Certified Legendary Thread Race for the flag, in squiggly lines

[d33my]

1996 was looking like a missed opportunity at one point during 2013.

Yeah, we really had a chance to drag those Hawthorn bastards down.

 

[Final Siren]

Do you have a HGA rating like Massey Ratings does? Or do you just flat it out as +12 no matter what?

It's always a simple +12 for games between teams from different states (+6 for the home team and -6 for the away team), and +0 otherwise. See squiggle -> Info -> The Model.

 

[old55]

Final Siren, what do you make of the "defence wins flags" theory. 8 premiership teams rating >=75 in attack vs 2 premiership teams rating >=75 in defence. Is this a trend or maybe a calibration error between defence and attack?

 

[Final Siren]

Final Siren, what do you make of the "defence wins flags" theory. 8 premiership teams rating >=75 in attack vs 2 premiership teams rating >=75 in defence. Is this a trend or maybe a calibration error between defence and attack?

There are so few flags (only one a year, in fact), it's impossible to conclude with certainty. You can always think of a reason why year X shouldn't count. But if "defence wins flags" means "When comparing teams of equal overall strength, the more defensive one is more likely to win the flag," there's nowhere near enough evidence to show this, and there's more evidence that the opposite is true.

There have been some great posts on this question earlier in the thread, including someone who made awesome graphs. But basically the last 15 years have given us a lot of very good defensive-first Grand Finalists who haven't won the flag, e.g. Sydney 2014, Fremantle 2013, Collingwood 2011, St Kilda 2009. The only years that support the "defence wins flags" theory are 2012, when Sydney defeated Hawthorn (a very attack-first team) despite being quite evenly matched overall, and 2005, which looks more like a series of minor miracles than regular finals.

There just seems to be this graveyard in the right-hand part of the chart where good teams go to not win flags. There's no equivalent on the attacking side: no teams that sit out to the left of the premiership cup cluster and make the Grand Final and lose.

So it's "better teams win flags," regardless of whether they're more attacking or defensive. If I have to choose, although I can't be sure, I'd rather be attacking.

In fact, here's a table:

Year|Premier|Squiggle Best Overall|Squiggle Best Attack|Squiggle Best Defence\2014| Hawthorn | Hawthorn | Hawthorn |Sydney\2013| Hawthorn | Hawthorn | Hawthorn |Fremantle\2012| Sydney | Sydney |Hawthorn| Sydney \2011| Geelong | Geelong | Geelong |Collingwood\2010| Collingwood | Collingwood | Collingwood | Collingwood \2009| Geelong |Adelaide| Geelong |St Kilda\2008| Hawthorn |Geelong|Geelong|Geelong\2007| Geelong | Geelong | Geelong | Geelong \2006| West Coast | West Coast | West Coast |Sydney\2005| Sydney |Adelaide|Western Bulldogs|Adelaide\2004| Port Adelaide | Port Adelaide |Brisbane|Sydney\2003| Brisbane | Brisbane | Brisbane |Collingwood\2002| Brisbane | Brisbane | Brisbane | Brisbane \2001| Brisbane | Brisbane | Brisbane |Carlton\2000| Essendon | Essendon | Essendon | Essendon

That's 11 years out of 15 where the premier had the best attack, compared to only 5/15 where they had the best defence.

 

[bomberclifford]

That's 11 years out of 15 where the premier had the best attack, compared to only 5/15 where they had the best defence.

... and 3 of those also had the best attack. Seems a pretty comprehensive stat to me.

Attacking teams win flags.

 

[Janus]

There are so few flags (only one a year, in fact), it's impossible to conclude with certainty. You can always think of a reason why year X shouldn't count. But if "defence wins flags" means "When comparing teams of equal overall strength, the more defensive one is more likely to win the flag," there's nowhere near enough evidence to show this, and there's more evidence that the opposite is true.

