Certified Legendary Thread Race for the flag, in squiggly lines

[Final Siren]

Round 20, 2016

"Fat Squiggle"

Animated!

The winner of this week's edition of WHO WANTS TO DROP A CRUCIAL GAME? is Hawthorn. Congratulations! You have won 3 weeks of gnawing uncertainty.

In fact let's skip ahead to the ladder predictor, since that's what's most interesting at the moment:

But this disguises the fact that the ladder is pretty well bedded down except for total chaos in the top 5. As better illustrated here:

Basically we get to play WHO WANTS TO DROP A CRUCIAL GAME? for the rest of the season, where all games involving the top 5 are crucial ones. If you drop one, you miss Top 4. Unless someone else also drops one.

And since those teams don't play each other, we're looking for upsets! It's tempting to think nobody will lose one, since the top 5 will be favourites against lower-ranked teams, and therefore the ladder should stay the same. But that's very unlikely. With 15 games, even if every top-5 team has a 95% chance of winning each individual game, it's more likely than not that someone will drop one.

This is why Geelong are predicted to finish so high, even though they're currently 5th.

But it also means it's near-impossible to predict. The top 4 will be whoever doesn't lose games they shouldn't.

In other news, Brisbane and Sydney both exited the chart, in opposite directions, so I bit the bullet and expanded the axis. So now we have a fat squiggle chart.

The squiggle tipped Adelaide by 94 points and was still well short, which pushed the Crows further into the ideal zone. Sydney had an excellent week, too, thrashing Port, even though they still haven't demonstrated a lot of scoring power.

Flagpole! Sydney shot up into 2nd place, which is especially impressive given that this algorithm cares a lot about scoring, which the Swans don't really do.

Live squiggling! Apologies for the delay in live squiggle updates over the weekend. By some strange coincidence the Fox Footy website stopped displaying live scores as well.

 

[nobbyiscool]

It's a brave squiggle that is predicting West Coast to beat Hawthorn!

I dunno what the opposite of a "king maker" is, especially in the sporting context - but as a supporter, I'm going to enjoy watching my teams role as the "anti-king maker" over the next 3 weeks.

Can I ask about the flagpole? Why are there 10 teams who are still being given some (even a miniscule) premiership chance? Shouldn't those 10 teams all be equally at rock bottom of the flagpole?

 

[Mningasimisonda]

There is one team sitting very much in the premiership zone and apart from their weakness against Geelong, they are the only team that really looks like a premiership team at the moment.

 

[Biggy_Boy]

There is one team sitting very much in the premiership zone and apart from their weekness against Geelong, they are the only team that really looks like a premiership team at the moment.

They certainly look good against some teams. Then they play against Geelong, and it seems like a lot of their weaknesses get exposed. And if Geelong can do it, I imagine a Sydney or Hawthorn could do so in the finals as well. Their lack of midfield depth could be a problem against those teams for one.

 

[kreglze]

Suns still moving in the right direction even with a loss. That is what I want to see. Except to see the Suns win of course...

 

[Bont2Bruce]

The total number of wins in your simulation (197) doesn't equal the total number of matches in the season (198). With no draws forecast, where did the extra win go?

 

[Bloody]

The total number of wins in your simulation (197) doesn't equal the total number of matches in the season (198). With no draws forecast, where did the extra win go?

Sent from my iPhone using Tapatalk

 

[Bloody]

Wasn't sure if Sydney moving or graph expanding on first view of animated squiggle


Sent from my iPhone using Tapatalk

 

[Roobs321]

Tower of Power is making its name via the top 5. To have five reasonably equal chances at the minor premiership with only a few games left is Gillon's wet dream.

West Coast play three of them, which makes it even more fun. I hope to god we haven't won our last game of 2016, and if we win just one more game, I'd have to take that!

 

[Hobbes]

It's a brave squiggle that is predicting West Coast to beat Hawthorn!

There's not much to go on for a form guide - West Coast haven't really been playing good teams at home this year - Brisbane, Fremantlex2, Richmond, Collingwood, St Kilda, Gold Coast, Essendon, Melbourne. That's pretty much a who's who of not playing finals. The only outliers are a horrifically out of form North and the surprise loss to Adelaide. West Coast's away form may be dodgy, but their home form has been pretty much exemplary - not just winning matches, but putting teams away by big margins.

 

[ok.crows]

The total number of wins in your simulation (197) doesn't equal the total number of matches in the season (198). With no draws forecast, where did the extra win go?

As was pointed out the squiggle forecast shows a probabilistic ladder which tries to account for the possibility of upsets.

If you fill out the ladder predictor on the AFL website for the last three rounds using the remaining home & away tips from the squiggle (i.e. you assume the squiggle's forecast tips are going to be 100% accurate from this point onward) then you end up with the following ladder:

1. Swans - 68 points - 142.7%
2. Adelaide - 68 points - 140.6%
3. GWS - 68 points - 138.7%
4. Geelong - 68 points - 136.7%
5. Hawks - 68 points - 118.2%
6. Bulldogs - 64 points - 118.2%
7. Eagles - 56 points - 123.9%
8. North - 48 points - 108.4%
9. Saints - 48 points - 98.7%

 

[Bont2Bruce]

As was pointed out the squiggle forecast shows a probabilistic ladder which tries to account for the possibility of upsets.

Why not zero sum the total wins though? The rounding clearly isn't working properly.

