Only emerged from behind the bar, still yet to leave the pub.....allegedly
Certified Legendary Thread The Squiggle is back in 2023 (and other analytics)
- Thread starter Final Siren
- Start date
- Tagged users None
threenewpadlocks
Brownlow Medallist
Port arguably recruited better than any other team in the off season, something the Squiggle doesn't predict, and people are saying that Port's ability to beat Fremantle in Round 1 is overrated?
WhiskeyTangoFoxtrot
Norm Smith Medallist
- Oct 26, 2012
- 8,528
- 12,477
- AFL Club
- Adelaide
There pre-season form doesn't really indicate that. I expect it to be closer than 95-53.
Golumless
A bigger deal out west.
People saying that was overrated beforehand really haven't paid the slightest attention to Port in the past two years. They are already very accomplished downhill skiiers (and the most predictable side in the league in terms of results).
Does these recruits make the difference against top 8 sides is another question entirely, seeing they are a side purely reliant on "talent" to get results.
- Aug 5, 2012
- 3,310
- 5,634
- AFL Club
- Richmond
Hell yeah! 2018 squiggle! IM READY BAYBEE
Only two games in and the new squiggle is already showing its colours. Those are some pretty significant moves for results that, when you look at the final scores, weren't *that* surprising. Sure, Richmond & Carlton both scored more than expected, but those squiggles are huge. Similarly, Essendon over Adelaide is surprising, but not *that* surprising given how they did last year (perhaps I am subconsciously factoring in off-field changes too much).
Will be interested to see how close teams stick to their first round positions over the next couple of weeks, or whether we see a bit of a reversal.
Will be interested to see how close teams stick to their first round positions over the next couple of weeks, or whether we see a bit of a reversal.
Lions_Prid3
Debutant
- Jan 10, 2018
- 95
- 356
- AFL Club
- Brisbane Lions
Yeah I'll put my hand up and say I got that one wrong! Turns out Port are still flat-track bullies, and Freo are still coached by Ross Lyon...
On [device_name] using BigFooty.com mobile app
treefingers
Cancelled
- Jan 20, 2008
- 6,514
- 8,396
- AFL Club
- Hawthorn
Live squiggle at half time of the GWS-WB game, wowee
That's insane, is it because it is so early in the season? Movement is more than what usually seems to happen.
treefingers
Cancelled
- Jan 20, 2008
- 6,514
- 8,396
- AFL Club
- Hawthorn
I think Final Siren weighted early-round games higher. Agree, the movement seems to be a bit much
correct, the afl and us knew where to schedule your home final
Hawks in the eight woo hoo
- Aug 18, 2009
- 4,229
- 17,496
- AFL Club
- Richmond
- Thread starter
- #89
Heh, it's a bug because I forgot how the mid-game predictor works. It extrapolates out for a final score but doesn't touch scoring shots, which Squiggle 2.0 now considers. So when it looks at a score that was extrapolated from early in the game, it sees only a few scoring shots for the match and thinks that's a super-strong defense.
Should be fixed now.
- Aug 18, 2009
- 4,229
- 17,496
- AFL Club
- Richmond
- Thread starter
- #90
It's supposed to update every 30 minutes, but it didn't work over the weekend and I only just noticed. It's up now and will continue to be live for the rest of the season.
https://live.squiggle.com.au/
Just two features I’d love to see and one thing can’t quite understand
First (and I know the disclaimer on finals tips), I don’t understand why the GF tip seems rather low scoring, compared to the GWS/Richmond game mid season (which is admittedly at a different ground). Is there some sort of difference between the way scores are calculated in the season compared to in the Gf?
Second, it’d be cool to have squiggle 1.0 presented somewhere, so we can compare 1.0 to 2.0 and see how they go. Maybe it’s something that’s already there and I’ve missed.
Third, I’d love a different sort of sliding doors - rather changing scores from the past and/or predictions from the future to those we think are more accurate, and seeing how this would affect the squiggle and other predictions.
First (and I know the disclaimer on finals tips), I don’t understand why the GF tip seems rather low scoring, compared to the GWS/Richmond game mid season (which is admittedly at a different ground). Is there some sort of difference between the way scores are calculated in the season compared to in the Gf?
