Analysis How much does experience in a side really matter when chasing the flag?

[Mattrox]

As people know, I have been collating this seasons data on age and total games experience fielded by each Club for all rounds up until now.

I have done a scatter-plot and linear regression on the data in an attempt to quantify how much experience really matters for a Club to be a flag contender.

Here are some assumptions I have made.

Age does not dictate experience. A mature age recruit, a player with a serious or multiple injuries will not have played as many games relative to other players their age. Also, depth players who are above state league level, but not quite AFL level will have less games as well.

Talent, skill and execution is an important factor. This is not directly measured in this data. However, players who are highly skilled and talented will debut at a younger age, play more often (if injury doesn't strike) and last longer on a team list. Teams with more experienced player are likely to have a higher number of skilled and talented players as they have remained on the list all this time.

Game plan or ability to execute a game plan is also a factor. This is ignored virtually completely by using this data.

The results:
I have graphed in a scatter plot the number of wins vs age and number of wins versus experience. In each case, wins is the dependent variable. Age and aggregate games played are the independent variable in each case. A linear regression was performed and the R-squared value claculated. The R-square value determines how good a fit the line is to the scattered points.

The 1st graph is all teams Wins vs Age and Wins Vs Experience.


There is only a weak correlation between age and wins. The correlation between experience and wins is a little better. The R-squared value indicates that experience has a 48% contributing factor to the number of wins. Other factors such as talent, skill, fitness and game plan etc also contribute to the number of wins.

What sticks out to me a lot is that Gold Coast on 9 wins and less than 1500 games experience on average is sitting well above the line. Gold Coast had priority access to the draft and also got Gary Ablett. It could be argued that skill and talent in Gold Coast's side are disproportionate to their collective experience. IE their side is stacked with talent. (GWS is also sacked with talent, but they are hugely inexperienced compared to every other side and they don't have Gary Ablett)

The other team is St Kilda. With 3 wins but approaching 1855 games experience on average they lie well below the line. They have far fewer wins than expected according to their level of experience. It could be argued that they lack talent in their team. They traded away a very good ruck and are relying on very much less talent in that division at least. Is there some hangover from Lyon not developing younger players on the list very well in his pursuit of a premiership.

I decided to treat these two teams as outliers and re-do the graphs without them. Something interesting happened.


In this new graph the correlation between experience and wins is stronger. Experience, as a factor accounts for 64% of the wins.

Bearing in mind, teams with large numbers of experienced players (eg 200+ games played) will have more talent as these experienced players only last because of their footballing ability. That is most likely why they lie above the line.

Why did I remove 2 teams for the second graph? I wanted to look at what happens "on average" without exceptional circumstances. I accept that GWS had special circumstances too, but their experience level is really, really low. They still do lie above the line, which means on average they are winning more games than expected for their level of experience.

BTW the Crows are the only team on 8 wins. So you can see where they sit with respect to the line.

Here is the table of data.



* Edit. This is not to be reproduced in print or electronically outside Bigfooty.

 

[Vader]

This analysis tells you how important it is, but if you really want to know what impact it has on winning the flag, then maybe a statistical examination of the last 10-20 years worth of Grand Final contenders might be of value?

 

[Mattrox]

This analysis tells you how important it is, but if you really want to know what impact it has on winning the flag, then maybe a statistical examination of the last 10-20 years worth of Grand Final contenders might be of value?

That is next.

I know from looking at 2007 there was a huge experience mismatch.

 

[Vader]

Nice analysis by the way. It's clear that experience is very important, but it's not the be-all and end-all. Talent has to play a part, as does good coaching.

Age is of course strongly correlated with experience, with older players generally being more experienced. For every JPod who made his AFL debut at 28, there are 50 players who have 150-200 AFL games under their belt by that age.

 

[Hank93]

As people know, I have been collating this seasons data on age and total games experience fielded by each Club for all rounds up until now.

