Opinion Connolly: Selwood needs more help

[sauron256]

I believe the tweet referred to therein is our very own SJ's!

So let me get this straight. SJ tweets Chris Scott about our fabulous forward press, and all of a sudden it turns to sh*t?

I think we've found who to blame.

 

[Lana]

How would his point change by using top 5??

It would change the outcome of his measure significantly.
For example a game where Selwood gets 13 CP, Johnson 11 and Kelly 10, results in Selwood and Johnson 1 point and Kelly nothing. When you collect all of the game data over a longer period of time it is very misleading.

It wouldn't change his point but it would make his "measure" look a lot less convincing. As instead of having something like
Selwood 70%
Kelly 30%
Johnson 25%
and a bunch of people grouping between 10-20%
etc

which implies he is a lone hand.

If you used top 5 you'd have something like this

Selwood 85%
Kelly 70%
Johnson 65%
Bartel 60%
and a bunch of people grouping between 20-40%

Which would paints a much better picture.

 

[Boxhead_31]

Its strange that he didn't write the same article in 2013 when Selwood played a similar game at a similar time of the year against the Swans when Selwood willed us to the win

 

[catempire]

It would change the outcome of his measure significantly.
For example a game where Selwood gets 13 CP, Johnson 11 and Kelly 10, results in Selwood and Johnson 1 point and Kelly nothing. When you collect all of the game data over a longer period of time it is very misleading.

It wouldn't change his point but it would make his "measure" look a lot less convincing. As instead of having something like
Selwood 70%
Kelly 30%
Johnson 25%
and a bunch of people grouping between 10-20%
etc

which implies he is a lone hand.

If you used top 5 you'd have something like this

Selwood 85%
Kelly 70%
Johnson 65%
Bartel 60%
and a bunch of people grouping between 20-40%

Which would paints a much better picture.

I don't agree. The point he is making is the drop off between first and second in the quoted categories. What the third, fourth, fifth, etc. did is irrelevant to that point and can't weaken it by definition. He said:

"Over that period, Selwood has played 93 games. In no fewer than 71 of those, 76 per cent of them, he has finished either Geelong's leading or second-highest clearance winner. The next on that "top two" list is James Kelly with just 32 games."

- so the point is the gap between Selwood (71) and Kelly (32)

Then similarly:

"In the area of contested ball meantime, the figures are similar. Selwood has been top two in 64 of those 93 games, the next highest Cat Steve Johnson with just 22 games. On Sunday, more than 60 per cent of Selwood's touches were won in a contest."

- again, the point being the gap between Selwood (64) and Johnson (22)

The overall point is the massive drop off between Selwood and the next. What happens after that is irrelevant to his (very well made) point.

 

[Biggy_Boy]

Rohan was talking about this on SEN 1116

listen to this on Audioboo

He mentions that we've lost the likes of Scarlett, Ling, Mooney, Corey, and Chapman as leaders, and that on-field leadership is therefore a problem. The problem with that narrative is that two years ago, we were 5 points off making a GF. That team didn't have Scarlett, Ling, Mooney, and throughout that year, Chapman only played eight games. Lack of leadership may be a problem now, but we were a top team two years ago with most of the guys he mentioned not in the team, so how much of an effect is that really having? In the 2011 grand final, Mooney didn't play, and Pods was gone in the first quarter, yet the forward line functioned pretty well. There are obviously issues with sharing the workload across the team right now, but I'm not sure that lack of leadership, specifically highlighting the losses of Ling, Mooney, Chapman, and Scarlett, is really relevant in assessing why the team has dropped off so much within the last two years.

 

[Lana]

I don't agree. The point he is making is the drop off between first and second in the quoted categories. What the third, fourth, fifth, etc. did is irrelevant to that point and can't weaken it by definition. He said:

"Over that period, Selwood has played 93 games. In no fewer than 71 of those, 76 per cent of them, he has finished either Geelong's leading or second-highest clearance winner. The next on that "top two" list is James Kelly with just 32 games."

