Certified Legendary Thread Race for the flag, in squiggly lines

[Juke]

http://maxbarry.com/squiggle/2015.html?door=379973356&luck=13&year=2015

Now this is podracing.

 

[cryptor]

I know you don't condone gambling at all Final Siren, but I was wondering if you have ever determined if you could actually make money via betting on the squiggle. And if so, how much you could make?
Purely for curiosity, but I think it would be cool to see

Pretty sure that the squiggle tips are only marginally better than simply tipping the home team in every game. Which means it's likely not that reliable as a betting aid. Though I have found the margins can be pretty good in games that are expected to be close.

Remembering of course the squiggle was/is primarily a tool to help you visualise teams form over a long period of time rather than being an infallible prediction machine.

 

[Shane Hird]

I found one with 30% luck! That's almost credible. It was the sole Melbourne flag in 25,000 simulations.

A series of unlikely coincidences conspire to hand the Demons the flag:

• Melbourne squeak into 8th spot with 11 wins and a terrible percentage.
• They face bunny sides Geelong and Richmond in the first two weeks of the finals and win by 2 and 17 points respectively.
  
• A 1-pt win after extra time in a horrendous Roos prelim slogfest, defeating Port Adelaide 30-29. Maybe it was raining.
• Hawthorn have the mother of all off days for the GF (maybe they celebrated too early), going down to Melbourne by three goals.

And what a grand final that was.

One for the ages..

 

[Final Siren]

I know you don't condone gambling at all Final Siren, but I was wondering if you have ever determined if you could actually make money via betting on the squiggle. And if so, how much you could make?
Purely for curiosity, but I think it would be cool to see

You definitely wouldn't make money by betting on every match with every squiggle tip! It's too crude for that. It would lose a few games where it didn't know about injuries and then the bookies' take is, what, 5%? Even I would take up punting if I had an objective algorithm that could beat the markets +5% betting on every game.

If it would work anywhere, it would be situations where the squiggle's opinion differs quite a lot from the market, and then you could use your human brain to decide whether that's due to a failing of the squiggle or mass hysteria in footy followers. Like in 2015:

• Hawks are gone; they're 4-4 and can't win close games
• West Coast are flat track bullies; will struggle to make finals
• Collingwood are a lock for finals; they're top 4 halfway through the season
• Fremantle are a top team

Only that last one turned into reality, and I still say they got lucky. So maybe there, if anywhere, there is scope for betting. But I dunno. Even then you have to time it right. Beating bookies is hard.

 

[Final Siren]

Pretty sure that the squiggle tips are only marginally better than simply tipping the home team in every game. Which means it's likely not that reliable as a betting aid. Though I have found the margins can be pretty good in games that are expected to be close.

Remembering of course the squiggle was/is primarily a tool to help you visualise teams form over a long period of time rather than being an infallible prediction machine.

They're a lot better than tipping the home team: you'd have had 111 tips (53.9%) from the home team in 2015 and 146 tips (70.9%) from the squiggle. But they're about the same as tipping the favourite in every game.

 

[cryptor]

They're a lot better than tipping the home team: you'd have had 111 tips (53.9%) from the home team in 2015 and 146 tips (70.9%) from the squiggle. But they're about the same as tipping the favourite in every game.

My mistake. I was going from memory. Meant to reference that tipping the favourite not home team stat.

 

[WCE_phil]

You definitely wouldn't make money by betting on every match with every squiggle tip! It's too crude for that. It would lose a few games where it didn't know about injuries and then the bookies' take is, what, 5%? Even I would take up punting if I had an objective algorithm that could beat the markets +5% betting on every game.

TBH mate if you had an objective algorithm that could beat the combined markets by 5% overall against all games you wouldnt need to bet. They'd but it off you for enough money so that you'd never need to work again.

 

[Simon_Nesbit]

TBH mate if you had an objective algorithm that could beat the combined markets by 5% overall against all games you wouldnt need to bet. They'd but it off you for enough money so that you'd never need to work again.

Or you could just make a hundred million, and set up a Museum for Old and New Art.

 

[Randomizor]

TBH mate if you had an objective algorithm that could beat the combined markets by 5% overall against all games you wouldnt need to bet. They'd but it off you for enough money so that you'd never need to work again.

Nah they'd all just ban or restrict your account

 

[cryptor]

Nah they'd all just ban or restrict your account

They'd buy it. Would allow them to set more accurate markets which would be tougher to beat.

 

[Randomizor]

They'd buy it. Would allow them to set more accurate markets which would be tougher to beat.

Cheaper to just follow the exchanges?

