Coach Justin Longmuir Pt 2

[Docker82]

So I picked a random game (of a finalist) from round 22 last year. In Port's team they had the following players in the 22-24 age bracket:

Butters, Rozee, Bergman, Duursma, Farrell, Williams, Hayes, McEntee, Evans

So leaving the two stars out of it (because Serong and Bray compare), the others are either plodders or blokes who I am sure would like to think their best football is ahead of them and significantly better than what it has been to date.

If Bergman and Farrell were playing for Freo this board would be talking them up for potential All Australians. They’re very good footballers.

Williams and McEntee are solid and play their role every week too.

Hayes and Duursma are already gone and Evans probably will be soon but there all bottom six players at best. Port made the decision these guys weren’t good at that age and moved two of them on - Exactly like I said happens.

I don’t think many AFL players that aren’t best 22 at 22 years old and have spent 3-4 years+ in the system have very long careers. Can you name one at Freo other than guys constantly on the fringe? I’m struggling

 

[Joao]

If Bergman and Farrell were playing for Freo this board would be talking them up for potential All Australians. They’re very good footballers.

Williams and McEntee are solid and play their role every week too.

Hayes and Duursma are already gone and Evans probably will be soon but there all bottom six players at best.

What this board would think of them is irrelevant. How good is what matters. Bergman averages 14 touches. Farrell averages 16. They are middling players at best getting carried by some stars and a bunch of more mature players.

 

[Docker82]

What this board would think of them is irrelevant. How good is what matters. Bergman averages 14 touches. Farrell averages 16. They are middling players at best getting carried by some stars and a bunch of more mature players.

I think you’re way off the mark with Bergman tbh. Was very good for Port in the second half of last season. I think he’ll be in their top ten players this year tbh.

 

[Snuffaluphagus]

What this board would think of them is irrelevant. How good is what matters. Bergman averages 14 touches. Farrell averages 16. They are middling players at best getting carried by some stars and a bunch of more mature players.

Bergman has the added benefit of helping JHF steal extra brownlow votes though

 

[Joao]

I think you’re way off the mark with Bergman tbh. Was very good for Port in the second half of last season. I think he’ll be in their top ten players this year tbh.

Unless he is kicking 3 goals a week, he wasn't doing much damage with 14 touches.

 

[WilloTree]

It isn't five years to win a flag. It is if, in your first coaching gig as an AFL coach, you haven't made a prelim in your first five years you are highly unlikely to win a flag. The negative predictive value is very high. The positive predictive value is not so high. In other words, quite a few coaches make a prelim in their first five years but don't go on to win a flag. But if you don't make one, you're virtually no hope, Hardwick the one exception.
You can argue all you want about confounding factors, but the data is the data. I'm at least trying to present a quantitative and qualitative evidence base for my view that the bloke isn't up to it. For those that think he's up to it, what are your reasons?

Thanks for that. Whilst it isn't a prelim, JL made a semi in his 3rd year so not a mile off.

I'd say that age profile and games experience would be a good indication. I'd hazard a guess Sydney, Hawks, Richmond, West Coast, Geelong were all in the top few clubs for both. Leigh Matthews says it's basically based on simply having the best players and knowing how to use them. Malcolm Blight says there are loads of factors including more than a bit of luck.

I also don't think the only measure of a coach is Premiership or bust. Ross Lyon and Rodney Eade for instance.

 

[The Feeder Club]

Thanks for that. Whilst it isn't a prelim, JL made a semi in his 3rd year so not a mile off.

I'd say that age profile and games experience would be a good indication. I'd hazard a guess Sydney, Hawks, Richmond, West Coast, Geelong were all in the top few clubs for both. Leigh Matthews says it's basically based on simply having the best players and knowing how to use them. Malcolm Blight says there are loads of factors including more than a bit of luck.

I also don't think the only measure of a coach is Premiership or bust. Ross Lyon and Rodney Eade or instance.

I can see logic in everything you say, even if I don't agree with it all. Was Bomber Thompson a great coach for winning two flags or a dud coach for leaving at least one on the table that he should have won with that list? That's a debate for another place and probably not for a club that has never won one. We need a premiership coach, badly. I don't think we've got one, hope I am wrong

 

[David Bubb]

JL interview on Backchat being aired on YouTube now for anyone interested

 

[The Lobbster]

It isn't five years to win a flag. It is if, in your first coaching gig as an AFL coach, you haven't made a prelim in your first five years you are highly unlikely to win a flag. The negative predictive value is very high. The positive predictive value is not so high. In other words, quite a few coaches make a prelim in their first five years but don't go on to win a flag. But if you don't make one, you're virtually no hope, Hardwick the one exception.
You can argue all you want about confounding factors, but the data is the data. I'm at least trying to present a quantitative and qualitative evidence base for my view that the bloke isn't up to it. For those that think he's up to it, what are your reasons?

