When it comes to predicting scores, it's like QD trying to bake—a bit hit or miss but always entertaining.

16 hours ago, Binmans PA said:

The most startling stat in that lot is Essendon...😳

And what a pity they aren't bottom 3.

Just a guess, that when half the AFL Board and half the TWSNBN Board know each other quite well, what should we expect.

1 hour ago, Lucifers Hero said:

I would agree the GF teams are likely to be higher in defence.  Which makes Freo interesting, making up the top 4 on the chart for defence with Syd, Geel and Dees.   In fact they rank higher than Syd and Geelong.  Our 2 games vs Freo will be very interesting.

Not their year but Freo are building.

Agree re Freo. They are dangerous and have played very well against us over the last few years. Not ideal to have them twice this year but such is the way the fixture works. 

49 minutes ago, deanox said:

That is such a lazy stat too.

It's simply top 6 in the points F and A. It's not even something like "within a set percentage of the leader". Meaning someone could be 7th, by 1 point, and they miss out, even though they are performing better than in previous years.

It also doesn’t take into account fixturing - particularly in assessing it at Round 8. So whilst Sydney and Geelong are top 6, Sydney has played 5 of the top 10 but Geelong has only played 2. 

It also of course doesn’t take into account injuries. I continue to believe Carlton’s defensive profile is impacted by the absences of Saad and McGovern, not sure they account for them being bottom 5 defensively but they would be better with the injured players back.

43 minutes ago, titan_uranus said:

I continue to believe Carlton’s defensive profile is impacted by the absences of Saad and McGovern, not sure they account for them being bottom 5 defensively but they would be better with the injured players back.

I agree with this, but they don't seem to have a great focus on developing defensive depth like we do. We've been actively developing replacements for Lever/May (in Adams and Turner), and have Petty who can swing backwards or forwards too.

They seem all in on their forward half, which is mostly a result of them nailing the Curnow and McKay picks in that same draft.

They have been unlucky with Jones, who has become a very good intercept marker. Put him next to Weitering, McGovern and even Marchbank and Kemp, and that's a good defensive line. So losing him back to the Dogs has hurt them, but they haven't really replaced him.

Docherty's injury has hurt their half back drive too.

Outside of Weitering in their current set up, all their defenders look vulnerable 1v1, and part of that is probably a lack of pressure from the rest of the team. 

When Carlton were on the up from the midway point of last year (after some Carlton fans wanted to sacked Voss), it was their all team defence that had clicked. They've lost that again, which suggests either they're not fit enough or it was a flash in the pan tactically, or both.

I still think Voss has a Rolls Royce that he is driving like my Hyundai Tucson.

Edited May 7, 2024May 7 by Binmans PA

16 hours ago, DubDee said:

love your work mate so not having a go. There are worst stats for sure but some people infer too much into this one. Not just in footy but soccer and others. 

So much comes down to who performs when under pressure, who kicks straight, so I’m not a fan of the expected score

14 hours ago, rpfc said:

I’m sorry, but this is a bad stat. It is just trying to place hard truth upon contests and shots at goal under pressure where there is only nuance and grey. 

There is very little insight with this particular statistic.

It's clearly not a meaningless or bad stat if the clubs put a lot of stock in it. Can you say that a team should have won a match because they won on expected scores by a small margin? No, you can't. So articles like the above from Fox Sports don't really help with interpretation of expected scores.

Goalkicking is not entirely about "who performs under pressure" - there's an element of that, but there is also an element of luck with goalkicking.

Regardless, expected scores provide two very useful and insightful metrics (even completely ignoring the "who should have won" perspective).

Average expected score per shot
This metric is a measure of average shot difficulty.

Melbourne supporters bemoan Melbourne's strategy of kicking it to the pocket because it leads to more difficult shots. This metrics quantifies this. Melbourne rank 18th for average shot difficulty in 2024 (16th last year) and 18th for set shots only (17th last year).

Average difference between actual and expected score per shot
This metric is a measure of goalkicking accuracy, and much better than raw accuracy (goals divided by shots).

People already look at the final tally of goals and behinds and draw conclusions such as a team should have won by more or were lucky to win. Two recent examples are:
 - Essendon defeating West Coast 11.11 to 11.5
 - Adelaide defeating Carlton 16.4 to 14.14

On the surface, you assume West Coast were lucky to get as close thanks to their accurate kicking and you assume Adelaide was very lucky to win thanks to their apparent ridiculously accurate kicking.