 

[The_Reaper]

Nah they'd all just ban or restrict your account

No because they'd want your algorithm so they could use it and also to prevent you getting mad and releasing it to the public

 

[Clems Knee]

How good are the previous year's' squiggles at predicting early games? I would assume the most movement is in the first few rounds, but that wasn't true of teams that continued their march all year, such as WC and the Bulldogs last season?

And yes I'm hoping that the squiggle will fall in love with my team again.

 

[catstorm]

And yes I'm hoping that the squiggle will fall in love with my team again.

I don't think the squiggle ever has loved your team, you don't score enough I can't remember Freo ever being in the flag hot spot.
Freo usually hangs around Sydney 05, but that was an aberration that has not been repeated.

 

[Marc_Remillard]

I found one with 30% luck! That's almost credible. It was the sole Melbourne flag in 25,000 simulations.

A series of unlikely coincidences conspire to hand the Demons the flag:

• Melbourne squeak into 8th spot with 11 wins and a terrible percentage.
• They face bunny sides Geelong and Richmond in the first two weeks of the finals and win by 2 and 17 points respectively.
  
• A 1-pt win after extra time in a horrendous Roos prelim slogfest, defeating Port Adelaide 30-29. Maybe it was raining.
• Hawthorn have the mother of all off days for the GF (maybe they celebrated too early), going down to Melbourne by three goals.

I laughed

 

[Clems Knee]

I don't think the squiggle ever has loved your team, you don't score enough I can't remember Freo ever being in the flag hot spot.
Freo usually hangs around Sydney 05, but that was an aberration that has not been repeated.

July 5 2014. Page 49 of this thread. I think we stayed at the top for a few rounds.

 

[iluvparis]

On that point is there an easy way to get each club's all time high squiggle rating?

 

[Final Siren]

July 5 2014. Page 49 of this thread. I think we stayed at the top for a few rounds.

Convenient link. That was pre-Flagpole, though, when it wasn't very good at tipping premiers.

I noticed this chart I posted around the same time, mapping First Round Priority Picks from 1997 to 2013. (Flags are Priority Picks, red crosses are none.) Really shows how much harder they've become to get:

Melbourne 2013 deserved one, and probably would've gotten it if GWS weren't stinking up the place at the same time.

There's talk of a Carlton PP this year, but to get it the Blues will probably have to be the worst team of the last 20 years.

 

[JaxsonB]

How good are the previous year's' squiggles at predicting early games? I would assume the most movement is in the first few rounds, but that wasn't true of teams that continued their march all year, such as WC and the Bulldogs last season?

And yes I'm hoping that the squiggle will fall in love with my team again.

As someone that just got into betting late last year and is looking to have a good full year this time around, I was curious about this too. So I've gone through and had a look at the predictions and outcomes of past years. Here's what I found (including predicted margins, as these tend to be an indicator of confidence in the tip for the Squiggle):

2015: 3/9 wrong; predicted margins 8, 8, 21; 70.87% tips correct over whole year
2014: 5/9 wrong; predicted margins 3, 76, 18, 9, 31; 71.98% tips correct over whole year
2013: 6/9 wrong; predicted margins 40, 0, 15, 43, 5, 30; 72.46% tips correct over whole year
2012: 4/9 wrong; predicted margins 12, 65, 12, 47; 77.78% correct tips over whole year
2011: 2/8 wrong; predicted margins 32, 3; 77.55% correct tips over whole year
2010: 2/8 wrong; predicted margins 5, 38; 68.28% correct tips over whole year

Compares those figures above to R7, when things have arguably started to settle, at least a bit:

2015: 1/9 wrong; predicted margin 38
2014: 2/9 wrong; predicted margins 63, 46
2013: 2/9 wrong; predicted margins 2, 7
2012: 5/9 wrong; predicted margins 32, 6, 13, 17, 14 (ok, this one's a bad example - one of those upset rounds that come along now and then. R6 got 8/9 right, and R8 got 7/9 right)
2011: 3/8 wrong; predicted margins 17, 6, 10 (like above, R6 got 7/8 and R8 got 6/8)
2010: 3/8 wrong; predicated margins 14, 9, 16

Essentially, since the introduction of the 18 team format, Squiggle has predicted the first round with about 50% accuracy overall. Basically, if you relied solely on it to tip in round 1, you'd more or less be flipping a coin. There's not even a safe enough margin (again, a measure of confidence) for R1 - as recently as 2014 it predicted Syndey would flog GWS, but GWS ended up winning by 5 goals.

Overall, I'd suggest using Squiggle only for games you're really unsure about in R1, and not as your first point of reference. Too much happens between seasons to rely on the round 1 prediction as your primary motivator.