- so the point is the gap between Selwood (71) and Kelly (32)

Then similarly:

"In the area of contested ball meantime, the figures are similar. Selwood has been top two in 64 of those 93 games, the next highest Cat Steve Johnson with just 22 games. On Sunday, more than 60 per cent of Selwood's touches were won in a contest."

- again, the point being the gap between Selwood (64) and Johnson (22)

The overall point is the massive drop off between Selwood and the next. What happens after that is irrelevant to his (very well made) point.

The nature of that system would reproduce similar results for almost any stat by it's nature. If you did it from 2007-2010 you'd get a similar result for Ablett. The measure doesn't mean what he implies it does.

Especially with that time frame of 4 years(going back to the start of 2011). With some of Geelong's most dominant clearance and CP players leaving half way through that time period and others spending half of it as forwards or defenders.

It's not a fair and relative statistical comparison.

 

[catempire]

Especially with that time frame of 4 years(going back to the start of 2011). With some of Geelong's most dominant clearance and CP players leaving half way through that time period and others spending half of it as forwards or defenders.

Err, that's kind of the point. The cupboard is bare but for Selwood.

It's not a fair and relative statistical comparison.

It is what it is: it clearly demonstrates there's a big gap behind Selwood.

If it's unfair I'm sure you'll be able to come up with some counter arguments to the central thesis...

 

[abet]

What a load of garbage by Connolly, Sellwod is a star and stars give it their all every time they step on that field. I wouldn't have expected anything different about the way Sellwood went about yesterday's game even if we had an Ablett and a Dangerfield in our side helping him, he would have still went just as hard and still marsheled the troops as he did. A champion footballer thrives in giving 100% and that wont take years from his career.

 

[Vdubs]

So let me get this straight. SJ tweets Chris Scott about our fabulous forward press, and all of a sudden it turns to sh*t?

I think we've found who to blame.

Which is our own SJ?

 

[Lana]

Err, that's kind of the point. The cupboard is bare but for Selwood.


It is what it is: it clearly demonstrates there's a big gap behind Selwood.

No it doesn't, just that others would have filled that role but not gotten enough to be on the board. And those that were on the board aren't higher up because they were doing something else or retired.

If it's unfair I'm sure you'll be able to come up with some counter arguments to the central thesis...

Naturally you deleted the part of my quote that addressed this and my first statement. I'll restate it and elaborate

1) For example a game where Selwood gets 13 CP, Johnson 11 and Kelly 10, results in Selwood and Johnson 1 point and Kelly nothing. Essentially it ignores almost identical contributions from players. In this example Kelly getting 10 is counted as much as him getting 0. When you collect all of the game data over a longer period of time it is very misleading.

2) Selwood gets a massive advantage by playing in the same role for the entire sample size where no other single player has. It doesn't mean that he hasn't had support, it just means that the people around him has changed over time.

But in general if you use a choose the top two and disregard the rest you will get a similar distribution of results for pretty much all outcomes, which on it's own makes this measure useless(and is the reason it isn't used anywhere else in the real world for measuring relative performance) and points 1 and 2 are perfect examples of a sample bias.

 

[catempire]

No it doesn't, just that others would have filled that role but not gotten enough to be on the board. And those that were on the board and higher up because they were doing something else or retired.

What?? "Others would have filled that role"? I guess we just choose not to support Selwood. That sounds, um, implausible.

And yes, the fact that there were retirements is entirely the point - there is no one there to support.

Naturally you deleted the part of my quote that addressed this and my first statement. I'll restate it and elaborate

I resent that insinuation. I deleted it because I couldn't see its relevance. If you want to remake the stats with Ablett go right ahead. Just claiming it would be so is meaningless.