There have been some great posts on this question earlier in the thread, including someone who made awesome graphs. But basically the last 15 years have given us a lot of very good defensive-first Grand Finalists who haven't won the flag, e.g. Sydney 2014, Fremantle 2013, Collingwood 2011, St Kilda 2009. The only years that support the "defence wins flags" theory are 2012, when Sydney defeated Hawthorn (a very attack-first team) despite being quite evenly matched overall, and 2005, which looks more like a series of minor miracles than regular finals.

There just seems to be this graveyard in the right-hand part of the chart where good teams go to not win flags. There's no equivalent on the attacking side: no teams that sit out to the left of the premiership cup cluster and make the Grand Final and lose.

So it's "better teams win flags," regardless of whether they're more attacking or defensive. If I have to choose, although I can't be sure, I'd rather be attacking.

In fact, here's a table:

Year|Premier|Squiggle Best Overall|Squiggle Best Attack|Squiggle Best Defence\2014| Hawthorn | Hawthorn | Hawthorn |Sydney\2013| Hawthorn | Hawthorn | Hawthorn |Fremantle\2012| Sydney | Sydney |Hawthorn| Sydney \2011| Geelong | Geelong | Geelong |Collingwood\2010| Collingwood | Collingwood | Collingwood | Collingwood \2009| Geelong |Adelaide| Geelong |St Kilda\2008| Hawthorn |Geelong|Geelong|Geelong\2007| Geelong | Geelong | Geelong | Geelong \2006| West Coast | West Coast | West Coast |Sydney\2005| Sydney |Adelaide|Western Bulldogs|Adelaide\2004| Port Adelaide | Port Adelaide |Brisbane|Sydney\2003| Brisbane | Brisbane | Brisbane |Collingwood\2002| Brisbane | Brisbane | Brisbane | Brisbane \2001| Brisbane | Brisbane | Brisbane |Carlton\2000| Essendon | Essendon | Essendon | Essendon

That's 11 years out of 15 where the premier had the best attack, compared to only 5/15 where they had the best defence.

So if thats the case, why is it that if you move Port to Geelong 2011 territory the squiggle still has them losing to Sydney in the predictor?

 

[lamaros]

If you consider 2005 and 2008 to be mistakes, then the flag has been won by the team with the best overall combination of attack at defence every time but once.

That other one time the best attack beat the best overall team was with Geelong beating Adelaide. In a year were neither was the best defensively.

Every time a team was the best overall and either the best defensively or attacking, they have won. Outside the two very obvious outliers.

From those results all you can really say is the best overall side normally wins, but that the best overall side is more often the best attacking side and not necessarily the best defensive one.

Which is to say that ultra defensive sides often seem to not be the best sides going around, not that they are the best and just can't win.

Also, Hawthorn in 2008 was the only time that the side that was the best overall, attacking and defensive has not won the GF.

And I'm pretty sure if that GF was played 10 times Geelong would have won 9 of them

 

[lamaros]

Or tl;dr

The best side normally wins. The best defensive sides just aren't normally the best sides.

 

[Rotayjay]

According to my taste, it's a good thing to hear that sides with the best attack have won the flag much more often than the side with the best defence. I suspected that this was true. You can shut down other sides all year and make it to the grand final, but you often can't shut down the best attack sufficiently when you come up against it at the big dance

 

[cryptor]

The game fundamentally requires you to score to win. Hypothetically if the perfect attack met the perfect defense in a GF then the result would be a series of replayed GFs that are all draws. Though with a rule that allows defenses to concede a behind to prevent a goal (which is still considered a good defensive option) there is in practice probably a slight advantage there to attacking sides.

 

[Simon_Nesbit]

...

Very nice table - interesting Sydney 05 and Hawthorn 08 weren't the best overall, best attacking, or best defensive sides.

Could you do a table of the top squiggle ratings in the same time frame? Who was the 'unluckiest' side? When did a club 'peak'?

 

[QuietB]

What happened to the squiggle at quarter time in Adelaide on Saturday night?

 

[Rotayjay]

Very nice table - interesting Sydney 05 and Hawthorn 08 weren't the best overall, best attacking, or best defensive sides.