I have done a scatter-plot and linear regression on the data in an attempt to quantify how much experience really matters for a Club to be a flag contender.

Here are some assumptions I have made.

Age does not dictate experience. A mature age recruit, a player with a serious or multiple injuries will not have played as many games relative to other players their age. Also, depth players who are above state league level, but not quite AFL level will have less games as well.

Talent, skill and execution is an important factor. This is not directly measured in this data. However, players who are highly skilled and talented will debut at a younger age, play more often (if injury doesn't strike) and last longer on a team list. Teams with more experienced player are likely to have a higher number of skilled and talented players as they have remained on the list all this time.

Game plan or ability to execute a game plan is also a factor. This is ignored virtually completely by using this data.

The results:
I have graphed in a scatter plot the number of wins vs age and number of wins versus experience. In each case, wins is the dependent variable. Age and aggregate games played are the independent variable in each case. A linear regression was performed and the R-squared value claculated. The R-square value determines how good a fit the line is to the scattered points.

The 1st graph is all teams Wins vs Age and Wins Vs Experience.
View attachment 68205

There is only a weak correlation between age and wins. The correlation between experience and wins is a little better. The R-squared value indicates that experience has a 48% contributing factor to the number of wins. Other factors such as talent, skill, fitness and game plan etc also contribute to the number of wins.

What sticks out to me a lot is that Gold Coast on 9 wins and less than 1500 games experience on average is sitting well above the line. Gold Coast had priority access to the draft and also got Gary Ablett. It could be argued that skill and talent in Gold Coast's side are disproportionate to their collective experience. IE their side is stacked with talent. (GWS is also sacked with talent, but they are hugely inexperienced compared to every other side and they don't have Gary Ablett)

The other team is St Kilda. With 3 wins but approaching 1855 games experience on average they lie well below the line. They have far fewer wins than expected according to their level of experience. It could be argued that they lack talent in their team. They traded away a very good ruck and are relying on very much less talent in that division at least. Is there some hangover from Lyon not developing younger players on the list very well in his pursuit of a premiership.

I decided to treat these two teams as outliers and re-do the graphs without them. Something interesting happened.
View attachment 68207

In this new graph the correlation between experience and wins is stronger. Experience, as a factor accounts for 64% of the wins.

Bearing in mind, teams with large numbers of experienced players (eg 200+ games played) will have more talent as these experienced players only last because of their footballing ability. That is most likely why they lie above the line.

Why did I remove 2 teams for the second graph? I wanted to look at what happens "on average" without exceptional circumstances. I accept that GWS had special circumstances too, but their experience level is really, really low. They still do lie above the line, which means on average they are winning more games than expected for their level of experience.

BTW the Crows are the only team on 8 wins. So you can see where they sit with respect to the line.

Here is the table of data.

View attachment 68210

* Edit. This is not to be reproduced in print or electronically outside Bigfooty.

Great stuff man, think you should reconsider the last bit though. This needs to be seen by the masses. Very interesting data.

 

[Mattrox]

Great stuff man, think you should reconsider the last bit though. This needs to be seen by the masses. Very interesting data.

Yeah, but without that last bit some bottom feeding journo will pick bits and skew it for their own purposes.

I'm still working on it and it needs a whole season as there might br wins that St Kilda are still to get that they haven't played yet which moves them up to the pack etc.

Sides that lie below the line by a "long way" have concerns because they are under performing.

 

[1970crow]

Nice try, but I still think we should drop pods for structural reasons.

 

[Mattrox]

Nice try, but I still think we should drop pods for structural reasons.

That's cool. One player doesn't do too much to the numbers over 22 players.


But this model explains why Port and Collingwood's drop off isn't entirely unexpected.

 

[Mattrox]

This table shows the age and experience in games played of the Grand Finalists since 2001. On only 2 occasions did the Premier have less than 2500 games experience. On only 3 occasions did the team with lesser experience won. Now I'm not a betting man, but given that 77% of the time the more experienced side wins, does this give an advantage when betting on the Premiership?

