The Only Stat That Matters - Disposal Efficiency

[bergholt]

I've run a few stats on the team's games this year to see what's important, and I reckon there are a few myths. Here are the things which don't matter, along with the correlation a high number in that stat has with margin.

Contested Possessions: -2% - no relationship at all
Long Kick Percentage: +5% - no relationship at all
One Percenters: -8% - no relationship
Hitouts: -11% - practically no relationship
Clearances: -19% - practically no relationship
Kick-to-Handball Ratio: +28% - not much relationship
Tackles: -30% - not much relationship, and what there is negative (high tackles means a loss)
Play-On Percentage: -37% - not much relationship, and what there is negative

Statistics show that none of these matter, despite media and fan focus on contested possessions, or on clearances, or on kick-to-handball ratio or play-on percentage.

Here's what matters:

Disposal Efficiency: +82%
Accuracy at Goal: +85%

If you don't know stats, this means that high numbers of either of these stats go with high margins in our favour, and vice versa. Don't believe me? Here are all our games so far this year, the margin, and the efficiency and accuracy stats:

R1: 8 point win, eff 71%, acc 71%
R2: 104 point win, eff 80%, acc 82%
R3: 28 point win, eff 73%, acc 56%
R4: 15 point win, eff 74%, acc 62%
R5: 10 point loss, eff 71%, acc 55%
R6: 3 point win, eff 76%, acc 54%
R7: 61 point loss, eff 67%, acc 41%
R8: 12 point loss, eff 76%, acc 46%
R9: 35 point win, eff 74%, acc 62%
R10: 47 point win, eff 79%, acc 70%
R11: 38 point win, eff 76%, acc 56%

Our four worst performances for the year were rounds 5-8, where our accuracy at goal was well down and disposal efficiency was pretty poor as well. On the other hand, our four biggest wins (round 2 + rounds 9-11) were some of our most accurate and efficient games.

So when the guy behind you at the footy shouts "Just hit a target!", he's right. It's not possible for us to do too much work on skills. Look at Geelong v West Coast last night. The Eagles would have won if they'd hit targets in the last quarter. But because Geelong have built a dynasty on good disposal, they could keep doing it under pressure.

It was also the difference on the 26th of September last year. And god forbid we should play Geelong in a final again this year, it will be again. Unless we start improving.

[Sainterone]

Milney set a new record on Friday night...

First time he has kicked 5 behinds in a game...

Shame he didn't get 7.3 or 8.2 or 9.1

still love ya Milney...

[Saintsfan]

I disagree with disposal efficiency. It is a good statistic if we win or get our fair share of the football however if we don't win clearances and don't cause turn overs we will not have the football in possession for long enough to win games.

Therefore disposal efficiency becomes ineffectual in arguing it is a good statistic we we never really had enough of the football or a fair share in the first place.

[samoht]

thanks bergholt ..

so from your stats - it looks like if you average the 2 important ones together (the disposal efficiency and acccuracy at goal) and you end up with a percentage of 65% or higher you most likely will end up with a win.

Accuracy at goal is the one we probably have most control of .. and it should improve when we're managing to take marks inside 50.

the disposal efficiency .. probably depends a lot on the effective chasing pressure the opposition puts on us and how effective their tackles are .. and a tackle stat probably doesn't indicate if it was an effective tackle.
You can tackle for example Chapman or Goddard and you could just end up bouncing off them... and so that shouldn't even count as a tackle... yet it probably is.

Also how do you quantify chasing pressure ?- it's not even a stat and yet we can see the result of a good chase.and pressure .. it definitely can affect disposal efficiency.

So there's probably other things that really matter that may not even be quantafiable... and these are not even counted as stats.

[BringBackMadDog]

Both those stats are heavily affected by the pressure placed on you by the opposition. That's why the % is lower in losses

[Teflon]

Both those stats are heavily affected by the pressure placed on you by the opposition. That's why the % is lower in losses

correct and conversely the pressure we put on opposition thats why IF we win clearances/1%ers/tackles et al it also means we are placing the opposition under extreme pressure..therefore......limiting the pressure we find ourselves under.

I do agree on 1 point with the OP - hitting targets by foot can alleviate pressure and damage opposition at same time (see Goddard/Dal Santo as 2 of our architects and most damaging players). Its why its now being scrutinised so heavily at draft camps.

[saint66au]

IMO its what Geelong are the best in the comp at..

You put them under extreme pressure, they get away what appears to be a wild kick out of trouble or a speculative handpass mid-tackle..but almost without exception they find a team-mate

Very very well drilled

[bergholt]

IMO its what Geelong are the best in the comp at..