Edit: as another measure, lets say that more than 21 points would be considered a 'big upset' in terms of Squiggle. From that viewpoint, for R1: 2015 had 1/9, 2014 had 2/9, 2013 had 3/9, 2012 had 2/9, 2011 had 1/8 and 2010 had 1/8. Compared to R7: 2015 had 1/9, 2014 had 2/9, 2013 had 0/9, 2012 had 1/9, 2011 had 0/8 and 2010 had 0/8. Basically, in R1 10 out of 52 games since 2010 have been big upsets according to Squiggle (using a predicted margin of 22 or more as the measure). In R7, that goes down to 4/52. Basically, one of the key factors of early season Squiggle tips is that big upsets are significantly more likely. After 1.5-2 months, things have settled a bit, and big upsets are less likely to come up.

 

[telsor]

Melbourne 2013 deserved one, and probably would've gotten it if GWS weren't stinking up the place at the same time.

I'd say Melbourne got one, and a big one at that, it just wasn't the usual on field 'pick'. (The AFL stepping in to fix them off field, including some guy called Roos).

 

[Final Siren]

As someone that just got into betting late last year and is looking to have a good full year this time around, I was curious about this too. So I've gone through and had a look at the predictions and outcomes of past years. Here's what I found (including predicted margins, as these tend to be an indicator of confidence in the tip for the Squiggle):

2015: 3/9 wrong; predicted margins 8, 8, 21; 70.87% tips correct over whole year
2014: 5/9 wrong; predicted margins 3, 76, 18, 9, 31; 71.98% tips correct over whole year
2013: 6/9 wrong; predicted margins 40, 0, 15, 43, 5, 30; 72.46% tips correct over whole year
2012: 4/9 wrong; predicted margins 12, 65, 12, 47; 77.78% correct tips over whole year
2011: 2/8 wrong; predicted margins 32, 3; 77.55% correct tips over whole year
2010: 2/8 wrong; predicted margins 5, 38; 68.28% correct tips over whole year

Compares those figures above to R7, when things have arguably started to settle, at least a bit:

2015: 1/9 wrong; predicted margin 38
2014: 2/9 wrong; predicted margins 63, 46
2013: 2/9 wrong; predicted margins 2, 7
2012: 5/9 wrong; predicted margins 32, 6, 13, 17, 14 (ok, this one's a bad example - one of those upset rounds that come along now and then. R6 got 8/9 right, and R8 got 7/9 right)
2011: 3/8 wrong; predicted margins 17, 6, 10 (like above, R6 got 7/8 and R8 got 6/8)
2010: 3/8 wrong; predicated margins 14, 9, 16

Essentially, since the introduction of the 18 team format, Squiggle has predicted the first round with about 50% accuracy overall. Basically, if you relied solely on it to tip in round 1, you'd more or less be flipping a coin. There's not even a safe enough margin (again, a measure of confidence) for R1 - as recently as 2014 it predicted Syndey would flog GWS, but GWS ended up winning by 5 goals.

Overall, I'd suggest using Squiggle only for games you're really unsure about in R1, and not as your first point of reference. Too much happens between seasons to rely on the round 1 prediction as your primary motivator.

Edit: as another measure, lets say that more than 21 points would be considered a 'big upset' in terms of Squiggle. From that viewpoint, for R1: 2015 had 1/9, 2014 had 2/9, 2013 had 3/9, 2012 had 2/9, 2011 had 1/8 and 2010 had 1/8. Compared to R7: 2015 had 1/9, 2014 had 2/9, 2013 had 0/9, 2012 had 1/9, 2011 had 0/8 and 2010 had 0/8. Basically, in R1 10 out of 52 games since 2010 have been big upsets according to Squiggle (using a predicted margin of 22 or more as the measure). In R7, that goes down to 4/52. Basically, one of the key factors of early season Squiggle tips is that big upsets are significantly more likely. After 1.5-2 months, things have settled a bit, and big upsets are less likely to come up.

Nice work!

Here is tip accuracy over the last 20 years:

Rounds 23 and 24 have only been around since 2011, and the comp has been relatively predictable since then. (Squiggle has 74.1% correct tips over 2011-2015 compared to 68.5% for 1996-2015.)

So it does looks like some fairly steady improvement over the season... which should be the case for all tippers, human and otherwise.

Definitely Round 1 is a weak point, since it's gone a long time with no data.

 

[Donakebab]

Love your work Final Siren. The powers that be should give you a Squiggle badge!

 

[Suma Magic]

Maybe a favourites r1-7 into squiggle r8-23 could be a good tipping strategy.

 

[Whats a Battle]

Maybe a favourites r1-7 into squiggle r8-23 could be a good tipping strategy.

I'm using this one for sure.

 

[cryptor]

Maybe a favourites r1-7 into squiggle r8-23 could be a good tipping strategy.

You'd do even better on top of this if you don't blindly follow it but adjust for injuries, etc.