 

[1970crow]

That's cool. One player doesn't do too much to the numbers over 22 players.


But this model explains why Port and Collingwood's drop off isn't entirely unexpected.

sorry mate, just making an irrelevant joke at my own expense.

i do like what you've done though, it forms a sound starting point in which to assess the relative performance of clubs. not sure about the above though, wouldn't you have to assume that the early season wins were when the age/experience were at their higher levels. whilst I think that's likely (port losing jacko/fatso, not sure about the pies though), I don't think that the graphs differentiate from game to game or periods within the total number of losses. although, once you've formed your hypothesis, I'm sure that it would be very easy to get the data to support. again, it really does seem like an excellent beginning point.

 

[beartoo]

Brilliant Mattrox, well done.

 

[Iron]

I love these sorts of statistical breakdowns. What is of particular interest is not that we are the average (we love to be ordinary) but that Carlton are floundering as an outlier. Pretty experienced, but not getting results. Lack of talent, lack of mental drive, lack of everything. Dare say there's a massive cleanout on the horizon.

 

[Mattrox]

I love these sorts of statistical breakdowns. What is of particular interest is not that we are the average (we love to be ordinary) but that Carlton are floundering as an outlier. Pretty experienced, but not getting results. Lack of talent, lack of mental drive, lack of everything. Dare say there's a massive cleanout on the horizon.

St Kilda is looking bad too. As is Richmond.

I dare say if Freo don't win this year or next, that they will have been Lyoned as well.



We have actually used 36 players so far. The only teams on this number or greater are the bottom teams. Brisbane 37, GWS 40, Melbourne 36, Richmond 36, St Kilda 41, Western Bulldogs 39.

Gold Coast has used 32, they are pumping games into fewer players.

Geelong has played 35, experienced side blooding new players, in addition to injuries.

Hawthorn, 34, As above.

Freo, 33.

Sydney, 30. A very stable side.


Our list of inexperienced players who would make our best 22 regularly.

Luke Brown 40
Brad Crouch 19
Josh Jenkins 42
Sam Kerridge 25
Rory Laird 31
Tom Lynch 35
(Brodie Martin 29) - Depth next year I reckon

Players 50 - 100 games
Ricky Henderson 58
Sam Jacobs 98
Matthew Jaensch 61
Andy Otten 76
(James Podsiadly 99) won't be there much longer
Rory Sloane 94
Brodie Smith 70
Daniel Talia 70
Taylor Walker 78
Matthew Wright 73

That is 17, yes 17, of those in contention for our best 22 under 100 games. We are a very inexperienced side.

Games experience lost next year
Rutten 227
Reilly 202 - arguably the game has gone past him, injury at wrong stage of his career.
Porplyzia 130 - cruelled by injury

Next year we lose ~560 games from the side but only add 484 over the course of the year. Based purely on games experience we would be doing very well to make the top 4. To be contending we really need 2400 - 2500 about 400 to 500 more games experience.

Of course this may be mitigated via trading and/or Free Agent pick ups.

 

[Mattrox]

Playing the Kids under Sando.

Jarryd Lyons Debut: 2012 19y 282d Games 18
Sam Kerridge Debut: 2012 18y 355d Games 25
Luke Brown Debut: 2012 19y 297d Games 40
Brad Crouch Debut: 2013 19y 83d Games 19
Rory Laird Debut: 2013 19y 113d Games 31
Matt Crouch Debut: 2014 18y 349d Games 7
Charlie Cameron Debut: 2014 19y 314d Games 4


"Not Kids" debuting under Sando
Kyle Hartigan Drafted in 2012 Rookie Draft Debut: 2013 21y 262d Games 10
Josh Jenkins Drafted by Essendon in 2010 Rookie Draft. Debut: 2012 23y 95d Games 42
Cam Ellis-Yolmen Drafted 2011 National Draft Debut: 2014 21y 132d Games 1
Mitch Grigg Drafted 2011 ND Debut: 2013 20y 201d Games 10

 

[Smithalicious]

Makes you wonder if CEY will be on the list next year. Everyone else has some games behind them except CEY.