You put them under extreme pressure, they get away what appears to be a wild kick out of trouble or a speculative handpass mid-tackle..but almost without exception they find a team-mate

Very very well drilled

yup, exactly. the pressure matters, sure, but the better your basic skills are, the better you'll be able to execute them under pressure.

[saintspremiers]

The Cats should kill us in a fortnight, even if we beat Freo.

The speed that they can move the ball in to their F50 is amazing.....poor Zac will be in a spot of bother!

[Milan Faletic]

The Cats should kill us in a fortnight, even if we beat Freo.

The speed that they can move the ball in to their F50 is amazing.....poor Zac will be in a spot of bother!

So then the midfield need to win the clearances or pressure Geelong's ball carriers. It's not necessarily the defence, it's the midfield pressure. If their supply is not as effective, our defence looks better. Shut down the engine and the train will stop.

[battye]

Tonight Finey was talking about Inside 50's being one of the most overrated stats, what do your calculations say about that bergholt?

[Con Gorozidis]

thanks bergholt. really good stuff. im sure it took u a while to work all that out.

are they instances where our contested ball and tackles are down but our DE and GE are high and we win anyway?

when i go to the footy and see teams lower down the ladder it is the first thing i notice and comment on. Their disposal and field kicking is 20-30% worse. and i reckon that is the main difference.

people have pointed out that DE is effected by pressure. But if u go and watch a game between the tiges and dees. They just miss targets all the time. even when there is no pressure. So i reckon youre stats really do tell a tale. look at the dogs this year. they are missing targets all the time and thats why they are finding themselves at the bottom of the 8.

[meher baba]

Intersting post, but I'm not sure what "important" means here?

All that you have shown, with a relativel small sample, is that a win on the scoreboard - especially a relatively easy win - is more likely to coincide with a higher % of accurate kicking both in general play and in front of goal. At least for us and at least in 2010.

This says nothing about cause and effect. I would suggest that, once a team gets truly on top in a game, the pressure goes off and kicking in general play gets a hell of a lot easier. And players become less nervous in front of goal and should be more accurate.

So, all being equal, I would expect that, the earlier a team gets on top of its opponent and the greater the disparity between the ability of the two teams, the better these stats would be. A bit like scoring rates in cricket.

For mine, winning the clearances, contested ball and, for the most part, the tackle count, are the key to winning the big games. Oh, and taking your genuine chances in front of goal: not always reflected in the goals/behinds ratio, because some behinds aren't real goal-scoring opportunities).

IMO FWIW.

But I love this sort of thread. So thanks to the OP.

[bergholt]

For mine, winning the clearances, contested ball and, for the most part, the tackle count, are the key to winning the big games.

incidentally, i'm not saying that these are general rules, applicable across all sides in the competition, or across all time. all i'm saying is that these are important to the st kilda team and gameplan that we have in 2010. that's all the stats tell me at this point.

clearances.

our lowest clearance numbers of the year came against richmond, freo, collingwood and the dogs. we won all those games. we had higher numbers in our three losses than in any of those four games.

but maybe looking at the difference in the clearances is more important.

R1: 8 point win, -7 clearances
R2: 104 point win, -4 clearances
R3: 28 point win, -4 clearances
R4: 15 point win, +21 clearances
R5: 10 point loss, +5 clearances
R6: 3 point win, +2 clearances
R7: 61 point loss, -6 clearances
R8: 12 point loss, +6 clearances
R9: 35 point win, +3 clearances
R10: 47 point win, +3 clearances
R11: 38 point win, -12 clearances

i defy you to find a pattern in that!

our biggest win came at a clearance deficit. our third biggest came at a substantial clearance deficit. on the other hand, our biggest clearance win only gave us a 15 point victory over freo. (incidentally, that was against the most dominant ruckman in the comp - what does that say about the importance of a ruckman?)

there's just no pattern.

how about contested possessions? here are our games sorted by contested possession count, with the differential from the opposition added in:

R8: 12 point loss, 168 cont poss, +24 diff
R5: 10 point loss, 164 cont poss, +5 diff
R4: 15 point win, 163 cont poss, +19 diff
R2: 104 point win, 161 cont poss, +28 diff
R9: 35 point win, 158 cont poss, +2 diff
R3: 28 point win, 155 cont poss, -15 diff
R1: 8 point win, 148 cont poss, -22 diff
R7: 61 point loss, 147 cont poss, -12 diff
R6: 3 point win, 144 cont poss, -3 diff
R10: 47 point win, 144 cont poss, +4 diff
R11: 38 point win, 128 cont poss, -17 diff

sometimes when we win, we smash the other team in contested possessions - north, freo. sometimes we smash them and lose! this happened against essendon and carlton. against both sydney and richmond we got smashed in contested possessions but won anyway.

