AFL 2019 Round 5: Western Bulldogs v Carlton, 1:10pm AEST MS

[Ron The Bear]

That's because it's a flawed debate. It places too high a value on players who are past their best such as Simpson, Murphy, Thomas, Kreuzer and Ed Curnow ... a value higher than 100+ game players in their footballing prime.

Well that's footy, isn't it? If an older team isn't winning games, it becomes a younger team.

 

[The Old Dark Navy's]

Well that's footy, isn't it? If an older team isn't winning games, it becomes a younger team.

We are a younger team. It's just that we have a dearth of 24-27 year olds so have had to retain older players to provide experience and guidance.

I don't think your observation goes anyway towards addressing the imbalance in your method.

 

[Ron The Bear]

We are a younger team. It's just that we have a dearth of 24-27 year olds so have had to retain older players to provide experience and guidance.

I don't think your observation goes anyway towards addressing the imbalance in your method.

Sure, a younger team that was older than the opposition.

Not sure why you lot want to nit-pick. Thought you’d be happy winning for a change.

 

[Monkey King]

Variance provides extra information, that's all.

Not at all. Variance doesn’t just provide extra information, but information that is required for most statistical formulas. Without it, the picture is too incomplete, because the mean of a sample doesn’t tell us a whole lot by itself.

I’m a real fan of your work, so don’t get me wrong. All I’m saying is plot those ages out and Carlton spike highest and higher than the Dogs in the youngest area around 18-21; the Bulldogs spike higher and highest where Carlton dip at 22-27; and then drop off 28+ where Carlton rise again. Carlton are both younger and older, with the Dogs with more in younger end of the middle group. In AFL terms, I’m not convinced on face value that mean age represents the “younger” team here, and if it does, it’s not by much.

And I appreciate the data you’ve just put up. There sure does appear a correlation between mean +age and win%. I think it would be even more revealing if the z-scored age was plotted out instead of just averages, because there’s a lot of info your awesome data set is still hiding.

I’m not sure what the second table means. If it’s older teams with less variance in age then there’s more genuine older fellas in there, which is maybe why it doesn’t go as deep in + years, coz everyone starts rebuilding by then. I’d also guess it’s why they don’t have a linear win%, because it would include teams who didn’t got their age demographics right, hence that last line appears to be the influx of youth line for genuinely older teams.

 

[ManOfClay]

Sure, a younger team that was older than the opposition.

Not sure why you lot want to nit-pick. Thought you’d be happy winning for a change.

Because they need to pump up their win as much as possible, it’s a bit weird. Carlton did great and deserved their win. No matter how inexperienced their opponents were.

Now nitpicking because you can’t stand the other team being called younger because you feel that detracts from the overall win makes you look sad.

 

[Ron The Bear]

I’m not sure what the second table means. If it’s older teams with less variance in age then there’s more genuine older fellas in there, which is maybe why it doesn’t go as deep in + years, coz everyone starts rebuilding by then. I’d also guess it’s why they don’t have a linear win%, because it would include teams who didn’t got their age demographics right, hence that last line appears to be the influx of youth line for genuinely older teams.

Generally there's some relationship between average age and variance, since a large variance in team ages almost always indicates skewness at the older end.

The three highest age variances for the round belonged to Carlton, Port & Geelong; the three lowest were St.Kilda, Collingwood & Adelaide. All winners.

In all games, the team with the greater age variance wins 50.6% of the time. On its own, variance just isn't a powerful statistic. Maybe in combination with something else it would become relevant but for now, Carlton being +0.90 on average age says a lot more to me.

 

[Monkey King]

Generally there's some relationship between average age and variance, since a large variance in team ages almost always indicates skewness at the older end.

The three highest age variances for the round belonged to Carlton, Port & Geelong; the three lowest were St.Kilda, Collingwood & Adelaide. All winners.

In all games, the team with the greater age variance wins 50.6% of the time. On its own, variance just isn't a powerful statistic. Maybe in combination with something else it would become relevant but for now, Carlton being +0.90 on average age says a lot more to me.

Yeah I’d imagine greater variance skews to the older side because the mean age is what, about 23yo odd?

Absolutely, variance doesn’t say much without the mean. And the mean of a sample is not very powerful without the variance either. Which is why this paring are the foundation of so many statistical tests.

