 10 or 20 times larger than the cheapest.

Quoted Post

Quoted Post
Quoted Post
Interesting, considering our list is skewed a bit by fletch. Wouldn't be surprised if without fletch were closer to port.
Always hated this argument. If Port took out their oldest guy, they'd probably be a bit lower down as well.

He still plays. He’s still 1 of our 22 most weeks. So why shouldn’t he count? That’s called cherry picking.

I’ll run with ‘it’s a stats thing’

Mean, median and mode are all simply measures of central tendency. Which can be heavily skewed by outlier one off result. Hence house values are measured in median rather than mean.

It’s interesting to see the average, but generally a poor statistical representation in this case. Hence my comment.

When about 98% of ~800 results are between 18 and 31 - and only 2 are over 35 - it's clearly a very very different set of numbers to the housing market, where you could have the top price could be 10 or 20 times larger than the cheapest.

Looking at starting team age, or best 25 or something is definitely more valuable than the age of the whole list. Every team drafts at least 3 players, normally more if you count the rookie list, which means 20% or so of everybody’s list is first and second year players, almost all of whom are 18-19 years old. Unsurprisingly, this flattens out squad ages to the point where it’s borderline meaningless. Last year the average age difference for entire squads between 6th (us) and 12th (Richmond) was 4 months.

Personally, I’d probably bucket player ages 18-22, 23-26, 27-30, 30+, or something like that. Ditto with games played, after all does a player with 350 games of experience really bring anything more than someone with 300?

But I’m not actually going to do it because screw that.

Edit: I also think our age profile has been skewed slightly older due to the draft sanctions. We’ve certainly historically held on to players for too long, but I think it’s been exacerbated by having less available good picks the last couple of years, so we’ve kept guys on the list that we normally would have delisted.

Looking at starting team age, or best 25 or something is definitely more valuable than the age of the whole list. Every team drafts at least 3 players, normally more if you count the rookie list, which means 20% or so of everybody's list is first and second year players, almost all of whom are 18-19 years old. Unsurprisingly, this flattens out squad ages to the point where it's borderline meaningless. Last year the average age difference for entire squads between 6th (us) and 12th (Richmond) was 4 months.
I just don't get this line of reasoning that "we need to look at all the numbers so let's ignore some of the numbers". Either with the oldies, or with youngsters. The numbers are the whole point!! We have 4 1st & 2nd year teenagers - 7 including rookies. Some clubs would have bucketloads (Dogs have 10/11, Brisbane have 10/15). Massive difference. When we have an injury we're calling in a Gwilt or Howlett, they're likely calling in a debutant.

Quoted Post

Quoted Post
Quoted Post
Interesting, considering our list is skewed a bit by fletch. Wouldn't be surprised if without fletch were closer to port.
Always hated this argument. If Port took out their oldest guy, they'd probably be a bit lower down as well.

He still plays. He’s still 1 of our 22 most weeks. So why shouldn’t he count? That’s called cherry picking.

its not cherry picking

for the sake of getting more accurate averages, these sorts of players should be left out

im not talking people who are 33/34, but maybe someone like harvey, fletch who are considerably older than the teams next oldest player

It’s simple statistics that I learned at uni. Fletch is an outlier. It’s not just about taking out the oldest bloke, it’s the fact that he is so far older than anyone else on the list, some 6 years older than Chappy, that he is a statistical anomaly and skews the average age for our list. Take him out and replace him with a draftee who is 18/19 and the average age on the list drop 6 months, which then puts us around AFL average (8th/9th).

I was always taught that the mean (average) age was not as meaningful stat as the median (the absolute middle age where 50% are over this figure and %50 are under) and also the interquartile ranges, which is the difference between the upper and lower quartiles. Take Geelong for instance, they have an average list age below us, yet they have 9 players above 30. We only have 4.

Just goes to show how consistent and reliable Stanton has been

Fletch and Harvey are not particularly meaningful statistical outliers for age. On games played, they add about 10 games to the average of the list, which isn’t significant either when you consider the vatiation of this stat across the league.

As HAP said, taking Fletch out is “cherry picking”. What about Goodes and Pavlich who have all played 300+.

Quoted Post

Quoted Post
Quoted Post
Quoted Post
Interesting, considering our list is skewed a bit by fletch. Wouldn't be surprised if without fletch were closer to port.
Always hated this argument. If Port took out their oldest guy, they'd probably be a bit lower down as well.

He still plays. He’s still 1 of our 22 most weeks. So why shouldn’t he count? That’s called cherry picking.

I’ll run with ‘it’s a stats thing’

Mean, median and mode are all simply measures of central tendency. Which can be heavily skewed by outlier one off result. Hence house values are measured in median rather than mean.