1) For example a game where Selwood gets 13 CP, Johnson 11 and Kelly 10, results in Selwood and Johnson 1 point and Kelly nothing. Essentially it ignores almost identical contributions from players. In this example Kelly getting 10 is counted as much as him getting 0. When you collect all of the game data over a longer period of time it is very misleading. This would be a

You are going down a completely different path with hypothetical stats. In connolly's example, the 3rd place in those stats categories would be even further away from Selwood. It's irrelevant to his point.

2) Selwood gets a massive advantage by playing in the same role for the entire sample size where no other single player has. It doesn't mean that he hasn't had support, it just means that the people around him has changed over time.

Wow. Selwood does so much better than others because he plays midfield more? This is breathtaking! You aren't seriously suggesting that he plays there more for any other reason than he's clearly out best midfielder?

But in general if you use a choose the top two and disregard the rest you will get a similar distribution of results for pretty much all outcomes, which on it's own makes this measure useless and points 1 and 2 are perfect example of a sample bias.

Again, no it doesn't. Nonsense:

Orchard A produced 500 apples. Orchard B produced 200. Orchard A is clearly the most productive orchard even if there's a third orchard producing 150.

 

[Lana]

What?? "Others would have filled that role"? I guess we just choose not to support Selwood. That sounds, um, implausible.

And yes, the fact that there were retirements is entirely the point - there is no one there to support.


I resent that insinuation. I deleted it because I couldn't see its relevance. If you want to remake the stats with Ablett go right ahead. Just claiming it would be so is meaningless.


You are going down a completely different path with hypothetical stats. In connolly's example, the 3rd place in those stats categories would be even further away from Selwood. It's irrelevant to his point.


Wow. Selwood does so much better than others because he plays midfield more? This is breathtaking! You aren't seriously suggesting that he plays there more for any other reason than he's clearly out best midfielder?


Again, no it doesn't. Nonsense:

Orchard A produced 500 apples. Orchard B produced 200. Orchard A is clearly the most productive orchard even if there's a third orchard producing 150.

Your example isn't what his metric was metric actually shows, but what he pretend that it does.
It cuts using a method that heavily exaggerates results in general. It pretends that every orchard that produces less than the 2nd most apples per day produced nothing. Under this system it pretend that Orchard a produced 500 when it produced 600, that B produced 200 when it produced 450 and C produced 150 when it produced 400. And then Claims that A produced 2.5 more then the next orchard and that the other orchards aren't nearly productive.

The data set used would be considered poor sampling statistically because the conditions of the time "the players and role" are not directly comparable over time, this could be reduced if he used the time frame of a single year. For example Guthrie performance in 2011 skews his % in the whole time period where his performance in 2014 which is the relevant time period(for him anyway) The technical term for this is "sample bias".

You wouldn't see this method used in any scientific study and it would be cut up in a peer review.

 

[Moneypenny]

Thank God it was Roh.Co writing this article, it actually had a point and offered deeper analysis than the no s**t Sherlock title let on. While I agree that Sel needs a lot of help, there are not many players of a similar skill set to him in the league, let alone our playing list. No one can get near his ability to set sail and win a game off his own back, and we haven't seen anyone else in the 22 achieve such levels of leadership - yet.

I'm heartened by the leadership role Mitch Clark has fallen/been placed with/taken on, it may encourage younger or less experienced players to be more vocal and show some talent in that area.

" don't reckon he's driven by his own performance as much as he is taking his own teammates with him," said coach Chris Scott. Independently, teammate Mitch Duncan echoed those sentiments. "He brought the whole 21 with him today to get us over the line. He's just an extraordinary player," he said.

I get that Joel is a freak, but did it ever occur to you to give him a chop out maybe? Geeeees

 

[Stan The Caddy]

I get that Joel is a freak, but did it ever occur to you to give him a chop out maybe? Geeeees

Hahaha!

Yep just because he can do it on his own, it doesn't mean that he should have to. C'mon boys, give him a hand once in a while!

