Hitouts vs Clearances - Is there any Correlation?

[Rosco]

Pearson Product Moment correlation = .42 (can be anything between -1.0 and +1.0; the further away from zero the more significant the correlation)

Hence moderate positive association between hitouts and clearances. Hitouts predict approx 18% of variability in clearances, so there are other factors operating, but the answer to the original question is Yes.

But correlation does not imply a particular causal direction. It might be that the ruckman is more likely to win the hitouts if he thinks that he is playing with mids who are likely to win the clearances!

Thanks SuperB for a wonderfully refreshing post.

We are so used to most posters taking the tried and proven approach of 'baffle them with BS', but you've taken the intellectual high ground and 'blinded us with science'.

Love it!


Question (to now expose my ignorance):
Since "correlation does not imply a particular causal direction",
then what you are really saying is that the answer is 'yes' (but only moderately so),
and even then, maybe not at all...?

I think he used science to say 'maybe'

[loris]

Pearson Product Moment correlation = .42 (can be anything between -1.0 and +1.0; the further away from zero the more significant the correlation)

Hence moderate positive association between hitouts and clearances. Hitouts predict approx 18% of variability in clearances, so there are other factors operating, but the answer to the original question is Yes.

But correlation does not imply a particular causal direction. It might be that the ruckman is more likely to win the hitouts if he thinks that he is playing with mids who are likely to win the clearances!

Thanks SuperB for a wonderfully refreshing post.

We are so used to most posters taking the tried and proven approach of 'baffle them with BS', but you've taken the intellectual high ground and 'blinded us with science'.

Love it!


Question (to now expose my ignorance):
Since "correlation does not imply a particular causal direction",
then what you are really saying is that the answer is 'yes' (but only moderately so),
and even then, maybe not at all...?

I think he used science to say 'maybe'

And they think the pros and cons for climate change have us 'great unwashed' confused................. Now I'm doing my head in over hitout ratios. Gee life gets complicated at times

[|Andy|]

Pearson Product Moment correlation = .42 (can be anything between -1.0 and +1.0; the further away from zero the more significant the correlation)

Hence moderate positive association between hitouts and clearances. Hitouts predict approx 18% of variability in clearances, so there are other factors operating, but the answer to the original question is Yes.

But correlation does not imply a particular causal direction. It might be that the ruckman is more likely to win the hitouts if he thinks that he is playing with mids who are likely to win the clearances!

Thanks SuperB for a wonderfully refreshing post.

We are so used to most posters taking the tried and proven approach of 'baffle them with BS', but you've taken the intellectual high ground and 'blinded us with science'.

Love it!


Question (to now expose my ignorance):
Since "correlation does not imply a particular causal direction",
then what you are really saying is that the answer is 'yes' (but only moderately so),
and even then, maybe not at all...?

He's saying just because there's correlation doesn't mean that clearances are the direct results of more hitouts. There may be other factors and reasons why a clearance was associated with more hitouts.

So a good example I found to understand this is: In a particular year, there was a direct correlation between the amount of ice cream sold and the number of shark attacks. This doesn't mean that buying ice cream will mean you'll get attacked by a shark. As it turns out, more ice cream is sold when the weather is hot, and people go to the beach when it's hot increasing the risk of getting attacked by a shark.

You can see there's a correlation, but buying ice cream isn't the causation of a shark attack.

[Rosco]

It doesn't "necessarily" mean that, but it doesn't rule it out either. Is there any research on flavours?

[samoht]

Good discussion... I don't think maths has a place here when we talk about the clearance stats as "quality" is being overlooked. All clearances are not the same, but are being counted as being the same (in the stats), as has been pointed out already.

There are a lot of factors - and we're only looking at raw numbers, after all, when we look at the stats - but I still can't overlook the fact that the awesome and formidable combination of Nat and Cox and the extra 15 or so hitouts they achieved on average simply did not translate into more clearances, even if it's in raw numbers.

On the issue of quality - North Melbourne and Wells with his damaging and dashing clearances was a stand-out vs us this year. Wells was unstoppable and he was the match winner.
Now if someone else was there for North instead of Wells - would they have read the hitout as well as he did and furthermore would they have been able to clear the ball and would they have been as damaging as Wells was ?