Second, it’d be cool to have squiggle 1.0 presented somewhere, so we can compare 1.0 to 2.0 and see how they go. Maybe it’s something that’s already there and I’ve missed.
Third, I’d love a different sort of sliding doors - rather changing scores from the past and/or predictions from the future to those we think are more accurate, and seeing how this would affect the squiggle and other predictions.
- Aug 18, 2009
- 4,229
- 17,496
- AFL Club
- Richmond
- Thread starter
- #92
Quick overview of R1:
- GWS very very good against the Dogs, and leap straight to #2 on the back of Squiggle's newfound early-rounds sensitivity.
- Richmond get plenty of credit for recording 38 scoring shots to 20 against Carlton. This one is interesting, because if you look at it in terms of Expected Score, both teams scored about what they should have, given from where & how they took each shot. And Expected Score is usually considered a more precise metric than scoring shots. But Carlton's shot positions are flattered by receiving seven goals from free kicks, including five from point-blank range. I doubt that that kind of free kick differential will follow either team around, so tend to think the scoring shots are a better reflection of underlying strengths in this game.
- Essendon and Hawthorn both had great weeks. So did Melbourne, despite losing.
- If you can take anything from the water-logged game in Cairns, and Squiggle did, it's that North Melbourne are spoon favourites.
Holy Rioli
Premiership Player
Why are JLT results being used?
- Aug 18, 2009
- 4,229
- 17,496
- AFL Club
- Richmond
- Thread starter
- #94
Yes, the GF tip is by the special Flagpole algorithm. I may well replace this with SQUIGGLE2 before we get much further in the season, because Flagpole hasn't exactly covered itself in glory over the last two years, and does have a couple of minor issues like those unrealistically low scores when it tips the GF a long way in advance. But I want to run some more analysis first.
There are historical v1 squiggles available here --> https://live.squiggle.com.au/v1/ <-- but they're not being auto-generated on an ongoing basis.
It's on the way!
- Aug 18, 2009
- 4,229
- 17,496
- AFL Club
- Richmond
- Thread starter
- #95
Because they're probably better than nothing. This is really just my guess, since there's not enough data to test it properly, but since JLT was more like the regular season this year, with no super-goals or weird rule trials, I thought I'd throw them in there.
It didn't wind up changing any tips, but did move the predicted margins 1.5 points closer to reality, so that's good.
Since Squiggle is in turbo mode in the early rounds, any hinkiness from JLT should be very easily washed out, too.
Lions_Prid3
Debutant
- Jan 10, 2018
- 95
- 356
- AFL Club
- Brisbane Lions
Not a tremendous amount of findings out of the first round of squiggles. The Suns movement was to be expected considering the game was played 6ft under water. The Giants regain their position with the front pack after trouncing the Bulldogs, who continually leave me scratching my head as to how they ever won a flag. Jeez 2016 was a crazy sporting year worldwide. Geelong I feel are hard done by, the squiggle did not rate the Demons at all before the start of the year, so a 3 point win leads to them regressing, but for most of us watching it felt like a battle between two top 8 teams. Outside of that it's just great to have some squiggle back in my life, love your work Max!
Suns now have almost a quarter of their nine predicted wins
I might as well use this thread for the algorithm I've been running for two or three years now.
I'll start with the initial ratings for teams. This is an end-of-2017 form rating for each team. There's one conspicuous number which it's difficult to not notice, but if they're not as good as all that then the number will come down. There's also a positive skew at the moment, meaning that the average team rating is about 3. (It should, in theory, average out as zero).
1. Richmond 42.6
2. Adelaide 18.0
3. Sydney 9.5
4. GWS 6.9
5. Port Adelaide 3.0
6. Hawthorn 2.5
7. Geelong 2.3
8. Collingwood 1.8
9. West Coast 0.43
10. St Kilda 0.40
11. Melbourne -0.8
12. Western Bulldogs -1.6
13. North Melbourne -2.4
14. Essendon -3.9
15. Brisbane -4.0
16. Carlton -5.7
17. Fremantle -6.4
18. Gold Coast -10.2
This would make the first round tips (Yeah, I know it's been and gone, but it's an algorithm so it shouldn't matter)
Richmond +48 v Carlton
Essendon v Adelaide +16
St Kilda +10 v Brisbane
Port Adelaide +22 v Fremantle
Gold Coast v North Melbourne +2
Hawthorn +1 v Collingwood
GWS +14 v Western Bulldogs
Melbourne v Geelong +3
Sydney v West Coast +3
6/9 is probably about average. Margins could use work, except for my first exact margin of the year, with Geelong.