 

[catempire]

Your example isn't what his metric was metric actually shows, but what he pretend that it does.
It cuts using a method that heavily exaggerates results in general. It pretends that every orchard that produces less than the 2nd most apples per day produced nothing. Under this system it pretend that Orchard a produced 500 when it produced 600, that B produced 200 when it produced 450 and C produced 150 when it produced 400. And then Claims that A produced 2.5 more then the next orchard and that the other orchards aren't nearly productive.

The data set used would be considered poor sampling statistically because the conditions of the time "the players and role" are not directly comparable over time, this could be reduced if he used the time frame of a single year. For example Guthrie performance in 2011 skews his % in the whole time period where his performance in 2014 which is the relevant time period(for him anyway) The technical term for this is "sample bias".

You wouldn't see this method used in any scientific study and it would be cut up in a peer review.

This is really surprising from you. You are now making no sense whatsoever. Sampling bias? Please! It is not even sampling!

Let me put it another way: no other way of manipulating the data would undermine Connolly's point. If he'd compare the top 3, Selwood would still be far and away ahead. Likewise 4,5,6, etc. Moreover, as that number increases it becomes less and less about the point Connolly is making which is Selwood is number one and 2nd is a long way behind him.

For example, if you ask, "how many times was player x in the top 10 for clearances in a match over the period" it's absolutely meaningless for determining whether there is a big gap between number 1 and number 2 over that period because player 1 could have been 1st in 99 games out of 100 and player 2 could have been 10th 98 games out of 100 and they would be separated only by 1 on that measure - clearly not showing the full picture.

 

[00VicWard001]

Bit surprised at the negative response to the article. He didn't just state the obvious point (which we all agree it is) but went to the trouble of digging into the stats behind it. That and the points about leadership made it a worthwhile read for me.

I agree. Wouldn't everyone rather articles about the GFC to read, to promote discussion and help analyse, rather than hear the same rubbish about the Hawks being the 2015 premiers, James Hird being a true hero etc etc?

It's an obvious topic, but not one so obvious as last weekend. An article like this reinforces the stark difference between the way Selwood plays and others do at the club. They might know it, but now they know that EVERYONE knows it. So what are they going to do? Say 'No Duh Connolly! We already knew that' and then go off and do the same thing?

 

[00VicWard001]

How would his point change by using top 5??

it would just add noise to it. I think that using the top two highlights his very point and his point is correct. The stats do not lie or mislead in this case.

 

[Strangled Cat]

I wonder if Bernie Vince will try his trucks on Selwood when we face Melbourne. More importantly, will our boys offer Joel protection? He's a sitting duck if he's left to do it by himself every week.

I'm expecting a lot from Josh Caddy this year and I hope that he starts to push up into the A grade midfielder category. We need him to help Joel in the guts as he's a big bodied mid with huge raps on him and just entering his prime.

The issue could be even further arrested if we get Danger next year.

 

[Cursed Cat]

the point Connolly is making which is Selwood is number one and 2nd is a long way behind him.

or, in BF-speak:

Selwood >daylight>2nd best>the rest.

or if you prefer:

Selwood ...... .>daylight>2nd best/the rest.

As a couple of posters have mentioned already in this thread, the current injuries to Jimmy Bartel and Matthew Stokes (and there must be some doubt on Johnson's ability to run out four quarters?) will potentially give opportunities for younger players to fast-track their development.

I don't think this will help Selwood's cause though! If anything he'll be carrying a bigger load than ever.
He'll have assistance from James Kelly (who's form has been steady); after that it'll all be sub 100-gamers, unless Scott and co. want to give Duncan a run in the guts.

Horlin-Smith should earn a reprieve; with Jimmy's absence perhaps the team can accommodate his lack of pace. He's still played so little football in the grand scheme of things, just 33 games. I've been a bit back and forth on my opinion of him to be honest; but if he wins a spot in the 22 I'd love to see him start to cement himself in the side.