Could you do a table of the top squiggle ratings in the same time frame? Who was the 'unluckiest' side? When did a club 'peak'?

The beauty of our game and the grand final is that the best team on the day wins, unlike in the EPL when the last few weeks of the season can be pure junk time waiting for Manchester United or whatever to claim the championship for the millionth time.

Your mob were better and more committed that day in 2008, and we won two flags finishing 4th and 5th respectively.

 

[Elder van Berlo]

In fact, here's a table:

Year|Premier|Squiggle Best Overall|Squiggle Best Attack|Squiggle Best Defence\2009| Geelong |Adelaide| Geelong |St Kilda

Right. In front. Of me.

 

[Borderline]

So how accurate did you say this was Final Siren?

 

[AlibiMonday]

There are so few flags (only one a year, in fact), it's impossible to conclude with certainty. You can always think of a reason why year X shouldn't count. But if "defence wins flags" means "When comparing teams of equal overall strength, the more defensive one is more likely to win the flag," there's nowhere near enough evidence to show this, and there's more evidence that the opposite is true.

There have been some great posts on this question earlier in the thread, including someone who made awesome graphs. But basically the last 15 years have given us a lot of very good defensive-first Grand Finalists who haven't won the flag, e.g. Sydney 2014, Fremantle 2013, Collingwood 2011, St Kilda 2009. The only years that support the "defence wins flags" theory are 2012, when Sydney defeated Hawthorn (a very attack-first team) despite being quite evenly matched overall, and 2005, which looks more like a series of minor miracles than regular finals.

There just seems to be this graveyard in the right-hand part of the chart where good teams go to not win flags. There's no equivalent on the attacking side: no teams that sit out to the left of the premiership cup cluster and make the Grand Final and lose.

So it's "better teams win flags," regardless of whether they're more attacking or defensive. If I have to choose, although I can't be sure, I'd rather be attacking.

In fact, here's a table:

Year|Premier|Squiggle Best Overall|Squiggle Best Attack|Squiggle Best Defence\2014| Hawthorn | Hawthorn | Hawthorn |Sydney\2013| Hawthorn | Hawthorn | Hawthorn |Fremantle\2012| Sydney | Sydney |Hawthorn| Sydney \2011| Geelong | Geelong | Geelong |Collingwood\2010| Collingwood | Collingwood | Collingwood | Collingwood \2009| Geelong |Adelaide| Geelong |St Kilda\2008| Hawthorn |Geelong|Geelong|Geelong\2007| Geelong | Geelong | Geelong | Geelong \2006| West Coast | West Coast | West Coast |Sydney\2005| Sydney |Adelaide|Western Bulldogs|Adelaide\2004| Port Adelaide | Port Adelaide |Brisbane|Sydney\2003| Brisbane | Brisbane | Brisbane |Collingwood\2002| Brisbane | Brisbane | Brisbane | Brisbane \2001| Brisbane | Brisbane | Brisbane |Carlton\2000| Essendon | Essendon | Essendon | Essendon

That's 11 years out of 15 where the premier had the best attack, compared to only 5/15 where they had the best defence.

I gather the best(s) also includes the result from the GF which is a little self-reinforcing. What would be marginally more interesting IMO would be this table before the GF is played.

 

[Final Siren]

So how accurate did you say this was Final Siren?

Looks legit

 

[Final Siren]

I gather the best(s) also includes the result from the GF which is a little self-reinforcing. What would be marginally more interesting IMO would be this table before the GF is played.

Well, yes, if we wanted to test prediction. But for the purposes of analysis, we're testing the proposition "Defence wins flags," not "Defence in games leading up to but not including the Grand Final wins flags."

 

[Final Siren]

Very nice table - interesting Sydney 05 and Hawthorn 08 weren't the best overall, best attacking, or best defensive sides.

Could you do a table of the top squiggle ratings in the same time frame? Who was the 'unluckiest' side? When did a club 'peak'?

Very subjective, and as Rotayjay says, in AFL there are no lucky teams, only premiers. But a few notable things from recent years...

