Certified Legendary Thread The Squiggle is back in 2023 (and other analytics)

[Final Siren]

I've been shifting stuff around servers lately so it might have been down for a bit. Should be back to normal now though.

 

[The Goon]

Probably crashed from all the Richmond fans going to check out that incredible sideways movement they got out of my boys...

We thank you for your sacrifice.

 

[WhiskeyTangoFoxtrot]

HERE COMES THE RED & BLUE SQUIGGLE!

Yep, the Eagles just managed to avoid the Demon dive bombing it from above!

The Hawks weren't so lucky, getting dive bombed by the Crows. Meanwhile Richmond, North and Collingwood appear to be enjoying their skiing holiday.

 

[WhiskeyTangoFoxtrot]

Squiggle now has Tiger B2B comfortably. Having said that we broke the squiggle last year and it may get broken again.

That's because you were playing the Crows last year in the GF. The Crows seem to exist just to defy the Squiggle's predictions.

 

[Final Siren]

There's a cool page here showing final ladder predictors from 9 different AFL computer models including Squiggle:

http://footymaths.blogspot.com.au/p/below-are-aggregated-end-of-season.html

Some notables:

• Richmond #1 according to all but two models
• The Crows have plunged, with most models now ranking them 5th-8th after starting the season as the most likely minor premiers
• The Hawks are top 4 according to all but two models
• The Cats are on the brink of the 8, with about half of the models putting them just inside and the others just outside
• Brisbane and Carlton to fight it out for the spoon, with Gold Coast and St Kilda rounding out the bottom 4

 

[XXRL]

There's a cool page here showing final ladder predictors from 9 different AFL computer models including Squiggle:

http://footymaths.blogspot.com.au/p/below-are-aggregated-end-of-season.html

Some notables:

• Richmond #1 according to all but two models
• The Crows have plunged, with most models now ranking them 5th-8th after starting the season as the most likely minor premiers
• The Hawks are top 4 according to all but two models
• The Cats are on the brink of the 8, with about half of the models putting them just inside and the others just outside
• Brisbane and Carlton to fight it out for the spoon, with Gold Coast and St Kilda rounding out the bottom 4

Would make for some cracking Qualifying finals in week 1:

Richmond vs Hawthorn
Sydney vs GWS

 

[HTT]

Would make for some cracking Qualifying finals in week 1:

Richmond vs Hawthorn
Sydney vs GWS

I hope not. Kind of funny we have never met in a final before.

 

[Crankyhawk]

There's a cool page here showing final ladder predictors from 9 different AFL computer models including Squiggle:

http://footymaths.blogspot.com.au/p/below-are-aggregated-end-of-season.html

Some notables:

• Richmond #1 according to all but two models
• The Crows have plunged, with most models now ranking them 5th-8th after starting the season as the most likely minor premiers
• The Hawks are top 4 according to all but two models
• The Cats are on the brink of the 8, with about half of the models putting them just inside and the others just outside
• Brisbane and Carlton to fight it out for the spoon, with Gold Coast and St Kilda rounding out the bottom 4

Does squiggle 2.0 view all scoring shots as equal or are goals still of more value than behinds (reflection better opportunity), I was surprised hawthorn didn't go up as well as to right (maybe due to freak accuracy)

 

[Pessimistic]

The gap between the leaders and the peloton seems easier to jump this year

 

[Final Siren]

Does squiggle 2.0 view all scoring shots as equal or are goals still of more value than behinds (reflection better opportunity), I was surprised hawthorn didn't go up as well as to right (maybe due to freak accuracy)

Goals are still worth more, but it's close: a goal is worth 4 points and a behind is worth 3 points. So there's a 33% premium for kicking a goal, but that's quite a comedown from the real-life 500% premium.

This ratio was chosen simply because it got the best results when tested against data from the last 5-10 years.

With a scoreline of Hawks 18.7 (115) to Melbourne 6.12 (48), it really highlights the differences!

According to...|The Hawks were...\The scores|140% better\Squiggle|55% better\Scoring shots|40% better

 

[Dr Tigris]

IMHO the Squiggle is trying to score a marathon by breaking it into 22 time periods. Runners surge and slow down, and (analogy slipping here) lose their stride and regain it. The Squiggle is the algorithm that best meets the criteria of predicting the eventual winner of the 22 game H&A marathon.

In finals some of the logic underpinning the Squiggle is broken. It's no longer a marathon, it's do or die each week. All teams are good, the differences between teams fade away. Differences in game plans, squad structures etc between individual teams becomes more important.

But I reckon the Squiggle gives us a good view of how good teams are, and in what way they are good. Simplified, but not too simple.

 

[Final Siren]

IMHO the Squiggle is trying to score a marathon by breaking it into 22 time periods. Runners surge and slow down, and (analogy slipping here) lose their stride and regain it. The Squiggle is the algorithm that best meets the criteria of predicting the eventual winner of the 22 game H&A marathon.

In finals some of the logic underpinning the Squiggle is broken. It's no longer a marathon, it's do or die each week. All teams are good, the differences between teams fade away. Differences in game plans, squad structures etc between individual teams becomes more important.

But I reckon the Squiggle gives us a good view of how good teams are, and in what way they are good. Simplified, but not too simple.

I actually think it has more to do with humans being better at tipping finals than regular games.

Squiggle's accuracy doesn't seem significantly different in finals vs H&A games. But I reckon humans have a much easier time tipping finals, because the amount of available data radically decreases. Instead of having to know a fair bit about 18 teams, you only have to know about 8 teams, then 6, then 4, then 2. And you're more likely to have seen those teams' recent form, since there aren't many finals matches to watch and they all involve two finalists.