Maybe what every team needs is a Wells in their team if they want quality clearances , more so than a Cox or a Nat ?

[kosifantutti]

Good discussion... I don't think maths has a place here when we talk about the clearance stats as "quality" is being overlooked. All clearances are not the same, but are being counted as being the same (in the stats), as has been pointed out already.

This is the key. Scruffy up and under kicks that don't advantage the forwards or even necessarily go forward are counted the same as someone running to the edge of the square and having a ping at goal. Too many times McEvoy is uncompetitive and allows the latter to happen.

Having said that, I'm always happy to see the use of Pearsons Correlation coefficient.

[samoht]

Good discussion... I don't think maths has a place here when we talk about the clearance stats as "quality" is being overlooked. All clearances are not the same, but are being counted as being the same (in the stats), as has been pointed out already.

This is the key. Scruffy up and under kicks that don't advantage the forwards or even necessarily go forward are counted the same as someone running to the edge of the square and having a ping at goal. Too many times McEvoy is uncompetitive and allows the latter to happen.

The key (to quality clearances) might be a Wells (or a Wells like player) - a damaging clearance player who can read the hitout and then use his smarts and dash and evasiveness to either clear a path for himself and then unleash a long accurate kick to his fullforward or .. who can spot and cleanly deliver to a midfield team mate who is in clear space.
The quality of the clearance may largely rest with the clearance player.

Nat, Cox, McEvoy hit the ball to hotly contested spots most of the time (95% of the time) - occasionally (once in a blue moon) they may manage to find someone in clear space (sometimes it may turn out to be an opposition player too) - but a Wells (like player) can create that smidgeon of required space anyway and can thus bring about the quality clearances - maybe we should look no further ?
On the one hand we know that Cox and Nat don't bring about more clearances (in terms of raw numbers and so I'm starting to doubt the quality/value of their extra 15 or so hitouts and therefore their hitouts, in general - or anyone else's hitouts for that matter) - but on the other hand we know that a Wells (like player) can and does bring about quality clearances - this is a given - an indisputable fact !!
q.e.d. (for the maths inclined).

[Superboot]

Question (to now expose my ignorance):
Since "correlation does not imply a particular causal direction",
then what you are really saying is that the answer is 'yes' (but only moderately so),
and even then, maybe not at all...?

Nice summary of the analysis Dave!!

In terms of the causality issue, Andy's shark attack analogy is terrific. Another to consider might be weight gain and depression. Do people get depressed because they've put on weight, do they get depressed first and then put on weight because of comfort eating, or are they simply genetically predisposed to both?!

It's true that we really need to look at quality rather than quantity of hitouts, but it would also be interesting to know whether those two things are completely independent of each other. I suspect they are not.

[cross2012]

The mids with the dominant ruckmen, knows where the ball is going, the others have to anticipate i.e. one team becomes offensive the other defensive. Great mids can anticipate the where the ball is going and get a scrubby, push foward clearance. Watching WCE, the have a number of runners that charge forward knowing the ball will end up in their path and as a result get a fast break clearance. Thats why stats cant be the defining argument, you need to measure the value of each.
If rucks are of similiar standard, then they probably negate each other, however if a ruckmen smashes his opponent, its a huge advantage.

[samoht]

Question (to now expose my ignorance):
Since "correlation does not imply a particular causal direction",
then what you are really saying is that the answer is 'yes' (but only moderately so),
and even then, maybe not at all...?

Nice summary of the analysis Dave!!

In terms of the causality issue, Andy's shark attack analogy is terrific. Another to consider might be weight gain and depression. Do people get depressed because they've put on weight, do they get depressed first and then put on weight because of comfort eating, or are they simply genetically predisposed to both?!

It's true that we really need to look at quality rather than quantity of hitouts, but it would also be interesting to know whether those two things are completely independent of each other. I suspect they are not.