The model generated end of season ladder cannot possibly be the final outcome after Round 23. At least get the model to generate an answer which is a possibility.

 

[Hobbes]

As was pointed out the squiggle forecast shows a probabilistic ladder which tries to account for the possibility of upsets.

If you fill out the ladder predictor on the AFL website for the last three rounds using the remaining home & away tips from the squiggle (i.e. you assume the squiggle's forecast tips are going to be 100% accurate from this point onward) then you end up with the following ladder:

1. Swans - 68 points - 142.7%
2. Adelaide - 68 points - 140.6%
3. GWS - 68 points - 138.7%
4. Geelong - 68 points - 136.7%
5. Hawks - 68 points - 118.2%
6. Bulldogs - 64 points - 118.2%
7. Eagles - 56 points - 123.9%
8. North - 48 points - 108.4%
9. Saints - 48 points - 98.7%

Much as I'd like to see North miss out (if only to laugh at the people that were gloating when they were 9-0), it's really not likely. It requires the perfect storm of North losing 3/3 (Hawthorn, Sydney, GWS - very plausible, but they do get three chances to avoid that fate) and St Kilda winning 3/3 (Sydney, Richmond and Brisbane - obviously, Sydney is the doozy) or, less likely, Port winning 3/3 and making up percentage (Melbourne, Adelaide and Gold Coast). Parlays which require six specific results are unlikely to happen, even if most of the constituent results are probable.

 

[Matele]

There's not much to go on for a form guide - West Coast haven't really been playing good teams at home this year - Brisbane, Fremantlex2, Richmond, Collingwood, St Kilda, Gold Coast, Essendon, Melbourne. That's pretty much a who's who of not playing finals. The only outliers are a horrifically out of form North and the surprise loss to Adelaide. West Coast's away form may be dodgy, but their home form has been pretty much exemplary - not just winning matches, but putting teams away by big margins.

The only thing I don't like about this point is it's contrast to last year. People were saying we couldn't take WC home form as our home ground wasn't were the business end of finals is played, thus it didn't mean much to be playing well there, even against good teams (think home advantage, umpiring, crowd effect etc). We get a bulk of the good teams away this season, and people still can't take the home record seriously, as it's against lesser opponents.

 

[chazwazza]

Stephen Hawking is currently going through all Squiggle algorithms trying to explain just how Sydney and Brisbane have stretched the fabric of time.

 

[The Gong]

Stephen Hawking is currently going through all Squiggle algorithms trying to explain just how Sydney and Brisbane have stretched the fabric of time.

Dark matter. It's a northern thing.

 

[GreyCrow]

Tower of Power is making its name via the top 5. To have five reasonably equal chances at the minor premiership with only a few games left is Gillon's wet dream.

West Coast play three of them, which makes it even more fun. I hope to god we haven't won our last game of 2016, and if we win just one more game, I'd have to take that!

Against Hawthorn would be nice thanks

 

[cmndstab]

Why not zero sum the total wins though? The rounding clearly isn't working properly.

Because that's not what it's designed to do. If there are three teams tipped to win 15.333 games arbitrarily giving one of them the carryover points would be stupid.

 

[Sid Ruckle]

What's the history say for a side that is as far right on squiggle as Sydney is currently?

 

[Yakker]

What's the history say for a side that is as far right on squiggle as Sydney is currently?

They won the flag in 2005 from closer to the left than they are this year. You would have to compare all sides for 2005 vs this year to get a handle on their chances I reckon

 

[Hobbes]

What's the history say for a side that is as far right on squiggle as Sydney is currently?

Adelaide in 2005 hit 84.
Collingwood in 2011 went over 95!!
Brisbane in 1996 hit 86.
West Coast in 1994 went over 85.
Collingwood in 1990 reached about 87.
Essendon in 1989 reached about 92.
Collingwood in 1988 reached about 89.

That's all there has been in my lifetime. Many years don't see anybody go past 80.

Fitzroy achieved a rating of 419 in 1899 when they held Melbourne to a score of 2. I still think that the squiggle overreacts to very low scores, which will often be largely a result of conditions rather than defensive excellence.

Adelaide in 2005 bombed out in the prelims.
Collingwood in 2011 lost the Grand Final.
West Coast in 1994 won the Grand Final.
Collingwood in 1990 won the Grand Final
Essendon in 1989 lost the prelim.
Collingwood in 1988 finished the minor round in 2nd but bombed out in straight sets.

 

[GreyCrow]

Sydney have created their own Squiggle Zone

 

[Sid Ruckle]

Adelaide in 2005 bombed out in the prelims.
Collingwood in 2011 lost the Grand Final.
West Coast in 1994 won the Grand Final.
Collingwood in 1990 won the Grand Final
Essendon in 1989 lost the prelim.
Collingwood in 1988 finished the minor round in 2nd but bombed out in straight sets.

5/6 (83.33%) - prelim or gf
3/6 (50%) - gf
2/6 (33.33%) - gf winner

History looks alright for the far right according to squiggle then.

 

[Gavin Excell]

Sydney is in a National Socialism level of far right at the moment

 

[Gavin Excell]

There is one team sitting very much in the premiership zone and apart from their weakness against Geelong, they are the only team that really looks like a premiership team at the moment.

Adelaide are entrenched in the zone arent they. Abundantly clear that they are the team to beat