Strategy 2016 Tacs Trailer

Remove this Banner Ad

Log in to remove this ad.

The Oshea 3rd qtr one shows red hot in the 3 spots where he had his 3 disposals. That means it gives you no real info on when he doesnt have the ball, ie doing shepards, blocks, intercept spoils, 1% etc.
 
I think the individual player heat maps can tell a story, especially with key position players. The heat maps in 2014 for our finals games against Hawthorn and Freemantle are interesting with Chad Wingard, Polec and White.
Port Adelaide unleashed Jared Polec and Matt White this season on either wing.

Polec had 10 bounces in the first final against Richmond (seventh in the competition for that stat with 53 bounces for the season) and looks every bit the classic left-footed wingman, whereas White has improved his running and is now more controlled at the point of delivery.

At half-time in the semi-final Chad Wingard went to the wing and kept driving the ball forward.

In the end his two-bounce, two-possession goal late in the last quarter when he outsprinted a seemingly exhausted Fremantle's Danyle Pearce, who appeared to give up the chase, sealed the game for Port Adelaide.
 
RussellEbertHandball said:
The Oshea 3rd qtr one shows red hot in the 3 spots where he had his 3 disposals. That means it gives you no real info on when he doesnt have the ball, ie doing shepards, blocks, intercept spoils, 1% etc.
Click to expand...

*ahem*

image.png
image.png
image.png
image.png
Doesn't explain the first quarter though does it?
 
Pull up the replay and look at those 2 possessions. They may be clangers or something?

Try Josh Walker in the brisbane game. He had 5 possessions for the match with 2 clangers.
 
bluemoon86 said:
Pull up the replay and look at those 2 possessions. They may be clangers or something?

Try Josh Walker in the brisbane game. He had 5 possessions for the match with 2 clangers.
Click to expand...

You've helped work it out...but it's nothing to do with clangers. The red and yellow represents player movement, just like I said it did:

image.png
image.png
image.png
image.png
 
As in player movement with the ball in hand? So a larger circle means they ran with the ball for a bit before disposing, but a small circle is maybe when they receive a handball and dish off straight away without moving much?

On the opposite side of the coin Pittard might be a good one to look at.
 
bluemoon86 said:
As in player movement with the ball in hand? So a larger circle means they ran with the ball for a bit before disposing, but a small circle is maybe when they receive a handball and dish off straight away without moving much?

On the opposite side of the coin Pittard might be a good one to look at.
Click to expand...

image.png

"Classic horseshoe pattern."

 
There are 2 bits of great defensive play I saw Oshea do in the 3rd Quarter as Brisbane kicked to the scoreboard end in the middle of his 2 red heat defensive areas tbat just comes up green. Thats why I questioned if its only possessions and if not what is included in these maps.
 

(Log in to remove this ad.)

So the shading is just an estimation of where possessions will take place based on empirical data? That is, a wide red range means that player/team has a high probability of a possession in that area? And yellow means that there was a medium probability whereas anything else there is a low probability (which is why Hombsch didn't record a "heat" signature on the left wing because he wasn't really expected to get a possession on that side?
 
That Hombsch data on my phone only came up as a blue dot. On my computer screen I see a little bit of yellow/green and red/orange around the blue dots. So what does that mean and the fact there is no other yellow or red on the first quarter map?? Does that mean he didn't do a spoil or other 1%er or that those things don't get recorded?
 
RussellEbertHandball said:
That Hombsch data on my phone only came up as a blue dot. On my computer screen I see a little bit of yellow/green and red/orange around the blue dots. So what does that mean and the fact there is no other yellow or red on the first quarter map?? Does that mean he didn't do a spoil or other 1%er or that those things don't get recorded?
Click to expand...

Since it's based on possession it doesn't record 1%ers. I'm pretty sure from what I've just read about Kernel Diversity Estimation that the shading is simply how likely it was for that possession to occur in that area of the ground based on previous data. Could be wrong - hopefully Champion_Data can explain it better :)
 
OneGreatClub said:
https://twitter.com/championdata?lang=en
Championdata on twitter.
Hotplots interesting. We have a way to go.
Click to expand...

When we actually played defense and pressed like we should have (St.Kilda, Essendon, Richmond, Brisbane) - 118.75 points for, 67.5 points against (premiership standard)

When we did our own thing and didn't play for each other (Adelaide, GWS, Geelong) - 70 points for, 134 points against

champion.png

Now, you can say that the first group of teams is much worse than the bottom group of teams...but they're not 115.25 points worse. That's what happens when you've got a group of players that have gone away from what makes them good (defense) and into trying to play 'sexy' football. The layman will look at Port Adelaide at their best and think that it's the free wheeling attack that makes them great...but it's actually defending as a unit that creates those attacks.

Yes, that's right - we were worse than Fremantle and Carlton in those games. The worst side in the competition. Worse than Essendon, who at least had something approaching defense. This is why I've always maintained it was effort. Remember - two of the sides that are in that premiership standard window are there partially because of us.
 
What elevates us into the premiership standard, Janus?

Schulz coming back into the side might elevate us into the 100 points for category, but will we be able to consistently defend sub 86 without a Carlisle, or a recognised ruckman allowing Trengove to defend?
 
