Pace & Possession Rates

mdc

Cancelled
Veteran 10k Posts
Jun 26, 2007
13,666
9
Melbourne
AFL Club
Collingwood
There's been a lot of discussion recently about the game becoming too fast, and cheap possessions inflating stats. So I thought it might be interesting to have a look at just how much quicker the game is getting.

Let's define League Pace: The average number of possessions in a game of football in a given year.

Looking at the last 20 years (1990-2008) we get the following graphic:
(League Pace on the Y-axis, Year on the X-axis)

c0f6aaa1ae.jpg



Likewise, we can have a look at Team Pace: The average number of possession in a game featuring Team X.

Or in terms of an eqn;
Team Pace = (Average team possessions + Average opposition possessions)

For example, for 2008 it looked like this:

f9e0778ba5.jpg


With adjusted pace, we're comparing Team Pace to League Pace. So basically, Geelong games were 7% faster than the League Pace, Sydney games were 10% slower, and everyone else is within 4% of the league average.

Now that we have Team Pace, we can also look at Possession Rate: The frequency with which a particular player finds the ball.

Mathematically, Possession Rate = (Player X's possession average)/(Team Pace) x 1000

(note: we multiply by 1000 simply to make the figures a bit neater - this changes nothing).

For example: Joel Corey led the league last year with 29.2 possessions per game. In 2002, Scott West led the league with 26.2 possession per game. But who had the higher Possession Rate?

Corey's PR = 29.2/762.7 x 1000 = 38.3

West's PR = 26.2/605.5 x 1000 = 43.3

So despite averaging 3 fewer possessions, West's Possession Rate is actually higher by 5, which is a pretty significant difference. This is because the Dogs circa 2002 played a lot slower than Geelong circa 2008.

Below is a table of the highest Possession Rates of the last 20 years:

c7b1f99cca.jpg


So what can we glean from this? Well, Robert Harvey was a freak. We can also see that it's very hard to register a high Possession Rate in a team that plays quickly - this is logical, as it's a lot easier for a mid to get to most contests if his team is playing slower.

In fact, the only two players on that list who played for teams that could be considered as fast-paced teams are Greg Williams in 1990 (although that was in just 11 games so could probably be ignored) and McDermott in 1992.

As a final thought - where would Ablett's 38/game appear on that table if the season finished today? Well, he'd come in with a PR of just under 50, which would be good for third behind Harvey's 2 Brownlow seasons. Considering he's doing it in a team that plays much, much faster than the late-90's Saints makes it all that much more impressive.
 

Nasma

Team Captain
Sep 12, 2007
506
7
Melbourne
AFL Club
Geelong
It's also interesting to note that none of the top scorers come from recent years - suggesting that midfields share the load a lot more than they used to.
 
May 21, 2008
1,386
5
Rent Free
AFL Club
Essendon
So what can we glean from this? Well, Robert Harvey was a freak. We can also see that it's very hard to register a high Possession Rate in a team that plays quickly - this is logical, as it's a lot easier for a mid to get to most contests if his team is playing slower.

In fact, the only two players on that list who played for teams that could be considered as fast-paced teams are Greg Williams in 1990 (although that was in just 11 games so could probably be ignored) and McDermott in 1992.

As a final thought - where would Ablett's 38/game appear on that table if the season finished today? Well, he'd come in with a PR of just under 50, which would be good for third behind Harvey's 2 Brownlow seasons. Considering he's doing it in a team that plays much, much faster than the late-90's Saints makes it all that much more impressive.

The only value I see in Possession Rate is in comparing a player relative to his teammates, rather than for comparing players from different teams.
 

mdc

Cancelled
Veteran 10k Posts
Jun 26, 2007
13,666
9
Melbourne
AFL Club
Collingwood
The only value I see in Possession Rate is in comparing a player relative to his teammates, rather than for comparing players from different teams.

The whole idea of possession rate is that it adjusts for the way different teams play.

It allows us to compare across teams and across eras, which is difficult to do with raw stats like "disposals" or "marks".
 
May 21, 2008
1,386
5
Rent Free
AFL Club
Essendon
The whole idea of possession rate is that it adjusts for the way different teams play.

It allows us to compare across teams and across eras, which is difficult to do with raw stats like "disposals" or "marks".

It's a good idea but all Possession Rate really gives is the percentage of possessions that player x averages against the total possessions in a game. Team Pace is well meaning but I think it assumes that the speed with which a game is played is the only variable contributing to the total possessions in a game.

The other key variable is the standard of players on the same team as player x. It's a similar weakness to the one that makes it hard for gun players in gun teams to win the Brownlow. I'm cautious of reading too much into this percentage figure for this reason.
 

mdc

Cancelled
Veteran 10k Posts
Jun 26, 2007
13,666
9
Melbourne
AFL Club
Collingwood
It's a good idea but all Possession Rate really gives is the percentage of possessions that player x averages against the total possessions in a game.

Right.

Team Pace is well meaning but I think it assumes that the speed with which a game is played is the only variable contributing to the total possessions in a game.

I'd like you to clarify this because;
a) I don't remember ever mentioning "speed", nor do I know where I may have implicitly implied it.
b) Even if it were the assumption, I don't really see the relevance one way or the other.

The other key variable is the standard of players on the same team as player x. It's a similar weakness to the one that makes it hard for gun players in gun teams to win the Brownlow.