a much better predictor is difference in uncontested possessions, believe it or not:

R2: 104 point win, 320 uncont poss, +101 diff
R11: 38 point win, 308 uncont poss, +100 diff
R3: 28 point win, 295 uncont poss, +98 diff
R10: 47 point win, 336 uncont poss, +79 diff
R8: 12 point loss, 269 uncont poss, +56 diff
R4: 15 point win, 298 uncont poss, +47 diff
R9: 35 point win, 248 uncont poss, -9 diff
R5: 10 point loss, 282 uncont poss, -19 diff
R6: 3 point win, 303 uncont poss, -19 diff
R1: 8 point win, 210 uncont poss, -42 diff
R7: 61 point loss, 218 uncont poss, -57 diff

when we're winning the uncontested possessions, we usually win. essendon is the exception here. when we're losing them, we usually lose or scrape over the line. west coast the only real exception here.

what does this mean?

when we're up and running to receive, we do well. we're tough enough and good enough at the contest that we're always going to go ok in contested possessions, but we need to have plenty of running outside that in order to actually score goals and win games.

to me this explains why players like ball and armitage have been marked hard. armo's currently ranked 19th for uncontested possessions, behind geary and mcqualter and baker and peake and blake. in order to be a midfielder in this side, he needs to get that number up, because it's important for us in winning.

long-winded response, but i think the stats can tell us some interesting things. our innate prejudices are not always accurate.

[samoht]

Analysing stats after the fact - after the game is won or lost, is probably only revealing things that you already sensed and saw first hand during the match itself.

It's what you can do during the match that counts - what you can control there and then ..e.g. getting the match ups right -the right team selection, covering your weaknesses and emphasising/using your strengths.

For instance if you see Harbrow get 36 running stats out of defence against you and you didn't have a quick player to negate that during the match or next time you play .. then maybe a more important stat may be good/bad team selection stat or a coach not taking action stat... or the recruiter not having the foresight to pick a Judd stat..

You can invent as many stats as you like ... but it's what you can control, such as taking a mark during the match instead of dropping it and negating Harbrow etc.. that counts.

[BAM! (shhhh)]

I've run a few stats on the team's games this year to see what's important, and I reckon there are a few myths. Here are the things which don't matter, along with the correlation a high number in that stat has with margin.

Contested Possessions: -2% - no relationship at all
Long Kick Percentage: +5% - no relationship at all
One Percenters: -8% - no relationship
Hitouts: -11% - practically no relationship
Clearances: -19% - practically no relationship
Kick-to-Handball Ratio: +28% - not much relationship
Tackles: -30% - not much relationship, and what there is negative (high tackles means a loss)
Play-On Percentage: -37% - not much relationship, and what there is negative

Statistics show that none of these matter, despite media and fan focus on contested possessions, or on clearances, or on kick-to-handball ratio or play-on percentage.

Here's what matters:

Disposal Efficiency: +82%
Accuracy at Goal: +85%

If you don't know stats, this means that high numbers of either of these stats go with high margins in our favour, and vice versa. Don't believe me? Here are all our games so far this year, the margin, and the efficiency and accuracy stats:

Great post (only key stat missing is turnovers... from Herald Sun match reports, that would essentially be Clangers - FA).

Which will raise the question of "why", these things are true.

My first question would be whether you're normalising the totals? If not, then you'll immediately skew the stats - a team like St Kilda, who play stoppage ball, will necessarily create high numbers for both teams in certain stats, which effectively creates "noise" in trying to create correlation.

Second, is that 11 rounds worth of AFL games, or 11 rounds worth of St Kilda games?

If the latter, as well as the sample size issue, you're running into tailored numbers. Different teams are trying to beat us in different ways.

For example, another touted "myth" stat is overall disposals. By an large, for individual games, winning the stat has little correlation to winning the game. However the high disposal teams overall tend to be the ones who spend more time with the ball, and these tend to win more games. Last years top 5 ladder teams were also the top 5 disposal teams.

So, onto furthering your model:

What about statistical correlation to each other. i.e. CP -> CL -> I50 -> Shots -> Goals -> Wins.

CP may not correlate highly to wins & goals, but if it correlates highly to the next stat in the chain, it's important.

Then there's removing the noise as well... personal theory of mine, Behinds -> I50. Teams set up their press, turn it over on the HFF, and get another opportunity.

As each piece of high correlation is found, remove it, and find the next piece of correlation, and the next, and the next etc. etc.

And you gradually come up with a gameplan.