The negative value is nowhere near as high as this makes it sound. If you look back from 2014-2024, there have been 16 coaches to coach 5 years or more (allowing for backdating as long as you coached in '14, only counting stints at each club individually).

12 have made a prelim within the first 5 years, with 6 going on to win a flag (C Scott, A Clarkson, J Longmire, A Simpson, L Beveridge and S Goodwin), giving a 50% flag rate. Its important to note that Hinkley and Fagan are still ongoing but you can't analyse the future.

Only 4 have coached 5 years without a prelim in those first 5. Of them, only 1 has won a flag (Hardwick), giving a 25% flag rate. However, you can't compare these two metrics given the vast difference in dataset size. Interestingly, if you include Longmuir and Nicks in the "5 years, no prelim" crew, only Longmuir has won a final within that time. Hardwick did not win a final within his first 5 years nor did A Richardson, J Worsfold (Essendon) or S Dew.

As for better data, age and experience relative to the rest of the comp is slightly more robust although recent trends are going even more experienced/older. Realistically though, you simply cannot quantify how well a team/coach is going. There are simply too many variables which is why at least IMO, AFL success is almost always judged in hindsight. Lots of clubs either underperform or overperform. It also why I disagree with looking at current succseful teams and just trying to copy what they are doing. Following that logic would have led to Gellong jettisoning some expereince prior to 2022, which obviously would have been a mistake.

TLDR; You basically cannot quantify coach performance to provide a reliable future predictor.

 

[arsesmart]

The somewhat astonishing thing is you saw fit to query it when you didn't have a working knowledge of what it showed, as evidenced by your prior post. I'll wait patiently for your more predictive data. As I have previously said, happy to be shown better

Is it the one where you put forward "no coach has won a GF if they haven't made a prelim in their first five years"?

Are we just trying to game play now?

 

[The Feeder Club]

The negative value is nowhere near as high as this makes it sound. If you look back from 2014-2024, there have been 16 coaches to coach 5 years or more (allowing for backdating as long as you coached in '14, only counting stints at each club individually).

12 have made a prelim within the first 5 years, with 6 going on to win a flag (C Scott, A Clarkson, J Longmire, A Simpson, L Beveridge and S Goodwin), giving a 50% flag rate. Its important to note that Hinkley and Fagan are still ongoing but you can't analyse the future.

Only 4 have coached 5 years without a prelim in those first 5. Of them, only 1 has won a flag (Hardwick), giving a 25% flag rate. However, you can't compare these two metrics given the vast difference in dataset size. Interestingly, if you include Longmuir and Nicks in the "5 years, no prelim" crew, only Longmuir has won a final within that time. Hardwick did not win a final within his first 5 years nor did A Richardson, J Worsfold (Essendon) or S Dew.

As for better data, age and experience relative to the rest of the comp is slightly more robust although recent trends are going even more experienced/older. Realistically though, you simply cannot quantify how well a team/coach is going. There are simply too many variables which is why at least IMO, AFL success is almost always judged in hindsight. Lots of clubs either underperform or overperform. It also why I disagree with looking at current succseful teams and just trying to copy what they are doing. Following that logic would have led to Gellong jettisoning some expereince prior to 2022, which obviously would have been a mistake.

TLDR; You basically cannot quantify coach performance to provide a reliable future predictor.

I presented thirty years of data, not ten. And I said association, not causation. Still waiting for actual better figures, not words like more robust, some figures

 

[Lach72]

I want some 2010 style footy. Attacking and enjoyable footy

Is that too much to ask?

 

[arsesmart]

Average age doesn’t show the break up of players.

Emmett was the only first year player that played last year and he’s a mature ager. We weren’t playing first and second year players on mass like North, West Coast and Hawthorn - I’m sorry but we just weren’t.

Walters was the only player over 30 that played many games last year. Teams with high average ages have 6-7+. I think that not having these veterans in the side rather than actually being young is why we had a low average age tbh.

The vast majority of who played were aged between 22 and 28. That’s either just before prime or prime age. There’s no excuses for those players - None.