Round 2 ratings
1. Richmond +35.5
2. Sydney +22.6 (+1)
3. GWS +19.9 (+1)
4. Adelaide +11.5 (-2)
5. Hawthorn +10.6 (+1)
6. Geelong +9.8 (+1)
7. Port Adelaide +8.7 (-2)
8. St Kilda 2.5 (+2)
9. Essendon +1.7 (+5)
10. Melbourne -0.6 (+1)
11. Carlton -1.9 (+5)
12. West Coast -4.7 (-3)
13. Collingwood -5.7 (-5)
14. North Melbourne -9.0 (-1)
15. Brisbane -10.2
16. Gold Coast -10.3 (+2)
17. Western Bulldogs -13.9 (-5)
18. Fremantle -14.8 (-1)
Round 2 predictions
Adelaide v Richmond +18
North Melbourne v St Kilda +12
Carlton +14 v Gold Coast
Collingwood v GWS +20
Brisbane v Melbourne +4
Fremantle v Essendon +7
Western Bulldogs v West Coast +1
Sydney +23 v Port Adelaide
Geelong v Hawthorn +1
3/8 so far should dispel any thoughts that I'm taking advantage of doing the tips after the event. I'll post-mortem properly when the round is finished.
I'll start with the initial ratings for teams. This is an end-of-2017 form rating for each team. There's one conspicuous number which it's difficult to not notice, but if they're not as good as all that then the number will come down. There's also a positive skew at the moment, meaning that the average team rating is about 3. (It should, in theory, average out as zero).
1. Richmond 42.6
2. Adelaide 18.0
3. Sydney 9.5
4. GWS 6.9
5. Port Adelaide 3.0
6. Hawthorn 2.5
7. Geelong 2.3
8. Collingwood 1.8
9. West Coast 0.43
10. St Kilda 0.40
11. Melbourne -0.8
12. Western Bulldogs -1.6
13. North Melbourne -2.4
14. Essendon -3.9
15. Brisbane -4.0
16. Carlton -5.7
17. Fremantle -6.4
18. Gold Coast -10.2
This would make the first round tips (Yeah, I know it's been and gone, but it's an algorithm so it shouldn't matter)
Richmond +48 v Carlton
Essendon v Adelaide +16
St Kilda +10 v Brisbane
Port Adelaide +22 v Fremantle
Gold Coast v North Melbourne +2
Hawthorn +1 v Collingwood
GWS +14 v Western Bulldogs
Melbourne v Geelong +3
Sydney v West Coast +3
6/9 is probably about average. Margins could use work, except for my first exact margin of the year, with Geelong.
Round 2 ratings
1. Richmond +35.5
2. Sydney +22.6 (+1)
3. GWS +19.9 (+1)
4. Adelaide +11.5 (-2)
5. Hawthorn +10.6 (+1)
6. Geelong +9.8 (+1)
7. Port Adelaide +8.7 (-2)
8. St Kilda 2.5 (+2)
9. Essendon +1.7 (+5)
10. Melbourne -0.6 (+1)
11. Carlton -1.9 (+5)
12. West Coast -4.7 (-3)
13. Collingwood -5.7 (-5)
14. North Melbourne -9.0 (-1)
15. Brisbane -10.2
16. Gold Coast -10.3 (+2)
17. Western Bulldogs -13.9 (-5)
18. Fremantle -14.8 (-1)
Round 2 predictions
Adelaide v Richmond +18
North Melbourne v St Kilda +12
Carlton +14 v Gold Coast
Collingwood v GWS +20
Brisbane v Melbourne +4
Fremantle v Essendon +7
Western Bulldogs v West Coast +1
Sydney +23 v Port Adelaide
Geelong v Hawthorn +1
3/8 so far should dispel any thoughts that I'm taking advantage of doing the tips after the event. I'll post-mortem properly when the round is finished.
Similar threads