Also, what defines a quality hitout (even if we can isolate its influence or causality on the clearances) ? -
and before we confer quality to a ruckman we may need to look at the ruckman's complete ledger ; maybe give a +1 score for a clear hit out to advantage, a -1 for an embarassing "hitout to disadvantage " and a zero for hitouts that are won but go to 50/50s !
Also, what we may need to factor in is that a quality clearance player can distort things (the clearance stats, by making the ruckman look good) whenever they turn a 50/50, go nowhere hitout, into a quality clearance.

I doubt that we or anyone else can do the level of analysis that's required to give definitive answers re: the hitouts and an individuals ruckmen's importance and overall influence and ranking among his peers (in regards to hitouts).
I'm more inclined to confer "quality" to the stand-out clearance player who can make things happen as far as bringing about quality clearances are concerned at this point - their influence is much more obvious and easier to define for me - as we can see it unfold in front of our eyes.
and so I remain reluctant to bestow greatness to a Nat and Cox and by extension to mark down the run of the mill tap ruckmen - at this stage - just based on the raw hitout numbers.
In other words I don't think that Nat and Cox are as great as they are made out to be or McEvoy is as bad as everyone thinks he is, based purely on their hitouts.

[matrix]

People who have actually player football as a midfielder will know the worth of a dominate ruckman.

Seems everyone on Saintsational must have been a forward.

I envy your height.

agreed spinner
i played in the middle for about two years (before going to the HBF) and having a ruckman that can palm the ball down into your moving path is absolutely vital and a great skill

[samoht]

People who have actually player football as a midfielder will know the worth of a dominate ruckman.

Seems everyone on Saintsational must have been a forward.

I envy your height.

agreed spinner
i played in the middle for about two years (before going to the HBF) and having a ruckman that can palm the ball down into your moving path is absolutely vital and a great skill

I do remember one game when Everitt had no opposition whatsoever vs Essendon (they must have had a crap ruckman at the time ?) when the mighty Nathan Burke made the most of his immaculate laser-guided taps, straight down the throat, to set up Hall with 3 or 4 last quarter goals and thereby set us up with a run away win.

That display was so rare - that it stands out.
Usually the ruckman are at full stretch and fully opposed - they rarely get it that easy.

At local footy you might get that sort of mismatch.

[gringo]

The thing is that i believe both Nic Nat and Cox are just as important away from the ruck contest. Cox can kill you on the forward line can take a back chop out role etc. Nic Nat has the ability to go as an extra rover in the centre and follows up his own tap work often. He is dangerous when on song because he gets the ball moving forward quickly and can kill you with his burst speed which often makes teams focus on negating him being able to do that. it means they are already one down in the centre guarding him often.

I noticed when we played Freo Sandilands just seemed to get every tap out but our centre players just watched where he tapped it and we seemed to win the clearances. Dominating hitouts is also only worth anything if your rovers are efficient. West coast once they get the ball are very clean and quick so probably don't need as many clearances. We seem to need double the clearances to overcome our sloppy ball use.

[Dis Believer]

So is the summary of all of this that a good ruckman makes a difference, but hte difference between ruckmen at AFL is not great enough for the it to translate into a clear advantage inthe game? Because we have all seen our own mids start roving to the opposition ruckman in games where our ruckman gets smashed.

Many of the guys here have played or do play still , whether that be at schoolboy or local club level, and know that the reality is that a good ruckman makes life much easier on you as a mid at a centre bounce, you simply don't have to work as hard. The consensus seems to be however, that at the highest level there is not enough of a differential betwen the competing rucks for it to be a significant advantage. There also appears to be a factor of the mids at this level being so good that unless there is a yawning chasm betwen the standard of the rucks, that the mids are able to compensate for an inferior ruckman.

[gringo]

So is the summary of all of this that a good ruckman makes a difference, but hte difference between ruckmen at AFL is not great enough for the it to translate into a clear advantage inthe game? Because we have all seen our own mids start roving to the opposition ruckman in games where our ruckman gets smashed.

Many of the guys here have played or do play still , whether that be at schoolboy or local club level, and know that the reality is that a good ruckman makes life much easier on you as a mid at a centre bounce, you simply don't have to work as hard. The consensus seems to be however, that at the highest level there is not enough of a differential betwen the competing rucks for it to be a significant advantage. There also appears to be a factor of the mids at this level being so good that unless there is a yawning chasm betwen the standard of the rucks, that the mids are able to compensate for an inferior ruckman.

something like that. A very good ruckman does make life easier and can make a few goals difference per game. A poor ruck can cost you the other way. Mids have to work in a more laborious fashion for the same effect when they are crap in the ruck.