In 2013, three teams finished ranked almost exactly evenly: Hawthorn (high attack), Fremantle (high defence), and Geelong (balanced). Very unusual.

Collingwood 2011 is the best team not to have won the flag this century. If the Pies had won, their premiership cup would rank better than every other except Geelong 2011. Although this is somewhat biased by the squiggle's love of keeping an opposition to a low score, which Collingwood did several times to great effect (West Coast 53, Fremantle 52, St Kilda 51, Melbourne 41, North Melbourne 30, and especially Port Adelaide 21 in Adelaide. Fun fact: In six out of eight quarters of football against the Pies in 2011, North Melbourne -- who won 10 games -- could score no more than one goal.) We tend to mark the Pies down for their fairly uninspiring end to the year; they were clearly sliding down from their peak. But still. They lost to one team all year.

Geelong 2009 was only the squiggle's third ranked team, after Adelaide and St Kilda (two teams that wound up in the defensive graveyard). It's also the worst ranked cup of the last 20 years.

Notwithstanding Hawthorn's excellent end to 2008, it's stupefying that Geelong lost to them. Geelong 2008 could play Collingwood 2011 for the biggest loser's poisoned chalice.

The gap between Geelong 2007 and everyone else is so large, they could have called off the season early and just given them the cup. The reason Geelong 2007 isn't rated a lot higher on the squiggle is they had no real opposition: no other teams could rise above the pack to provide a challenge.

Sydney 2005 looks incredibly anomalous and it's amazing they even made the Grand Final, let alone won it. Adelaide were ranked better, but in the defensive graveyard again; St Kilda were in a great spot.

All the other years from 2000 look about right.

 

[Bender571]

I gather the best(s) also includes the result from the GF which is a little self-reinforcing. What would be marginally more interesting IMO would be this table before the GF is played.

Completely off topic post, but are you still doing your output rankings of the 2008 draftees?

 

[B4Bear]

It would be interesting to see a version of 'THE SQUIGGLE" where every team starts from the same position.
If you removed last year's influence it would show how each team is performing relative to each other from round 1. Because it is pretty obvious that the results from last July for example, are having way to much impact on some teams, such as Melbourne and the Bulldogs.

Anyway. Food for thought.

 

[SJ]

It would be interesting to see a version of 'THE SQUIGGLE" where every team starts from the same position.
If you removed last year's influence it would show how each team is performing relative to each other from round 1. Because it is pretty obvious that the results from last July for example, are having way to much impact on some teams, such as Melbourne and the Bulldogs.

Anyway. Food for thought.

But if everyone started in the same spot, it would mean that a given win over St Kilda moves a team just as much as a the same win scoreline over Hawthorn.

 

[Final Siren]

So if thats the case, why is it that if you move Port to Geelong 2011 territory the squiggle still has them losing to Sydney in the predictor?

You found a bug! Thanks for reporting this.

What went wrong: When you moved a team, the predictor was starting it there, then playing through 2015's actual matches, then predicting the rest. So Port would fall back quite a long way from where you put them because of Rounds 1-4 before the prediction happened. Now fixed!

 

[Final Siren]

Public Squiggle Announcement: I've updated the ladder predictor to use a different algorithm for the Grand Final, since the default isn't very good at tipping those, and it's both wrong and confusing to assign home ground advantage there.

Grand Finals are now tipped by OFFDEF-75. This ignores home ground advantage and is more sensitive to recent form. It has a pretty good record at tipping Grand Finals over the last couple of decades, despite getting 2014 wrong.

It's still somewhat misleading, since the reason for OFFDEF-75's good record is that for a Grand Final, recent form is finals form, and that's not the case here, when we're using it to tip a winner five months out. It will probably jump around depending on who's had a good week. But that's better than assigning home ground advantage in a Grand Final, and more fun besides.

 

[Clems Knee]

I'm loving the Squiggle today. First time since just before the St. Kilda game that the squiggle has predicted a Fremantle grand final victory.