It’s interesting to see the average, but generally a poor statistical representation in this case. Hence my comment.

When about 98% of ~800 results are between 18 and 31 - and only 2 are over 35 - it's clearly a very very different set of numbers to the housing market, where you could have the top price could be 10 or 20 times larger than the cheapest.

Yes you are correct, median possibly is again not the best. But I never said it was a good fit, just that house prices are skewed and hence why they us a different test in the median.

You are correct, until somebody does a full stats work up on the age and games of players at clubs and whether that is a statistical measure of the GF success then we will never know if it’s worth publishing. It’s why I typically go with a ‘that’s interesting’ but there are problems comment.

Apologise if my fletcher comment hit a statistical nerve!

Quoted Post

Fletch and Harvey are not particularly meaningful statistical outliers for age. On games played, they add about 10 games to the average of the list, which isn't significant either when you consider the vatiation of this stat across the league.

As HAP said, taking Fletch out is “cherry picking”. What about Goodes and Pavlich who have all played 300+.

if you keep the discussion confined to age, which is the measurement in contention, neither goodes nor pavlich would be outliers i believe

Quoted Post

Quoted Post
Quoted Post
Quoted Post
Interesting, considering our list is skewed a bit by fletch. Wouldn't be surprised if without fletch were closer to port.
Always hated this argument. If Port took out their oldest guy, they'd probably be a bit lower down as well.

He still plays. He’s still 1 of our 22 most weeks. So why shouldn’t he count? That’s called cherry picking.

its not cherry picking

for the sake of getting more accurate averages, these sorts of players should be left out

im not talking people who are 33/34, but maybe someone like harvey, fletch who are considerably older than the teams next oldest player

It’s simple statistics that I learned at uni. Fletch is an outlier. It’s not just about taking out the oldest bloke, it’s the fact that he is so far older than anyone else on the list, some 6 years older than Chappy, that he is a statistical anomaly and skews the average age for our list. Take him out and replace him with a draftee who is 18/19 and the average age on the list drop 6 months, which then puts us around AFL average (8th/9th).

I was always taught that the mean (average) age was not as meaningful stat as the median (the absolute middle age where 50% are over this figure and %50 are under) and also the interquartile ranges, which is the difference between the upper and lower quartiles. Take Geelong for instance, they have an average list age below us, yet they have 9 players above 30. We only have 4.

Outliers can only discarded if you can establish them as being an outliers, given North have Harvey and there are a few others that are roaming around, but lesser obviously, it may prove difficult.

As I said above, a full stats work up would establish this and it’s relevance as a causal measure rather than just a correlating statistic.

Somehow I think if you go back in time and look at games/age it may prove difficult to find a causal relationship.

Quoted Post

Looking at starting team age, or best 25 or something is definitely more valuable than the age of the whole list. Every team drafts at least 3 players, normally more if you count the rookie list, which means 20% or so of everybody's list is first and second year players, almost all of whom are 18-19 years old. Unsurprisingly, this flattens out squad ages to the point where it's borderline meaningless. Last year the average age difference for entire squads between 6th (us) and 12th (Richmond) was 4 months.
I just don't get this line of reasoning that "we need to look at all the numbers so let's ignore some of the numbers". Either with the oldies, or with youngsters. The numbers are the whole point!! We have 4 1st & 2nd year teenagers - 7 including rookies. Some clubs would have bucketloads (Dogs have 10/11, Brisbane have 10/15). Massive difference. When we have an injury we're calling in a Gwilt or Howlett, they're likely calling in a debutant.

I’m not saying “ignore the young players”, I’m saying is that the age of the team that actually runs out (the stat you used) is more important than the average age of the total list. If people want to talk about list age, then they should absolutely use the whole list, I just don’t think average list age is important or interesting, at least in part because the AFL mandates that list age skews lower than week to week team age.

Edit: obviously you’d have to exclude some players, eg Hogan and Martin in 2013, but that’s a pretty specific exception.

Quoted Post

Quoted Post
Looking at starting team age, or best 25 or something is definitely more valuable than the age of the whole list. Every team drafts at least 3 players, normally more if you count the rookie list, which means 20% or so of everybody's list is first and second year players, almost all of whom are 18-19 years old. Unsurprisingly, this flattens out squad ages to the point where it's borderline meaningless. Last year the average age difference for entire squads between 6th (us) and 12th (Richmond) was 4 months.
I just don't get this line of reasoning that "we need to look at all the numbers so let's ignore some of the numbers". Either with the oldies, or with youngsters. The numbers are the whole point!! We have 4 1st & 2nd year teenagers - 7 including rookies. Some clubs would have bucketloads (Dogs have 10/11, Brisbane have 10/15). Massive difference. When we have an injury we're calling in a Gwilt or Howlett, they're likely calling in a debutant.
I'm not saying "ignore the young players", I'm saying is that the age of the team that actually runs out (the stat you used) is more important than the average age of the total list. If people want to talk about list age, then they should absolutely use the whole list, I just don't think average list age is important or interesting, at least in part because the AFL mandates that list age skews lower than week to week team age.

