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

[wowlace]

Do not question the Squiggle!

Yep, did squiggle predict hawthorn being this high come seasons end?

 

[Crankyhawk]

Yep, did squiggle predict hawthorn being this high come seasons end?

We are going out in straight sets

 

[Headless]

Squiggle currently has Melbourne playing three finals, all of which are decided by less than a goal.

Don't think I could cope.

 

[Final Siren]

The mathematics, as stated previously, are flawed. It has nothing to do with pushing out numbers i don't like. There are very few real indicators at the end of the home and away season that point to who will win the premiership outside of the basic eye test, but arguably the most definitive is finishing top 4. It simply doesn't make sense to completely ignore a key statistic that (outside the bulldogs) has held true for nearly 2 decades. The evidence from decades worth of data would indicate that your chance of winning the premiership is greatly, if not nearly completely reduced if you don't finish top 4.

Now, if the assertion of the data was "power rankings of the most in form teams in the AFL", sure, Melbourne would but up and around the top spot, as would Geelong, but its not. The Flagpole (as the name indicates) is a representation of who is closer to a premiership which, as previously stated, finishing top 4 is a huge factor. To ignore that seems pretty silly and the to suggest that factor is irrelevant is intellectually dishonest.

The Flagpole seems more than capable of eliminating teams flagpole height if their chance of finishing top 8 disappears (as their rating is given a 0 across the board), which would indicate ladder position does play a role. Not sure why top 4 can't have some sway on the data as well.

Yes, that's all fair enough.

I did look at it, but without a bit of sophistication, it makes Flagpole measurably worse. For example, the past two years' ratings get less accurate, because Richmond wasn't expected to make Top 4 until quite late in the season and the Bulldogs never made it. The year before that, it had already pegged the Hawks above everyone else, but would have rewarded Fremantle for locking up a Top 4 spot early, rather than calling them out as pretenders. So it threatens to turn Flagpole into a Top 4 predictor rather than a form rating.

In murky situations like this, I'd rather keep things simple and let people understand the limitations of the system, not try to account for everything and do it badly.

But I do definitely agree that there is a real benefit to finishing top 4 that Flagpole should factor in. The only reason it doesn't is I haven't found the time to do it properly.

 

[Mcveigh]

Yes, that's all fair enough.

I did look at it, but without a bit of sophistication, it makes Flagpole measurably worse. For example, the past two years' ratings get less accurate, because Richmond wasn't expected to make Top 4 until quite late in the season and the Bulldogs never made it. The year before that, it had already pegged the Hawks above everyone else, but would have rewarded Fremantle for locking up a Top 4 spot early, rather than calling them out as pretenders. So it threatens to turn Flagpole into a Top 4 predictor rather than a form rating.

In murky situations like this, I'd rather keep things simple and let people understand the limitations of the system, not try to account for everything and do it badly.

But I do definitely agree that there is a real benefit to finishing top 4 that Flagpole should factor in. The only reason it doesn't is I haven't found the time to do it properly.

Just add 4000 smidgeons to the top four teams once their top 4 is secured, pretty simple really.

 

[Dr Tigris]

Yes, that's all fair enough.

I did look at it, but without a bit of sophistication, it makes Flagpole measurably worse. For example, the past two years' ratings get less accurate, because Richmond wasn't expected to make Top 4 until quite late in the season and the Bulldogs never made it. The year before that, it had already pegged the Hawks above everyone else, but would have rewarded Fremantle for locking up a Top 4 spot early, rather than calling them out as pretenders. So it threatens to turn Flagpole into a Top 4 predictor rather than a form rating.

In murky situations like this, I'd rather keep things simple and let people understand the limitations of the system, not try to account for everything and do it badly.

But I do definitely agree that there is a real benefit to finishing top 4 that Flagpole should factor in. The only reason it doesn't is I haven't found the time to do it properly.

Simple is good.