 - West Coast had 24 shots to 23 and won expected scores 89.9 to 80.7.
 - Adelaide had 25 shots to 26 (despite it being 20 "scores" to 28) and only lost expected scores 92.8 to 95.8.

I'm not convinced about the "expected scores" stat because (apart from shots at goal after the siren at any quarter), the result of every kick for goal changes the next play. Every time a team kicks a behind, they give the ball back to the opposition.

A more impressive analysis would point out that Melbourne is still trying to work out what its best forward mix is while all the while having won 6 out of 8 games. (Credit where it's due: This was stated by David King after round 7, obviously when it was a 5 out of 7 score line). 

Expected score counts for nothing at the end of the day.

We beat the Pies and Blues in last years finals on expected score. Unfortunately Pies won the flag.

Expected score win and you lose the actual game = bad kicking is bad football!!!

2 hours ago, WheeloRatings said:

It's clearly not a meaningless or bad stat if the clubs put a lot of stock in it. Can you say that a team should have won a match because they won on expected scores by a small margin? No, you can't. So articles like the above from Fox Sports don't really help with interpretation of expected scores.

Goalkicking is not entirely about "who performs under pressure" - there's an element of that, but there is also an element of luck with goalkicking.

Regardless, expected scores provide two very useful and insightful metrics (even completely ignoring the "who should have won" perspective).

Average expected score per shot
This metric is a measure of average shot difficulty.

Melbourne supporters bemoan Melbourne's strategy of kicking it to the pocket because it leads to more difficult shots. This metrics quantifies this. Melbourne rank 18th for average shot difficulty in 2024 (16th last year) and 18th for set shots only (17th last year).

Average difference between actual and expected score per shot
This metric is a measure of goalkicking accuracy, and much better than raw accuracy (goals divided by shots).

People already look at the final tally of goals and behinds and draw conclusions such as a team should have won by more or were lucky to win. Two recent examples are:
 - Essendon defeating West Coast 11.11 to 11.5
 - Adelaide defeating Carlton 16.4 to 14.14

On the surface, you assume West Coast were lucky to get as close thanks to their accurate kicking and you assume Adelaide was very lucky to win thanks to their apparent ridiculously accurate kicking.

 - West Coast had 24 shots to 23 and won expected scores 89.9 to 80.7.
 - Adelaide had 25 shots to 26 (despite it being 20 "scores" to 28) and only lost expected scores 92.8 to 95.8.

What would be interesting is going back through history and seeing how goal kicking accuracy has changed in the way you define it (ie. adjusted for difficulty) . Simply looking at raw accuracy is not a robust method.   

For every day Schwartz was kicking them from both scg boundaries there was surely days where Mooney and his team mates were missing from the goal square.  

Other sports do this quite well, particularly baseball.  I also wonder if team scientifically assess the best kick in their teams from varying positions  Eg,  have every player take shots from varying positions inside 50 and record their results.   Of course match day pressure is not factored in but it does provide a benchmark of a players goal kicking skill from certain positions.   Eg max from outside 50 and Fritter from a right forward pocket 😃

32 minutes ago, D4Life said:

Expected score counts for nothing at the end of the day.

We beat the Pies and Blues in last years finals on expected score. Unfortunately Pies won the flag.

Expected score win and you lose the actual game = bad kicking is bad football!!!

Ah, but how do you define/measure "bad kicking"? Teams will always miss some shots at goal - at what point do you say it was bad kicking?

It's not about celebrating an expected score win, it's about understanding where/how you lost the game. Expected scores helps to differentiate whether you were getting good shots on goal but not executing, or you weren't getting enough good shots at goal.

I would hope the post-game analysis by coaching staff is more in depth than "we lost, let's not do that next time".

Single stats in isolation don't define a game. Nobody says they do. And data doesn't explain things, it just provides the opportunity to search for observable patterns and interpret that information. Outright dismissing the 'expected score' stat is like saying that 'disposals' are meaningless, although everybody already knows some disposals are more damaging than others. I get the sense that the hostility toward 'expected score' comes back to the AFL fan aversion to 'excuses'. 

Anyway, back to the topic at hand. From memory, the almost inexplicable run of inaccuracy on Saturday from both teams coincided with an unusually long period of very few stoppages. Is this correct? Would love to see the relationship between measures of fatigue and accuracy and how rotations etc. might impact expected score. 