Edit: obviously you’d have to exclude some players, eg Hogan and Martin in 2013, but that’s a pretty specific exception.


I think they’re all important to some degree - if not quite as important as each other.
ie a side that has a bunch of 20-25 year olds outside the 22 could reasonably expect to get more output than a team that’s only got 18-19 year olds (ie better depth), and probably look a little more likely to maintain or improve in the medium term future.
If I had heaps and heaps of time, and could be arsed, I’d make up a mega spreadsheet that does all this stuff - mean/median, games and age, most experienced 22 & list minus outliers & whole list.

Far more fun just to argue about it!

Not for the rest of us.

o
m
g

Quoted Post

Quoted Post
Quoted Post
Looking at starting team age, or best 25 or something is definitely more valuable than the age of the whole list. Every team drafts at least 3 players, normally more if you count the rookie list, which means 20% or so of everybody's list is first and second year players, almost all of whom are 18-19 years old. Unsurprisingly, this flattens out squad ages to the point where it's borderline meaningless. Last year the average age difference for entire squads between 6th (us) and 12th (Richmond) was 4 months.
I just don't get this line of reasoning that "we need to look at all the numbers so let's ignore some of the numbers". Either with the oldies, or with youngsters. The numbers are the whole point!! We have 4 1st & 2nd year teenagers - 7 including rookies. Some clubs would have bucketloads (Dogs have 10/11, Brisbane have 10/15). Massive difference. When we have an injury we're calling in a Gwilt or Howlett, they're likely calling in a debutant.
I'm not saying "ignore the young players", I'm saying is that the age of the team that actually runs out (the stat you used) is more important than the average age of the total list. If people want to talk about list age, then they should absolutely use the whole list, I just don't think average list age is important or interesting, at least in part because the AFL mandates that list age skews lower than week to week team age.

Edit: obviously you’d have to exclude some players, eg Hogan and Martin in 2013, but that’s a pretty specific exception.


I think they’re all important to some degree - if not quite as important as each other.
ie a side that has a bunch of 20-25 year olds outside the 22 could reasonably expect to get more output than a team that’s only got 18-19 year olds (ie better depth), and probably look a little more likely to maintain or improve in the medium term future.
If I had heaps and heaps of time, and could be arsed, I’d make up a mega spreadsheet that does all this stuff - mean/median, games and age, most experienced 22 & list minus outliers & whole list.

Far more fun just to argue about it!

Or you could do the right thing and just admit you are wrong!

Quoted Post

Put in langford for fletcher its the only way.
Why not? Like for like.

Quoted Post

Quoted Post
Quoted Post
Quoted Post
Looking at starting team age, or best 25 or something is definitely more valuable than the age of the whole list. Every team drafts at least 3 players, normally more if you count the rookie list, which means 20% or so of everybody's list is first and second year players, almost all of whom are 18-19 years old. Unsurprisingly, this flattens out squad ages to the point where it's borderline meaningless. Last year the average age difference for entire squads between 6th (us) and 12th (Richmond) was 4 months.
I just don't get this line of reasoning that "we need to look at all the numbers so let's ignore some of the numbers". Either with the oldies, or with youngsters. The numbers are the whole point!! We have 4 1st & 2nd year teenagers - 7 including rookies. Some clubs would have bucketloads (Dogs have 10/11, Brisbane have 10/15). Massive difference. When we have an injury we're calling in a Gwilt or Howlett, they're likely calling in a debutant.
I'm not saying "ignore the young players", I'm saying is that the age of the team that actually runs out (the stat you used) is more important than the average age of the total list. If people want to talk about list age, then they should absolutely use the whole list, I just don't think average list age is important or interesting, at least in part because the AFL mandates that list age skews lower than week to week team age.

Edit: obviously you’d have to exclude some players, eg Hogan and Martin in 2013, but that’s a pretty specific exception.


I think they’re all important to some degree - if not quite as important as each other.
ie a side that has a bunch of 20-25 year olds outside the 22 could reasonably expect to get more output than a team that’s only got 18-19 year olds (ie better depth), and probably look a little more likely to maintain or improve in the medium term future.
If I had heaps and heaps of time, and could be arsed, I’d make up a mega spreadsheet that does all this stuff - mean/median, games and age, most experienced 22 & list minus outliers & whole list.

Far more fun just to argue about it!

Or you could do the right thing and just admit you are wrong!

He is not, and can’t be assed proving it, which also gives the possibility he could be wrong.