This is a common, yet complete misguided, assumption. Just looking at the Brownlow winners this decade, by team ladder position, we have:

2008: 3rd, 2007: 1st, 2006: 4th, 2005: 2nd, 2004: 7th, 2003: 2nd/4th/6th, 2002: 2nd, 2001: 2nd, 2000: 3rd

So clearly, playing in a good team is an advantage here, not a disadvantage.

With possession rate, I don't really believe it would make much difference one way or the other. If you look at the all-time table, there are players on good teams (including Harvey who had Burke to compete with, the latter also on the all-time table), average teams, and bad teams.

As an aside, if you look at the PR table, 5 of the top 10 won a Brownlow in that very year. ;)

I'm cautious of reading too much into this percentage figure for this reason.

I mean, it's not supposed to be some super-important measure. It's simply an improvement on the normal "possession" statistic, as it more accurately gauges who's best at finding the ball.

But I'm not for a minute suggesting that you use PR to argue Harvey > Voss or whatever.
 
May 21, 2008
1,386
5
Rent Free
AFL Club
Essendon
I'd like you to clarify this because;
a) I don't remember ever mentioning "speed", nor do I know where I may have implicitly implied it.
b) Even if it were the assumption, I don't really see the relevance one way or the other.

a) The first line of your OP implied it.
b) It's relevant because when you use team pace (FWIW I'd prefer the term Match Pace) in an equation to evaluate an individual then it discounts the standard of opposition and the standard of teammates of player x.

This is a common, yet complete misguided, assumption. Just looking at the Brownlow winners this decade, by team ladder position, we have:

2008: 3rd, 2007: 1st, 2006: 4th, 2005: 2nd, 2004: 7th, 2003: 2nd/4th/6th, 2002: 2nd, 2001: 2nd, 2000: 3rd

So clearly, playing in a good team is an advantage here, not a disadvantage.

With possession rate, I don't really believe it would make much difference one way or the other. If you look at the all-time table, there are players on good teams (including Harvey who had Burke to compete with, the latter also on the all-time table), average teams, and bad teams.

I didn't say that it couldn't happen (gun players on gun teams winning the brownlow); I just said it was harder. Here's a list of gun players on gun teams who I believe were the best in the league in that year but didn't win due to the quality of their teammates:

2008: Gary Ablett
2007: Gary Ablett (Bartel the exception)
2006: Chris Judd
2005: Chris Judd
2004: Warren Tredrea

As an aside, if you look at the PR table, 5 of the top 10 won a Brownlow in that very year. ;)

All that means is that the stat supports other existing stats for that discreet year that indicate that player x was better than his peers, but it doesn't assist in comparing player x to player y across eras.

I mean, it's not supposed to be some super-important measure. It's simply an improvement on the normal "possession" statistic, as it more accurately gauges who's best at finding the ball.

But I'm not for a minute suggesting that you use PR to argue Harvey > Voss or whatever.

Fair enough. I agree it's an interesting one but I'd be cautious to not read too much into it. It's a good job and a very good OP so sorry if it seems like I'm having a go :thumbsu:. Btw, is it just a hobby of yours or do you work in this field?
 

mdc

Cancelled
Veteran 10k Posts
Jun 26, 2007
13,666
9
Melbourne
AFL Club
Collingwood
It's relevant because when you use team pace (FWIW I'd prefer the term Match Pace) in an equation to evaluate an individual then it discounts the standard of opposition and the standard of teammates of player x.

The standard of the opposition is the same for everyone...(obviously some players play some teams twice and others don't, but whatever, it's not a big thing).

I didn't say that it couldn't happen (gun players on gun teams winning the brownlow); I just said it was harder. Here's a list of gun players on gun teams who I believe were the best in the league in that year but didn't win due to the quality of their teammates:

2008: Gary Ablett
2007: Gary Ablett (Bartel the exception)
2006: Chris Judd
2005: Chris Judd Cousins, his team-mate won
2004: Warren Tredrea He's a forward. And aforementioned Chris Judd won.



All that means is that the stat supports other existing stats for that discreet year that indicate that player x was better than his peers, but it doesn't assist in comparing player x to player y across eras.

Does your argument here boil down to the quality of players varying over eras?

If that's the case, I'd suggest that the variation is going to be very minor in a league involving 600 or so players, and if anything, the quality improves slightly with time (as is the case with almost every sport).

Fair enough. I agree it's an interesting one but I'd be cautious to not read too much into it. It's a good job and a very good OP so sorry if it seems like I'm having a go :thumbsu:. Btw, is it just a hobby of yours or do you work in this field?

I just have a maths background and interest in sports statistics. So to answer your qn - hobby.

btw, as much as I'd like to take credit for the OP, it's not really my idea per se. I simply adapted something that's been used for a long time in basketball (and probably other sports).
 
May 21, 2008
1,386
5
Rent Free
AFL Club
Essendon
The standard of the opposition is the same for everyone...(obviously some players play some teams twice and others don't, but whatever, it's not a big thing).

Does your argument here boil down to the quality of players varying over eras?

If that's the case, I'd suggest that the variation is going to be very minor in a league involving 600 or so players, and if anything, the quality improves slightly with time (as is the case with almost every sport).

I just have a maths background and interest in sports statistics. So to answer your qn - hobby.

btw, as much as I'd like to take credit for the OP, it's not really my idea per se. I simply adapted something that's been used for a long time in basketball (and probably other sports).

Yeah that's what I'm getting at. My concern was that Possession Rate overstates the impact of gun players in lower-quality eras compared to the modern greats (Ablett). When you see a circa 99 Buckley get a similar possession rate to 2009 Ablett then you know something's wrong.
 
Back