What you're doing with the Squiggle, to me anyway, is providing analytical input to thinking about the AFL. I now know that attack wins premierships - where previously I thought defense ruled. It always a balance, but the Squiggle clearly shows that being good at defense and better at attack is a great combo. Because it is dynamic you see not just form, but teams are getting better/worse. And the flagpole is just a decent indication of likelihood of finals success. That's all. I find it really insightful. Not because it tells me the answers, but it tells me more about what is going on that I would work out myself. Grist for the mill.

And for that grist to be really good it has to be understandable. So I like simple. It enables me to adjust my thoughts fairly easily to what is going on outside the modelling world.

 

[Hobbes]

Richmond +13 v Essendon
Collingwood +14 v Port Adelaide
Geelong +44 v Fremantle
GWS +22 v Sydney
Gold Coast v Brisbane +35
St Kilda v Hawthorn +35
Carlton v Western Bulldogs +24
West Coast +14 v Melbourne
Adelaide +14 v North Melbourne

7/9.

1. Richmond 30.6
2. Geelong 19.3 (+5)
3. Collingwood 17.2 (+3)
4. Essendon 16.3
5. Melbourne 15.8 (+3)
6. GWS 15.1 (-4)
7. Hawthorn 13.9 (-4)
8. West Coast 8.2 (-3)
9. Sydney 8.1 (+2)
10. Adelaide 3.1 (-1)
11. North Melbourne -3.4 (+2)
12. Port Adelaide -4.1 (-2)
13. Brisbane -6.2 (-1)
14. St Kilda -14.8 (+1)
15. Western Bulldogs -17.8 (-1)
16. Gold Coast -31.6 (+1)
17. Fremantle -32.4 (-1)
18. Carlton -41.5

Port Adelaide v Essendon +11
Geelong +57 v Gold Coast
Richmond +48 v Western Bulldogs
Fremantle v Collingwood +41
Carlton v Adelaide +39
Sydney v Hawthorn +1
Brisbane v West Coast +1
Melbourne +6 v GWS
St Kilda v North Melbourne +11

 

[Dacksaredown]

Squiggle is still bullish on the Dees making finals. Squiggle about to look a fool.

Or not.

 

[JohnZ]

So from the Def/Att graph, is it fair to say Squiggle rates Richmond first, Geelong/Melbourne ~eq 2nd (switched numbers for Att and Def 58 and 63) and then the others except Sydney all similarly defensively who are behind the pack?

 

[Hobbes]

Port Adelaide v Essendon +11
Geelong +57 v Gold Coast
Richmond +48 v Western Bulldogs
Fremantle v Collingwood +41
Carlton v Adelaide +39
Sydney v Hawthorn +1
Brisbane v West Coast +1
Melbourne +6 v GWS
St Kilda v North Melbourne +11

Woohoo. 9/9 heading into the finals.

1. Richmond 24.3
2. Melbourne 22.8 (+3)
3. Geelong 22.7 (-1)
4. Essendon 17.0
5. Hawthorn 16.1 (+2)
6. West Coast 13.2 (+2)
7. Collingwood 12.1 (-4)
8. Adelaide 9.5 (+2)
9. GWS 9.3 (-3)
10. Sydney 6.7 (-1)
11. North Melbourne -0.0
12. Port Adelaide -4.9
13. Brisbane -10.1
14. Western Bulldogs -10.5 (+1)
15. St Kilda -13.7 (-1)
16. Fremantle -28.5 (+1)
17. Gold Coast -35.6 (-1)
18. Carlton -46.6

Plenty of movement despite picking 9/9, thanks to good results for Melbourne, West Coast and Hawthorn, and disappointingly small wins for Collingwood and Richmond.

Richmond +8 v Hawthorn
Melbourne +1 v Geelong (more like a tip of +0.1) Both teams are in great form, and one of them will be very disappointed next week.
Sydney v GWS +3
West Coast +10 v Collingwood

Looks like four good games - I don't make any team more than a 60% chance. Running a ladder predictor, I make Melbourne to be the favourite this week, then against Hawthorn, then away against West Coast. I guess this is where Melbourne fans dare to dream.