1 hour ago, WheeloRatings said:

Ah, but how do you define/measure "bad kicking"? Teams will always miss some shots at goal - at what point do you say it was bad kicking?

It's not about celebrating an expected score win, it's about understanding where/how you lost the game. Expected scores helps to differentiate whether you were getting good shots on goal but not executing, or you weren't getting enough good shots at goal.

I would hope the post-game analysis by coaching staff is more in depth than "we lost, let's not do that next time".

I was very interested in expected score last year, but after the finals lost interest!

I’m sure they do analyse all aspects!

3 hours ago, La Dee-vina Comedia said:

I'm not convinced about the "expected scores" stat because (apart from shots at goal after the siren at any quarter), the result of every kick for goal changes the next play. Every time a team kicks a behind, they give the ball back to the opposition.

A more impressive analysis would point out that Melbourne is still trying to work out what its best forward mix is while all the while having won 6 out of 8 games. (Credit where it's due: This was stated by David King after round 7, obviously when it was a 5 out of 7 score line). 

As @WheeloRatings says, you can use it to understand whether or not your misses were shots you should have taken or were just tough shots. That gives it value. It then gets mis- and over-used such as is the case by Fox in the discussed article. 

4 hours ago, WheeloRatings said:

It's clearly not a meaningless or bad stat if the clubs put a lot of stock in it. Can you say that a team should have won a match because they won on expected scores by a small margin? No, you can't. So articles like the above from Fox Sports don't really help with interpretation of expected scores.

Goalkicking is not entirely about "who performs under pressure" - there's an element of that, but there is also an element of luck with goalkicking.

Regardless, expected scores provide two very useful and insightful metrics (even completely ignoring the "who should have won" perspective).

Average expected score per shot
This metric is a measure of average shot difficulty.

Melbourne supporters bemoan Melbourne's strategy of kicking it to the pocket because it leads to more difficult shots. This metrics quantifies this. Melbourne rank 18th for average shot difficulty in 2024 (16th last year) and 18th for set shots only (17th last year).

Average difference between actual and expected score per shot
This metric is a measure of goalkicking accuracy, and much better than raw accuracy (goals divided by shots).

People already look at the final tally of goals and behinds and draw conclusions such as a team should have won by more or were lucky to win. Two recent examples are:
 - Essendon defeating West Coast 11.11 to 11.5
 - Adelaide defeating Carlton 16.4 to 14.14

On the surface, you assume West Coast were lucky to get as close thanks to their accurate kicking and you assume Adelaide was very lucky to win thanks to their apparent ridiculously accurate kicking.

 - West Coast had 24 shots to 23 and won expected scores 89.9 to 80.7.
 - Adelaide had 25 shots to 26 (despite it being 20 "scores" to 28) and only lost expected scores 92.8 to 95.8.

Hang on a second. I am talking about insight, the stat itself can help paint a picture as much as my eyes and subjective (yet educated) opinion can, but the insight gleaned by some from this stat- including our coach apparently - is patently misguided. 

Any team that prioritises territory over deliberate movement is going to have more entries, more shots at goal, more ‘repeat’ shots at goal that make the previous ‘miss’ irrelevant to this particular stat as it would not have existed if not for the previous ‘miss’.

It is by definition a flawed stat, and, taken on its own, irrelevant.

But that’s all stats I suppose, but none get treated like this one, there is an ‘expected score’ ladder floating around for crying out flamin’!

5 minutes ago, titan_uranus said:

As @WheeloRatings says, you can use it to understand whether or not your misses were shots you should have taken or were just tough shots. That gives it value. It then gets mis- and over-used such as is the case by Fox in the discussed article. 

So is it a lie or a damned lie?

In general the media does a very bad job of interpretation of stats.

XScore does not represent what the score "should" have been and it can't be interpreted as who "should" have won. When Fox creates a ladder like this it's simplistic at best and bad for public understanding of the art of statistical analysis at worst.

 

XScore tells us about the number of shots generated.

XScore tells us about the quality of the shots generated (good or difficult scoring areas).

XScore effectively gives us a measure of the expected score (based on expected accuracy) of the shots generated. 

It doesn't try to predict the future or alternative outcomes.

 

Why does Goody use it? I expect he is using it as a measure of "is the game plan working? Are we generating sufficient shots of the right quality? Are we restricting the opposition shot quality?".