From the Wiki page the colour coding represents the probability of whatever data points it is that CD are using occurring in an aggregated region of space. I'm pretty sure that they won't want to tell us exactly what these data are. The point is to make inferences from spacially represented data. It's probably reasonably meaningless in some of the examples discussed above looking at individual or small numbers of possessions for a player in a quarter, and more powerful with larger data sets.

I hope it's OK to post these two paragraphs:

https://en.m.wikipedia.org/wiki/Multivariate_kernel_density_estimation

In the left figure below, a kernel (represented by the grey lines) is centred at each of the 50 data points above. The result of summing these kernels is given on the right figure, which is a kernel density estimate. The most striking difference between kernel density estimates and histograms is that the former are easier to interpret since they do not contain artifices induced by a binning grid. The coloured contours correspond to the smallest region which contains the respective probability mass: red = 25%, orange + red = 50%, yellow + orange + red = 75%, thus indicating that a single central region contains the highest density.


Construction of 2D kernel density estimate. Left. Individual kernels. Right. Kernel density estimate.

The goal of density estimation is to take a finite sample of data and to make inferences about the underlying probability density function everywhere, including where no data are observed. In kernel density estimation, the contribution of each data point is smoothed out from a single point into a region of space surrounding it. Aggregating the individually smoothed contributions gives an overall picture of the structure of the data and its density function. In the details to follow, we show that this approach leads to a reasonable estimate of the underlying density function.

https://en.m.wikipedia.org/wiki/Multivariate_kernel_density_estimation
 
Janus said:
Since it's based on possession it doesn't record 1%ers. I'm pretty sure from what I've just read about Kernel Diversity Estimation that the shading is simply how likely it was for that possession to occur in that area of the ground based on previous data. Could be wrong - hopefully Champion_Data can explain it better :)
Click to expand...
It's not a prediction, it's just a summary of the possessions that actually happened. Once you add 50 dots to a map of the ground it is near impossible to get any information about the patterns or density of the dots. By putting a kernel density estimate over the top of it, you get areas of high density represented as red, areas of medium density as yellow and no density as green. The more spread out the dots are on the field, the wider the shading on each point is. If none of the dots are close enough that they overlap, all of them will have a red centre spreading out to yellow and then green. If you have two dots next to each other, and others spread out and not overlapping, the colour underneath the two dots will be red and underneath all other dots will only be yellow, since the "height" of the density under the two dots is double the "height" under the single dots. Colour represents the height of the density estimate. It has nothing to do with efficiency or how far the players run.
 
Macca_1870 said:
What elevates us into the premiership standard, Janus?

Schulz coming back into the side might elevate us into the 100 points for category, but will we be able to consistently defend sub 86 without a Carlisle, or a recognised ruckman allowing Trengove to defend?
Click to expand...

Nothing to do with personnel. Everything to do with team defense and offense.



This video shows how strong something can be when it is built with solid support - a paper bridge, held together with tape and glue, with the right formation, can hold over 480 kg. The shape used is more important than the materials you build with. Our problems in our losses were systemic - based around players believing they had to be the one to pull the side back into playing the type of football we are capable of playing instead of just trusting in the formation and being support players when required.

That's why I find it amusing that there are a lot of people wanting to drop Toumpas. Dimitri has done everything he has been asked to do when it comes to being a support player for his teammates - he tracks back and defends, he pushes up in attack. He's not a star - not yet - but we don't need a lot of stars. You need the inky blackness of space - the players doing the things that you can't see - to make the stars shine brightly.

This is why I think Gray's 'injury' was just that - an excuse to get the players back thinking not on individual performances, but on team care. As Ken said in that coaches round table on Fox - "I tell the boys, 'We're not that much better than anyone else'". And it's true - if you're not going to adhere to the game plan, you're going to fold. A championship winning team is greater than the sum of its parts, with everyone playing their role. We remember the stars because they give us the flashy performances - but without the sacrifice of personal glory from others, we will never be successful.
 
Champion_Data said:
It's not a prediction, it's just a summary of the possessions that actually happened. Once you add 50 dots to a map of the ground it is near impossible to get any information about the patterns or density of the dots. By putting a kernel density estimate over the top of it, you get areas of high density represented as red, areas of medium density as yellow and no density as green. The more spread out the dots are on the field, the wider the shading on each point is. If none of the dots are close enough that they overlap, all of them will have a red centre spreading out to yellow and then green. If you have two dots next to each other, and others spread out and not overlapping, the colour underneath the two dots will be red and underneath all other dots will only be yellow, since the "height" of the density under the two dots is double the "height" under the single dots. Colour represents the height of the density estimate. It has nothing to do with efficiency or how far the players run.
Click to expand...

Ok I get it now. So what happened with Hombsch in the first quarter of the Brisbane game? Was that just an error? Cause there are two possessions close to each other that look as though they should have been red. And then I look at O'Shea's third quarter in the same game and he has huge areas of red around his three possessions that are spaced all over the field, whereas Josh Walker only gets a tiny little dot (which is what I would have expected it to be for O'Shea).

Just trying to understand how it works - I never thought it was about efficiency or how far players run, more where they are receiving/disposing of the ball on the field. :)
 

Remove this Banner Ad

Strategy 2016 Tacs Trailer

Similar threads

GremioPower
Resource Footy Tactics: Could Relationism work in Australian Football?
Replies
18
Views
1K
GremioPower
GremioPower

Remove this Banner Ad

Back
Top