Loving the analysis mate, absolutely awesome.

 

[mattymac]

Regressions aren't great for this sort of thing. You can show an association between the decline of pirates and global warming using regression analyses.

But yes, I don't think there was ever any doubt that experience is key and the reason we have been poor is due to very poor recruiting in the early 2000s we lack that experienced senior cohort that the best sides have. Our best seniors are Thommo and Rutten both at retirement age, behind them is virtually nothing until a glut of talented 22-24 year olds. When they get to their late 20s we will be a juggernaut if we can hold onto them

 

[Mattrox]

Regressions aren't great for this sort of thing. You can show an association between the decline of pirates and global warming using regression analyses.

But yes, I don't think there was ever any doubt that experience is key and the reason we have been poor is due to very poor recruiting in the early 2000s we lack that experienced senior cohort that the best sides have. Our best seniors are Thommo and Rutten both at retirement age, behind them is virtually nothing until a glut of talented 22-24 year olds. When they get to their late 20s we will be a juggernaut if we can hold onto them

Except pirates and globsl warming are completely unrelated variables.

Games experience by a team and wins by the team are related variables because they are the same team. It's a perfect example to use regression analysis.

Of course correlation is not causation. Being experienced doesn't just make a win happen as explained in the 1st post.

 

[mattymac]

Except pirates and globsl warming are completely unrelated variables.

Games experience by a team and wins by the team are related variables because they are the same team. It's a perfect example to use regression analysis.

Of course correlation is not causation. Being experienced doesn't just make a win happen as explained in the 1st post.

ANOVA that shit

FWIW I agree with all your discussion. Its accepted wisdom that you need an experienced senior cohort in their late 20s to mount a flag tilt, but they have to have earned that experience with talent not just bradbury it. Its also accepted wisdom that Lyons let St Kilda wilt on the vine in his short sighted flag lust.

 

[Mattrox]

ANOVA that shit

FWIW I agree with all your discussion. Its accepted wisdom that you need an experienced senior cohort in their late 20s to mount a flag tilt, but they have to have earned that experience with talent not just bradbury it. Its also accepted wisdom that Lyons let St Kilda wilt on the vine in his short sighted flag lust.

No pretty graphs with ANOVA.

 

[spexau]

This is some moneyball shit right here

 

[jenny61_99]

View attachment 68319
This table shows the age and experience in games played of the Grand Finalists since 2001. On only 2 occasions did the Premier have less than 2500 games experience. On only 3 occasions did the team with lesser experience won. Now I'm not a betting man, but given that 77% of the time the more experienced side wins, does this give an advantage when betting on the Premiership?

Were their teams that didn't make the grand final that had more experience than the two that did?

 

[Mattrox]

Were their teams that didn't make the grand final that had more experience than the two that did?

Probably, particularly when the GF team has low 2000 games exp. I should go and look at the other 2 prelim finalists too. But its trawling through AFL tables and will take ages.

I do intend on looking at AFC in finals years though.

 

[Mattrox]

This is some moneyball shit right here

This could be why Sydney has historically traded for servicable players with experience and stuck them in a role.

Looks like Richmond has tried this too.

 

[fabs33]

I believe experience is vital towards wining a premiership. Which is why I want to see us at this moment in time putting games into the players we see as part of our next premiership challenging team, i.e. the players who will still be at the club when we approach the experience level associated with challenging, at the expense of senior players whose performances are barely above or equal to that of their younger peers. I realise it's somewhat cruel, but it's necessary.

 

[Draft Bored]

Wow. Interesting stats.