A quick example and then I’ll stop banging on about it. If you had safety parameters to test on equipment, the mean alone wouldn’t help you. The variance alone wouldn’t help you. Put them together and you can actually test if the equipment is operating within parameters.

Anyway, as I said, fan of your work. You’re the BF stats king.

 

[Ron The Bear]

A quick example and then I’ll stop banging on about it. If you had safety parameters to test on equipment, the mean alone wouldn’t help you. The variance alone wouldn’t help you. Put them together and you can actually test if the equipment is operating within parameters.

I don't doubt it. Horses for courses.

 

[Monkey King]

I don't doubt it. Horses for courses.

More about appreciating what certain data can and cannot tell us and how we can get it to tell us more.

Anyway cheers.

 

[Outback Footy]

Stats geeks, maybe start your own thread?

PotAto Potarto, TamAto Tomarto, we were soundly beaten. It really doesn't matter if we are slightly younger or slightly older. I expected much better and Im disappointed but congrats Blues, you wanted that much more than my boys did, and that's the bottom line to me.

Oh and I wish Bevo would put more height in defense and push caleb into the half forward line and add another ruck and...

Again, congrats Blues.

 

[JustaBattler]

Sure, a younger team that was older than the opposition.

Not sure why you lot want to nit-pick. Thought you’d be happy winning for a change.

I don't think normative statistics has much relevance to the discussion ( ie mean variance analysis ) - given the fact that sample sizes are less than 30...
Leaving that aside for discussion's sake- another problem with comparisons as far as age goes is that with Carlton there is a strong binodal distribution to consider as well - overlayed with a strong skewness and Kurtosis - again not enough of a sample size to say much - but the eyeball test of any chart descriptor would confirm...Carlton is fielding a few (very) old players and a lot more (very) young players and games played than typical .

So basically I am agreeing on the basis of 'text book' with the poster above who stated potatoes portatos etc....

 

[Ron The Bear]

Sheesh, no shortage of people wanting to take issue with a simple post, the likes of which I've been putting up for years. There are two rules of thumb concerning critics that are more reliable than the figures themselves:

i) people who want to argue the point are invariably aggrieved that the figures don't post their club in a better light
ii) they never come back to acknowledge the fact after they've been proven wrong

It's valid to point out that the distribution of Carlton's 22 was an unusual case. But so too was the Dogs' - three players older than 26 (Dickson/Wood/Lloyd, 31/29/29).

Re the above: it wasn't that long ago that I studied stats and although there is of course no set number that constitutes a statistically valid sample size, the figure my instructors taught was 20. Educators now teach that boys can be girls, so if they're also saying 20 = 30, who am I to argue?

Whatever the case, I will always point any doubters to the table of results for older teams, which indisputably depicts a linear relationship between age advantage and win percentage over 123 seasons. The table is less relevant for individual games but over a period of time, if your club is below that line then it suffers from substandard list management and/or substandard player development and/or substandard coaching. You can argue all you like, but you will be wrong.

A couple of examples...

In 2016, Carlton won 6 out of 7 games in April/May and there were threads lauding Bolton as a supercoach. So I say hang on, Carlton has one of the oldest 22's in the competition, they are winning games by virtue of more seasoned bodies/more experience than the opposition. These senior players failed to take the club anywhere under Malthouse and they will fail under Bolton. No, you're a ******* idiot they cry. We're winning and might is right. You're just jealous, stick to commenting on your own useless ******* team, Bolts is da best! Seeing similar here; a win over the youngest team of the season has been pumped up out of all proportion by the media. Just reinforces my belief that you lot haven't suffered enough!

In 2012 I had a running battle with St.Kilda supporters who came from the trees to attack my assertion that sitting at the bottom of the eight with the oldest team in the comp was a sure sign they were going over the cliff. They pissed and moaned and spat, even after Watters' disastrous trading plunged them into the abyss from which they're still trying to crawl. No apologies, no regrets forthcoming.

The methodology is not flawless and if you stick your neck out in footy, you're going to be wrong a lot of the time. But I'll be wrong less often than you.

NB I'm not protective of my data and have handed out the database to more people than you could shake a stick at. If anyone knows their way around SQL Server and thinks they can extract something more meaningful, message me and I'll arrange to get a copy to you.

 

[Ron The Bear]

Actually there has been a solitary exception to the above.

Big shout out to Ron The Bear.
Hope you are well.
You were right all along.
Hope you have a great weekend.
Well played good sir.