[samoht]

Can we conclude that - ?

1. The major determinant re: quality clearances , at the highest level - may be having quality mids/clearance players.
The Wells, Mitchells, Kennedys are the key to quality clearances - not the rucks so much.

2. The rucks more or less negate each other at hitout contests - even if the raw numbers seem to show an advantage - 15 more hitouts than the average in: Nat and Cox's and WCE's case - nevertheless WCE clearances numbers are only about average. So the (raw number hitout) advantage is effectively lost, as it doesn't translate into more clearances.

3. Ruckmen (especially the mobile, athletic types who gather their 20 possessions on average) can make a real difference around the ground.
This is where they can really shine.

[bergholt]

Hence moderate positive association between hitouts and clearances. Hitouts predict approx 18% of variability in clearances, so there are other factors operating, but the answer to the original question is Yes.

Solid analysis. I know very little about statistics but I do know well the principle garbage in, garbage out.

I think this is samoht's argument. The hitout and clearance data we get given takes no account of the quality of each one, so even if there is a correlation between them, it doesn't mean anything for actual results. However, don't confuse the public data with the data the clubs have - there's ten times as much data which we never see.

[samoht]

Hence moderate positive association between hitouts and clearances. Hitouts predict approx 18% of variability in clearances, so there are other factors operating, but the answer to the original question is Yes.

Solid analysis. I know very little about statistics but I do know well the principle garbage in, garbage out.

I think this is samoht's argument. The hitout and clearance data we get given takes no account of the quality of each one, so even if there is a correlation between them, it doesn't mean anything for actual results. However, don't confuse the public data with the data the clubs have - there's ten times as much data which we never see.

Well summarised bergholt...
Good analysis by Superboot - and he was right to qualify what the result (moderate positive association) actually meant and to point out that there were other factors operating.

The next stage of the analysis would be to up the anti and to look for the major determinant for "quality" clearances, not just clearances... I wonder how far the club data goes on that score and if they have determined the answer to this seemingly inscrutable topic, and the rhetorical questions (in effect) that I have raised here.

[SainterK]

I love Ben, think he'll turn out fine...but I miss Gardi

[loris]

I love Ben, think he'll turn out fine...but I miss Gardi

Me too SainterK................. I see a lot of upside for Ben.

Hey I notice Wells name being mentioned in this thread alot........................ we picked Goddard over Wells in the draft.............. you win some you lose some eh? We could do with Wells' silky skills nowadays.

I can remember being at Rottnest Island watching the Grand Final at the Rotto Pub that draft year, and a very young, shy Daniel Wells was sitting near me. I got to chat to him over the afternoon and he was hoping the Saints would pick him up in the draft, alas not to be.

I remember posting about that meeting on the SS site and it ended up in a big debate over who to go with Goddard or Wells. Most seemed to think Wells would want to return to WA (I don't know why they thought that as he lived most of his life in SA and his family lived there, he had gone to WA solely for education and footy reasons) and Goddard was the choice of the majority of SS

Maybe because I was so impressed by this polite, shy unassuming kid (Daniel Wells) so much, was the reason I never warmed to BJ as a person. I originally wanted the Saints to go for Wells as he wanted the Saints, plus there was all that 'pushy mum' stuff with Goddard's garralous mother at the time that turned me right off BJ.

Strange what makes one like or be just ho hum about a player. I always acknowledged BJ skills, but never was part of his fan club............ So my Daniel Well's meeting possibly created my ho hum attitude towards BJ.

[sunsaint]

You can see there's a correlation, but buying ice cream isn't the causation of a shark attack.

I can see that your hypothesis may be sound in genreal terms, but i think your methodology was not thorough enough to eliminate another more sound conclusion.
If there are shark attacks, then there will also be a corresponding increase in ice cream sales.
No one will go in swimming.