Average age is a robust enough stat to cover those variances in the majority of cases. It's not a forgone conclusion of course (and I'll need to revisit my exact outcomes), but it was based on about 4 seasons of games.

What you could infer is that across the four quarters of footy (then a season) having a certain level of experience and physical maturity will likely break the dam of naive enthusiasm and comparatively underdeveloped physique.

 

[arsesmart]

I presented thirty years of data, not ten. And I said association, not causation. Still waiting for actual better figures, not words like more robust, some figures

It's still not a big sample size and the confounding variables muddy the water to such an extent it renders any meaningful interpretation from it near impossible.

 

[The Feeder Club]

It's still not a big sample size and the confounding variables muddy the water to render any meaningful interpretation from it.

Still waiting for your improved data, That's how progressive statistical modelling works you know. You have a base model, and someone either refines it or presents an entirely new paradigm that's better. Sniping from the sidelines that "your model is shit" is not an advancement

 

[The Lobbster]

I presented thirty years of data, not ten. And I said association, not causation. Still waiting for actual better figures, not words like more robust, some figures

The AFL finals system only changed to its current iteration in 2000, so any data before that is basically pointless. Furthermore, I would argue an 18 team competition is also relevant which rules out anything prior to 2012.

Regardless, I ran the numbers from 2006 to 2024 (using '06 because Wikipedia stopped listing coaches on its season page beyond this point for some stupid reason).

Now there have been 19 coaches to make a prelim within their first 5 years. Of them, 9 won a flag at some point with that club, giving a flag rate of 47%

There are now 7 coaches to not make a prelim who coached at least 5 years for the same club. Hardwick is still the only premiership coach between them for a flag rate of 14%. However, again you cannot compare a size of 7 to 19, especially considering most of the added data is prior to an 18 team comp, changing how effective draft picks are. Not to mention changing equalisation policies, changes in rules being better for certain game styles and a further infinite amount of factors.

As for the better dataset, the one you've based your opinion on is not much more than a curiosity. Why 5 years, why a prelim? Basically all data for judging performance can only be very basic, (win %, PF, PA etc) and doesn't take the list into account. Some coaches do well with bad teams and some do bad with good ones. At this point all data is pretty much the same, that being not very useful. You simply cannot predict who is going to be a premiership coach.

 

[arsesmart]

Still waiting for your improved data, That's how progressive statistical modelling works you know. You have a base model, and someone either refines it or presents an entirely new paradigm that's better. Sniping from the sidelines that "your model is s**t" is not an advancement

You want me to produce a better model than one that doesn't work?

Well specify what is it that your model tells us and we can try to figure if that is accurate or if there might be another possible way (because there is always the possibility we don't have access to that data).

 

[The Feeder Club]

You want me to produce a better model than one that doesn't work?

Well specify what is it that your model tells us and we can try to figure if that is accurate or if there might be another possible way (because there is always the possibility we don't have access to that data).

I've already told you that, you can easily produce a 2x2 table of my basic data to enable you to calculate a positive and negative predictive value and a sensitivity/specificity if you like. Outcome = premiership. Exposure = any little old variable you choose to put up against my one of making a prelim in their first five years of being a coach. Or you want to be more sophisticated than running a univariate model to match my little univariate one? Than run a multivariate one. See, easy for a smart guy like you

 

[The Feeder Club]

The AFL finals system only changed to its current iteration in 2000, so any data before that is basically pointless. Furthermore, I would argue an 18 team competition is also relevant which rules out anything prior to 2012.

Regardless, I ran the numbers from 2006 to 2024 (using '06 because Wikipedia stopped listing coaches on its season page beyond this point for some stupid reason).

Now there have been 19 coaches to make a prelim within their first 5 years. Of them, 9 won a flag at some point with that club, giving a flag rate of 47%

There are now 7 coaches to not make a prelim who coached at least 5 years for the same club. Hardwick is still the only premiership coach between them for a flag rate of 14%. However, again you cannot compare a size of 7 to 19, especially considering most of the added data is prior to an 18 team comp, changing how effective draft picks are. Not to mention changing equalisation policies, changes in rules being better for certain game styles and a further infinite amount of factors.

As for the better dataset, the one you've based your opinion on is not much more than a curiosity. Why 5 years, why a prelim? Basically all data for judging performance can only be very basic, (win %, PF, PA etc) and doesn't take the list into account. Some coaches do well with bad teams and some do bad with good ones. At this point all data is pretty much the same, that being not very useful. You simply cannot predict who is going to be a premiership coach.