 

[Final Siren]

So from the Def/Att graph, is it fair to say Squiggle rates Richmond first, Geelong/Melbourne ~eq 2nd (switched numbers for Att and Def 58 and 63) and then the others except Sydney all similarly defensively who are behind the pack?

Yep. At the moment it's essentially:

Best: Richmond

Then: Melbourne & Geelong

Then: Collingwood, West Coast, GWS

Then: Hawthorn

Then: Sydney

 

[99cents]

Yep. At the moment it's essentially:

Best: Richmond

Then: Melbourne & Geelong

Then: Collingwood, West Coast, GWS

Then: Hawthorn

Then: Sydney

Rich 136.1%
Gee 131.6%
Mel 131.4%

 

[BEaston]

Lots of % inflation against crap teams for two of the above, which we know squiggle rates highly.

 

[Komrade_Tiger]

Lots of % inflation against crap teams for two of the above, which we know squiggle rates highly.

Only if they maintain it though, therein lies the issue. Especially with both being in knockout finals

It’s impressive but they have to back it up, Port failed to do so and the squiggle reacted harshly

 

[kickazz]

Lots of % inflation against crap teams for two of the above, which we know squiggle rates highly.

Only if they maintain it though, therein lies the issue. Especially with both being in knockout finals

It’s impressive but they have to back it up, Port failed to do so and the squiggle reacted harshly

Which two are you talking about? Melbourne had the second easiest, Geelong the third hardest and Richmond around the middle?

 

[spinynorman]

Given the squiggle so highly rates 5th and 8th, I'd be interested what it sees as the chances for which teams to make the Grand Final/win the whole shebang, if that's possible?

 

[Komrade_Tiger]

Which two are you talking about? Melbourne had the second easiest, Geelong the third hardest and Richmond around the middle?

Huh? If you’re referring to the draw that’s a different discussion to the one here.

Squiggle has shown in the past to react to floggings, but in doing so, teams have to prove it wasn’t a flash by maintaining squiggle’s new predictions, port had huge movement to the right but couldn’t maintain that and immediately started sliding left at a rate of knots again.
Ergo, to have such a high % but be in the bottom half of the 8 indicates that Squiggle has reacted but isn’t certain on them yet.

It’s a knockout final so all that % is going to count for nothing for whoever loses so let’s not blow it out of proportion, especially as Flagpole doesn’t account for ladder position, it’s a pure power ranking

 

[kickazz]

Given the squiggle so highly rates 5th and 8th, I'd be interested what it sees as the chances for which teams to make the Grand Final/win the whole shebang, if that's possible?

I've read similar analysis elsewhere that rates them higher on form and ability but the mathematics of winning the extra game gives them an overall lesser chance.

 

[The Swans Blog]

Squiggle might need adjusting if 8th bombs out tonight

 

[PHX]

Giants just hulk-smashed the Squiggle.

 

[RookiePick]

West Coast less likely to win the flag than Geelong? Squiggle is a hard taskmaster

 

[Pessimistic]

Now predicting both away teams to win. Has this ever happened?

Both Melbourne and GWS had relatively easy wins

Hawks and tigers went into preservation mode late in their game. Eagles and pies slugged it out to the last.

Let’s see shall we? Hawks had comfortable wins v Melbourne and Collingwood early in the year, but all three have improved since then

 

[evo]

`I reckon Squiggle has it right.

Only question is Melbourne v West Coast Prelim. I reckon Melbourne can do a repeat of R23 and beat the Weags again in Perth.

Richmond Dees GF would be tremendous.

 

[PerthBoy86]

`I reckon Squiggle has it right.

Only question is Melbourne v West Coast Prelim. I reckon Melbourne can do a repeat of R23 and beat the Weags again in Perth.

Richmond Dees GF would be tremendous.

Big question is can we make up for the likely dominance of Gawn and the rovers in the midfield.

 

[evo]

Big question is can we make up for the likely dominance of Gawn and the rovers in the midfield.

Your rucks actually did pretty well against Grundy. Was surprised by Vardy.