Like I said, very interesting post

[samoht]

Some Stats may look like a strength but may actually disguise a weakness - e.g. you can get the ball inside your F50 say 70 times only to see it turned over and rebound straight out of your half forward line 50 times - so that only leaves you 20 F50's to work with.

Maybe inside 50 stats need to be read alongside opposition rebound stats to get a more accurate picture of how effective they are.

Also a F50 stat where you've used your pace and the corridor to deliver a pass that gets marked in the goal square beats 3 inside 50's where the ball zig zags in and out and gets turned over or marked 60 metres out.

So stats may hide more than they reveal..
Bottom line quality always beats quantity... and the scoreboard doesn't lie.


Also -
re :Disposal efficiency .. it's easy to have a high % disposal efficiency if you're just chipping the ball around and backwards - but will that win you games ?

[rexy]

Some Stats may look like a strength but may actually disguise a weakness - e.g. you can get the ball inside your F50 say 70 times only to see it turned over and rebound straight out of your half forward line 50 times - so that only leaves you 20 F50's to work with.

Maybe inside 50 stats need to be read alongside opposition rebound stats to get a more accurate picture of how effective they are.

Also a F50 stat where you've used your pace and the corridor to deliver a pass that gets marked in the goal square beats 3 inside 50's where the ball zig zags in and out and gets turned over or marked 60 metres out.

So stats may hide more than they reveal..
Bottom line quality always beats quantity... and the scoreboard doesn't lie.


Also -
re :Disposal efficiency .. it's easy to have a high % disposal efficiency if you're just chipping the ball around and backwards - but will that win you games ?

Agree with your point, not sure on your mathematics though, inside 50s that get turned over 60 metres from goal? Never actually got inside 50 to begin with so are not inside 50s at all? I would have thought.

Dont read to much into stats IMO, easy to find stats to support why you lost, not so easy to prove stats that people claim are the reason for winning.

Kicking straight, contested ball, tackles, clearances, inside 50s, scoring shots, yada yada yada are all more or less important depending on the relativity of the exact mment in time that they happen IMO.

1 tackle or turnover or amazing field kick, snap from the boundary line, intercept or ruck tap can turn the whole game on its head if it is at a crucial time.

Capitalising on opportunities might sum it up the best.

[degruch]

Regardless of the stat's importance, I believe Freo were pretty poor around the ground efficiency wise when pressured by Adelaide (I didn't see much of the game, going on reports), which is encouraging.

IMO, we've a long way to go before we're as bad as Port and the Hawks were on the weekend, and we know we have the improvement in us.

[BAM! (shhhh)]

Some Stats may look like a strength but may actually disguise a weakness - e.g. you can get the ball inside your F50 say 70 times only to see it turned over and rebound straight out of your half forward line 50 times - so that only leaves you 20 F50's to work with.

Maybe inside 50 stats need to be read alongside opposition rebound stats to get a more accurate picture of how effective they are.

Also a F50 stat where you've used your pace and the corridor to deliver a pass that gets marked in the goal square beats 3 inside 50's where the ball zig zags in and out and gets turned over or marked 60 metres out.

So stats may hide more than they reveal..
Bottom line quality always beats quantity... and the scoreboard doesn't lie.


Also -
re :Disposal efficiency .. it's easy to have a high % disposal efficiency if you're just chipping the ball around and backwards - but will that win you games ?

What you're talking about - and you're right - is context.

Frequently you hear commentators tell us we can throw whatever stats away/stats aren't telling the whole story/yada yada yada.

The purpose of stats is to put some science behind our opinions; the OP does it brilliantly. Your point about chipping it around the backline may actually be backward in context of the OP.

Ball retention via kicking around the backline bothers a lot of people at the games. The Moorabbin wing itself has been known to erupt in boos. However, switching the ball and moving it backward serves to make the opposition cover more of the field rather than simply flooding. This opens up further uncontested kicks, and safer roads to enter the forward 50.

So we can look at the metres gained for Blake and Dawson and shake our heads, or we can look at their ridiculous efficiency numbers and nod.

The question is which one really drives scoring and winning. The OP's numbers at the very least hint that Blakey kicking it backwards to Dawson going sideways to Backer chipping it up to Raph may not present the excitement of the long bomb kick... but that at the very least, boosting those numbers will anecdotally correlate to a better chance of scoring, even if the long bomb kick lands inside 50.

Btw, there are stats to represent most of what you're talking about... they just aren't made available to the public (for free anyway), and most of the commentators don't use them. Stats are kept on where on the ground the ball is received, disposed of, etc.

Marks inside 50 are more important than inside 50s. Opposition rebound 50s are nearly as important. All these serve to tell us things.