So the big advantage of a computer model - being able to effortlessly juggle lots of data - goes away.

 

[A Swallow]

There's a cool page here showing final ladder predictors from 9 different AFL computer models including Squiggle:

http://footymaths.blogspot.com.au/p/below-are-aggregated-end-of-season.html

Some notables:

• Richmond #1 according to all but two models
• The Crows have plunged, with most models now ranking them 5th-8th after starting the season as the most likely minor premiers
• The Hawks are top 4 according to all but two models
• The Cats are on the brink of the 8, with about half of the models putting them just inside and the others just outside
• Brisbane and Carlton to fight it out for the spoon, with Gold Coast and St Kilda rounding out the bottom 4

I'd be stoked with this. North 10th or 11th in all models. Considering it's year 2 of a complete rebuild, in which we were tipped to be almost certainly wooden spooners for years to come on here, in the media and the general public. I was sure we weren't as bad as the majority were making out, but I guess it's still only been 4 rounds.

 

[Final Siren]

I'd be stoked with this. North 10th or 11th in all models. Considering it's year 2 of a complete rebuild, in which we were tipped to be almost certainly wooden spooners for years to come on here, in the media and the general public. I was sure we weren't as bad as the majority were making out, but I guess it's still only been 4 rounds.

Four rounds can be enough to have a fair idea of what's going on! After Round 4 last year, the ladder looked like this:

1. Geelong
2. Adelaide
3. Richmond
4. GWS
5. West Coast
6. Western Bulldogs
7. Port Adelaide
8. Melbourne

That's the same top 4 as the end of the year, almost in the same order (Geelong and Adelaide are switched).

West Coast and Port made finals, and while the Bulldogs and Melbourne didn't, they didn't miss by much: the Dogs were 1 game and percentage out, and the Demons, famously, missed by half a percentage point.

North are currently 7th with a percentage of 134%, and while they've certainly had a gentle run with the fixture so far - they may be yet to play a finals-bound opponent - they don't look much like a bottom team. Their ladder position has been earned off the back of two monsterings, and you can even argue that the conditions against Gold Coast in Round 1 didn't give them much of a chance to play their natural game.

There's good reason to think the Roos will fall down the ladder once they come up against real opposition, but it no longer seems likely that they'll fall very far.

 

[Happy_Hawker]

Four rounds can be enough to have a fair idea of what's going on! After Round 4 last year, the ladder looked like this:

1. Geelong
2. Adelaide
3. Richmond
4. GWS
5. West Coast
6. Western Bulldogs
7. Port Adelaide
8. Melbourne

That's the same top 4 as the end of the year, almost in the same order (Geelong and Adelaide are switched).

West Coast and Port made finals, and while the Bulldogs and Melbourne didn't, they didn't miss by much: the Dogs were 1 game and percentage out, and the Demons, famously, missed by half a percentage point.

North are currently 7th with a percentage of 134%, and while they've certainly had a gentle run with the fixture so far - they may be yet to play a finals-bound opponent - they don't look much like a bottom team. Their ladder position has been earned off the back of two monsterings, and you can even argue that the conditions against Gold Coast in Round 1 didn't give them much of a chance to play their natural game.

There's good reason to think the Roos will fall down the ladder once they come up against real opposition, but it no longer seems likely that they'll fall very far.

So what you're saying is predicted top 4 after round 4 are certain to finish there...

*Looks at prediction*

Good, about time Hawks finished top 4!

 

[Magpie Monopoly]

Going by the week to week projection, Pies are favourite in 14 of the remaining games, with the Cats game being 50/50.

16 wins, where do I sign?

 

[Pessimistic]

I’m a squiggle believer, but spot the clear error in this projection

 

[Underarm]

I’m a squiggle believer, but spot the clear error in this projection

View attachment 485974

Giants playing finals at the SCG? Collingwood getting within 3 goals of a grand final? Port Adelaide winning a showdown?

 

[brabs]

I’m a squiggle believer, but spot the clear error in this projection

View attachment 485974

A game where Adelaïde didn’t turn up, catapults the pies to the prelim.
The only other game they won this year was Carlton?

 

[Isaac Cumming No 1]

I’m a squiggle believer, but spot the clear error in this projection

View attachment 485974

My blood just boiled at the thought of us playing a QF at the SCG. No way no how. Spotless or a shock to most, Manuka

 

[roacheee]

a tiges v pies prelim!!!!!

when was the last 100k attendance

 

[Foley#41]

a tiges v pies prelim!!!!!

when was the last 100k attendance

30/9/17

 

[kaiserchief13]

After round 6 isn't it that the top 8 only changes with 1 or 2 teams?

 

[Hobbes]

There's a cool page here showing final ladder predictors from 9 different AFL computer models including Squiggle:

http://footymaths.blogspot.com.au/p/below-are-aggregated-end-of-season.html

Some notables:

• Richmond #1 according to all but two models
• The Crows have plunged, with most models now ranking them 5th-8th after starting the season as the most likely minor premiers
• The Hawks are top 4 according to all but two models
• The Cats are on the brink of the 8, with about half of the models putting them just inside and the others just outside
• Brisbane and Carlton to fight it out for the spoon, with Gold Coast and St Kilda rounding out the bottom 4

I know my model doesn't take this into account either, but Adelaide has a lot of major outs. For example, this week they were missing Sloane, Betts, Smith and two Crouches (and a few more). I suspect that reports of their demise have been exaggerated.

 

[A Swallow]

North having another solid wander diagonally up & right. Drastic change over the course of the first 5 rounds.