I would also like to know if superboot made allowances in his statistics for obvious factors like roof open or closed. And did he conduct a controlled sample to eliminate any statistical deviations.

PS Ben's ok!

[elizabethr]

I love Ben, think he'll turn out fine...but I miss Gardi

Me too SainterK................. I see a lot of upside for Ben.

Hey I notice Wells name being mentioned in this thread alot........................ we picked Goddard over Wells in the draft.............. you win some you lose some eh? We could do with Wells' silky skills nowadays.

I can remember being at Rottnest Island watching the Grand Final at the Rotto Pub that draft year, and a very young, shy Daniel Wells was sitting near me. I got to chat to him over the afternoon and he was hoping the Saints would pick him up in the draft, alas not to be.

I remember posting about that meeting on the SS site and it ended up in a big debate over who to go with Goddard or Wells. Most seemed to think Wells would want to return to WA (I don't know why they thought that as he lived most of his life in SA and his family lived there, he had gone to WA solely for education and footy reasons) and Goddard was the choice of the majority of SS

Maybe because I was so impressed by this polite, shy unassuming kid (Daniel Wells) so much, was the reason I never warmed to BJ as a person. I originally wanted the Saints to go for Wells as he wanted the Saints, plus there was all that 'pushy mum' stuff with Goddard's garralous mother at the time that turned me right off BJ.

Strange what makes one like or be just ho hum about a player. I always acknowledged BJ skills, but never was part of his fan club............ So my Daniel Well's meeting possibly created my ho hum attitude towards BJ.

You seem to make a lot of decisions while on licensed premises !

[SainterK]

I love Ben, think he'll turn out fine...but I miss Gardi

Me too SainterK................. I see a lot of upside for Ben.

Hey I notice Wells name being mentioned in this thread alot........................ we picked Goddard over Wells in the draft.............. you win some you lose some eh? We could do with Wells' silky skills nowadays.

I can remember being at Rottnest Island watching the Grand Final at the Rotto Pub that draft year, and a very young, shy Daniel Wells was sitting near me. I got to chat to him over the afternoon and he was hoping the Saints would pick him up in the draft, alas not to be.

I remember posting about that meeting on the SS site and it ended up in a big debate over who to go with Goddard or Wells. Most seemed to think Wells would want to return to WA (I don't know why they thought that as he lived most of his life in SA and his family lived there, he had gone to WA solely for education and footy reasons) and Goddard was the choice of the majority of SS

Maybe because I was so impressed by this polite, shy unassuming kid (Daniel Wells) so much, was the reason I never warmed to BJ as a person. I originally wanted the Saints to go for Wells as he wanted the Saints, plus there was all that 'pushy mum' stuff with Goddard's garralous mother at the time that turned me right off BJ.

Strange what makes one like or be just ho hum about a player. I always acknowledged BJ skills, but never was part of his fan club............ So my Daniel Well's meeting possibly created my ho hum attitude towards BJ.

Oh well, we could of ended up with Brennan

[Superboot]

You can see there's a correlation, but buying ice cream isn't the causation of a shark attack.

I can see that your hypothesis may be sound in genreal terms, but i think your methodology was not thorough enough to eliminate another more sound conclusion.
If there are shark attacks, then there will also be a corresponding increase in ice cream sales.
No one will go in swimming.

I would also like to know if superboot made allowances in his statistics for obvious factors like roof open or closed. And did he conduct a controlled sample to eliminate any statistical deviations.

PS Ben's ok!

Parsimony is important in science and I think sunsaint's shark attack hypothesis provides the most straightforward and feasible explanantion yet.

I wasn't provided with roof info but would be happy to factor it in as a covariate if anyone has the data. There may well be other confounds that need to be controlled; colour of opposition players' shorts, for example

I miss Gardi too!!

[samoht]

superboot ..
I got a Pearson Product Moment correlation of 0.38 compared to your 0.42 - but I could only include the stats for 15 teams - as the online calculator, I used, only allows for 15 variable sets (input-wise).
So I've sort of rechecked your calculation and your 0.42 sounds right - as I'm thereabouts !

Here is the online link I referred to - for the statistically minded.

https://statistics.laerd.com/calculator ... ulator.php