A base metric won't lie... we'll just interpret it wrong or use it out of context. Hence there being lies, damn lies, and statistics.

[OneEyedSainter77]

bergholt I would love to read this thread and I will... but at the moment, I can't make out how you get the percentages... can you (or someone) dumb it donw for me as I really am stupid.

[samoht]

BAM

You make a good point - our "round the cape" chipping style is aimed at someone finally delivering a break out kick (a Goddard say) to someone in space and therefore lead to some goals we would otherwise perhaps miss out on.

So I probably shouldn't be totally disparaging or dismissive of that style... although that style (as you concede too) obviously clearly and perhaps artificially improves (and therefore maybe even makes a mockery) of our disposal efficiency stats.

However .. (you must agree too)
It's a pity we need to kick a large % of our goals in this laborious strategic manner (like a German soccer team) - whereas Essendon say had 70 fewer possessions and were able to use their pace to take control of the corridor and be more spontaneous (not to mention exciting) and effective - it happened a lot easier for them.

So at the end of the day .. the best stat for efficiency may be the possession/goal stat - the lower that is, the more efficient you are and the less hard you had to work for each goal - but then again you'd rather work harder (if racking up possessions is considered harder work) as long as you win.

Winning is the best stat of all... and the end (a winning result) justfies the means no matter how painful , laborious or inefficient they are.

[BAM! (shhhh)]

BAM

You make a good point - our "round the cape" chipping style is aimed at someone finally delivering a break out kick (a Goddard say) to someone in space and therefore lead to some goals we would otherwise perhaps miss out on.

So I probably shouldn't be totally disparaging or dismissive of that style... although that style (as you concede too) obviously clearly and perhaps artificially improves (and therefore maybe even makes a mockery) of our disposal efficiency stats.

However .. (you must agree too)
It's a pity we need to kick a large % of our goals in this laborious strategic manner (like a German soccer team) - whereas Essendon say had 70 fewer possessions and were able to use their pace to take control of the corridor and be more spontaneous (not to mention exciting) and effective - it happened a lot easier for them.

So at the end of the day .. the best stat for efficiency may be the possession/goal stat - the lower that is, the more efficient you are and the less hard you had to work for each goal - but then again you'd rather work harder (if racking up possessions is considered harder work) as long as you win.

Winning is the best stat of all... and the end (a winning result) justfies the means no matter how painful , laborious or inefficient they are.

Let's be clear: there is no "artificial" inflation of anybody's numbers. The first rule of stats is to look at them for what they are, not what we want them to be (which is where the disparagement of disposals tends to come from).

All an effective disposal means is that you retained the ball. It doesn't mean that it's been a 75 metre torp that was so good that it landed in the hands of a player who was looking the wrong way and falling down - or anyone else playing in the forward line with Kosi at the time either.

The numbers are what the numbers are. If you want to find something out the numbers you're using aren't telling you, use different numbers (e.g. your scoring efficiency calc).

All the OP is telling us is that when teams retain the ball from their own possession, they're more likely to outscore the opposition. Sound intuitive? But didn't we just say that efficiency was "artifically inflated".

Your possession/goal stat is case in point. If I'm low possession/goal, it probably means I'm a high scoring team - which is great... but that's going to leave more time for the opposition to hold the ball. Which is better? Well, the top 5 teams for disposals last year were the Saints, Cats, Dogs, Pies and Crows.

Check out where those 5 teams finished on the ladder... I'm told disposals-goals correlated to wins early in the decade. They don't now (IIRC, the leader in that stat last year was Carlton, but I wouldn't swear to it).

FWIW, I don't consider "around the horn" either boring or painful. Done well, I love seeing the execution of a hapless opposition defense. The reverse is true as well - I love seeing the Saints defense repel the Bulldogs whenever they try to outpossess.

[degruch]

FWIW, I don't consider "around the horn" either boring or painful. Done well, I love seeing the execution of a hapless opposition defense. The reverse is true as well - I love seeing the Saints defense repel the Bulldogs whenever they try to outpossess.

I agree.

[degruch]

Interesting for stats fans:

http://www.afl.com.au/news/newsarticle/ ... fault.aspx

Our mids (those chosen) get a lot of the ball for a pathetic return. I'm sure this would read differently if split into pre-Roo hammy, post-Roo hammy wobbly games, and now.

Interesting comparison...I wouldn't have chosen Goddard CJ and Lenny for the comparison (against Geelong's Chapman Selwood and Bartel), would have thought Monty and NDS were more responsible for carrying it out of the centre, plus CJ's a tagger and more akin to Ling. Am I imagining things?