Josh Caddy's midfield minutes will need to increase. Has he got the stamina to handle that increased workload?
He has undoubted talent, but he needs to learn to get those easier kicks with smarter running (patterns).
It's all death-or-glory with this dude; even the Supreme Leader has the smarts to run to the right spots without getting the ol' ocular's knocked about every time.

Guthrie's role in the team continues to provoke some division among supporters; he generally gets a fair slab of time on the ball anyway so he'll be up to scratch as far as the workload goes.
Some argue that he doesn't do enough as a midfielder and is better served as a defender; personally I've always rated him really highly and think he's probably the best young (less than 100 games) player at the club.
He's getting close to the 70 game mark, he's averaging 20 touches and 5 tackles so far this season.
His run-with/accountable roles have been a bit confusing in that it's sometimes unclear just how close he's supposed to be minding his opponent (imo).
To me his development has been excellent though and he's a really well-rounded player; he may not always beat his opponent at this stage but at least he has a strong defensive ethic. I've seen people on this very board imply (or flat-out state) that he's 'soft' and I find that perplexing.
Round 3 indicated that he *can* still go to the backline and have an impact as he did in 2013, but right now might not be the time for that re-shuffle, and personally, if the club is keen to continue turning him into a midfielder then I'm pretty content to accept the learning curve might be steep at times.

Steve Motlop has shown a bucketload of ability in his 70 games, and to give him his due he has played some big hands in winning the club games post-2011.
Increasingly his hard and smart running allows him to pick up plenty of the ball; unfortunately the decision-making and even the execution has let him down at times this year. But if anyone on the list has the potential to be our next Gary Ablett jnr, it's this bloke.
Is his knee a problem?
Picked up a lazy 32 touches (27 kicks) against the premiers; sure many of them were cheap and ineffectual, but he can run and knows where to run.

So Joel Selwood, in the words of the immortal Little River Band, "Hang on. Help is on it's way".

 

[Teriyakicat]

That's about the time our ruckmen started getting injured and stopped being strong in the clearances, hence we couldn't employ the press.

We played McCarthy in R1&2, and Burbury R3-7, with Murdoch and Varcoe also applying a lot of forward pressure through those early rounds too, 10lana. Murdoch and Varcoe spent more time in defence later in the season and we effectively lost about 12 tackles from the forward line.

 

[Generalissimo]

Guthrie, Duncan, and Caddy need to be the ones to lift. They're big enough and have enough games experience to do the job.

 

[Moneypenny]

Hahaha!

Yep just because he can do it on his own, it doesn't mean that he should have to. C'mon boys, give him a hand once in a while!

Exactly!! That needs to be plastered across the change rooms somehow.

Doesn't even need to be every match. Maybe every other match! Lend him a hand when you think he might need it. It may become the new normal eventually. I hope. These are professional athletes right?

 

[Bobby_]

I reckon the cruel fact amongst all of this is that Selwood will be over 30 when he does finally have a terrific midfield surrounding him. It's just the way it has panned out having so many wonderful players who were able to maintain a very high standard for so long.
Losing Mumford, Prismall, Ablett & Christensen certainly hasn't helped. You can make a case for Cowan & Hogan, given they were two highly rated footballers who showed promise early on in their careers.

 

[Footy Smarts]

I reckon the cruel fact amongst all of this is that Selwood will be over 30 when he does finally have a terrific midfield surrounding him. It's just the way it has panned out having so many wonderful players who were able to maintain a very high standard for so long.
Losing Mumford, Prismall, Ablett & Christensen certainly hasn't helped. You can make a case for Cowan & Hogan, given they were two highly rated footballers who showed promise early on in their careers.

You'd hope that in 2 years Duncan, Motlop, Caddy and Guthrie are the core of a very good midfield. It was the 2001 draft's 6th year that Bartel, Ablett, Kelly and Johnson exploded. Time for those guys to do similar.

 

[manboob]

I reckon the cruel fact amongst all of this is that Selwood will be over 30 when he does finally have a terrific midfield surrounding him.

It's not gonna take 4 years.