It helps Goody discuss outcomes with the team without worrying about the scoreboard. "Don't worry about the score, here is proof we are generating the right opportunities" or "don't get ahead of yourself, we've been lucky so far, let's get back to basics".

 

Of course XScore is influenced by game plan. It will take into a lot of measures but not all of them. I'm not sure but I doubt it includes:

- player specific data (just AFL averages);

- fatigue at the macro (what time of the quarter/game is it) or micro (what work rate has the player shooting for goal output over the last minute);

- weather conditions like wind, rain, humidity;

- other external factors like home crowd advantage and pressure;

- the impact of repeat entries building fatigue or mental pressure etc versus a more relaxed game plan.

So it can be improved. But it is probably something that the club's can track for themselves. Ie last year we often "won" XScore based on our repeat entries, bit maybe or XScore should only be compared against our XScore, not against Geelong's.

 

Overall it is a helpful metric that tells you some things, particularly in conjunction with other metrics. It's not an alternative future predictor.

 

Edited May 7, 2024May 7 by deanox

43 minutes ago, rpfc said:

It is by definition a flawed stat, and, taken on its own, irrelevant.

But that’s all stats I suppose, but none get treated like this one, there is an ‘expected score’ ladder floating around for crying out flamin’!

That was partly my point - that you can use it in conjunction with actual score and number of shots to derive useful metrics. Ok, maybe not so insightful, but still useful. And it's a good thing to be able to use data to quantify things that your eye tells you. It's certainly flawed, but that doesn't mean it's irrelevant.

Re the expected scores ladder, they literally have "free kick" ladders which people use as evidence of bad or biased umpiring. Expected scores are more meaningful than free kick counts. But I agree it's overdone in the media.

2 minutes ago, deanox said:

I'm not sure but I doubt it includes:

- player specific data (just AFL averages);

- fatigue at the macro (what time of the quarter/game is it) or micro (what work rate has the player shooting for goal output over the last minute);

- weather conditions like wind, rain, humidity;

- other external factors like home crowd advantage and pressure;

- the impact of repeat entries building fatigue or mental pressure etc versus a more relaxed game plan.

You are correct, it doesn't take any of that into account.

4 hours ago, La Dee-vina Comedia said:

A more impressive analysis would point out that Melbourne is still trying to work out what its best forward mix is while all the while having won 6 out of 8 games. (Credit where it's due: This was stated by David King after round 7, obviously when it was a 5 out of 7 score line). 

Personally I love this narrative. I also have a theory that the FD also love this narrative.

The club is not trying to work it out - nor has it done so for the last half a decade, sure new people (JVR) have rotated through, but that's the process. Just like we laud the depth of the backline to stand up with main stays who are KPD's, there is an expectation that this is the same for the forwardline.

The mainstays are Fritsch and Pickett, sure two bigger bodies are required, but JVR, Brown, TMAC, Petty, have all rotated through for 4/5 years - with the interpretation that it's unsettled. And that it lets Fritsch, off the leash so to speak... 

Not one other team has this narrative, so lazy journalism just perpetuates....

All the journalists are 40+ years of age, and hark back to their halcyon days of big lead up forwards.... it's not done like that anymore with the zone defenece...,we all know it, but we still desire the big key forward as an exemplar.

but that's still the narrative pushed by the media...

It's why Fritsch is still underrated even AFTER he kicked 6 goals in a grand final...

Anyway I love the fallacy that we are still trying to sort out the mix, and I bet Goodwin does too.

and that's my take.

 

 

Edited May 7, 2024May 7 by Engorged Onion

I’m still waiting for the club to release its ‘Expected Scores Premiers 2023’ cheeseboard.

It comes with a pack of Kraft Singles when you were expecting Stilton Gold.

Top of 22

Feels like 14.

It also fails to take into account who the various teams have played.  Playing North, Hawthorn and WCE already will give a far greater opportunity to be in the top right corner than having played only one of these.

14 hours ago, Skuit said:

Anyway, back to the topic at hand. From memory, the almost inexplicable run of inaccuracy on Saturday from both teams coincided with an unusually long period of very few stoppages. Is this correct? Would love to see the relationship between measures of fatigue and accuracy and how rotations etc. might impact expected score. 

No goals in the second quarter resulted in a really short quarter, 25 minutes. There was a period of 11 minutes straight without a stoppage.