Basically same reply as above

 

[arsesmart]

I've already told you that, you can easily produce a 2x2 table of my basic data to enable you to calculate a positive and negative predictive value and a sensitivity/specificity if you like. Outcome = premiership. Exposure = any little old variable you choose to put up against my one of making a prelim in their first five years of being a coach. Or you want to be more sophisticated than running a univariate model to match my little univariate one? Than run a multivariate one. See, easy for a smart guy like you

So it's it just telling us that it's hard to coach for over 5 years at the one AFL club?

That it's difficult surviving a rebuild as a coach?

Or, that it is a make or break year for Longmuir?

There are more direct measures surely. Less the stat tests. More the variables measured.

For example, most simply we could just calculate the average tenure of an AFL coach. The trending age of and/or list turnover might be other measures to add and see how that impacts the tenure duration.

What does it add to the discussion though, is the question.

 

[The Feeder Club]

What does it add to the discussion though, is the question.

That's for others to judge. I've seen posts on here that suggest he is the right man to coach the club because he once leapt into the crowd clutching his jumper after kicking a goal after the siren to win us a game. So perhaps more than that?

As I said I have an opinion that he isn't a premiership coach in the making, and am trying to back that up with qualitative and quantitative arguments. Rather than take pot shots, I invite everyone with a contrary opinion to do the same. Note, making excuses for him is not a particularly great argument - tell us all why you think we have a premiership coach on our hands

 

[estibador]

I've already told you that, you can easily produce a 2x2 table of my basic data to enable you to calculate a positive and negative predictive value and a sensitivity/specificity if you like. Outcome = premiership. Exposure = any little old variable you choose to put up against my one of making a prelim in their first five years of being a coach. Or you want to be more sophisticated than running a univariate model to match my little univariate one? Than run a multivariate one. See, easy for a smart guy like you

Trigger warning please.... you're giving me PTSD flashbacks to my uni stats unit.

 

[arsesmart]

That's for others to judge. I've seen posts on here that suggest he is the right man to coach the club because he once leapt into the crowd clutching his jumper after kicking a goal after the siren to win us a game. So perhaps more than that?

As I said I have an opinion that he isn't a premiership coach in the making, and am trying to back that up with qualitative and quantitative arguments. Rather than take pot shots, I invite everyone with a contrary opinion to do the same. Note, making excuses for him is not a particularly great argument - tell us all why you think we have a premiership coach on our hands

That's fine to have that opinion of Longmuir as a coach, but that stat doesn't really speak to that.

Likewise, calling something an "excuse" doesn't address the quality of the evidence or the impact it actually has.

If I were looking for evidence of Longmuir's coaching capacity right now, I'd say 2022, where we surprised the comp isn't a bad place to start. That was a young team that went a long way. Harnessing the personnel/resources available to you is a lot of the art of coaching.

He doesn't have a lot of time to prove he's more than a one hit wonder though. The team has gotten younger. Which is what your stat speaks to. And yes his survival as coach will likely be decided on crude terms. Wins and losses in 2024.

 

[The Feeder Club]

That's fine to have that opinion of Longmuir as a coach, but that stat doesn't really speak to that.

Likewise, calling something an "excuse" doesn't address the quality of the evidence or the impact it actually has.

If I were looking for evidence of Longmuir's coaching capacity right now, I'd say 2022, where we surprised the comp isn't a bad place to start. That was a young team that went a long way. Harnessing the personnel/resources available to you is a lot of the art of coaching.

He doesn't have a lot of time to prove he's more than a one hit wonder though. The team has gotten younger. Which is what your stat speaks to. And yes his survival as coach will likely be decided on crude terms. Wins and losses in 2024.

So still no figures of your own, just another quiet snipe at mine.
I'm glad I have your permission to have my opinion though, ever so grateful for that.

 

[arsesmart]

So still no figures of your own, just another quiet snipe at mine.
I'm glad I have your permission to have my opinion though, ever so grateful for that.

What? You're quite sensitive on that thing. It wasn't quiet by the way, it was saying what it actually spoke to. If you think that's wrong tell me how.

When I've got more time and opportunity maybe I'll look up my average age/win ratio stats for you.

Further to that, you say he's not a "premiership coach" (hmmmm what's the stat for that? premierships?). Should we be more worried about that if we had a more typical profile of a premiership list?