Certified Legendary Thread Race for the flag, in squiggly lines

[lamby]

Hi Final Siren,

The Crows finally did a squiggly move to the right on the back of a soft game at home (our first one for the year!) But the 'soft' teams won't actually be there in September so play no part in winning the flag.

So I would love to see a squiggle starting at 50/50 just showing games against other top 8 teams. Does it change the positions much?

 

[LeverPuller]

The 9-9 split is very interesting

 

[WhiskeyTangoFoxtrot]

Hi Final Siren,

The Crows finally did a squiggly move to the right on the back of a soft game at home (our first one for the year!) But the 'soft' teams won't actually be there in September so play no part in winning the flag.

So I would love to see a squiggle starting at 50/50 just showing games against other top 8 teams. Does it change the positions much?

Trouble is, how do you know who the top 8 teams will be each year? Gold Coast were top 8 for the first 3 rounds but then died away.

This year is unusual in that the top 8 looks fairly set (with Port probably a sneaky chance), but most years there are a number of teams that rise or fall from the 8 as the year goes on, so results would drop in and out of the squiggles on a week-to-week basis, resulting in a lot of movement in the squiggles that didn't have anything to do with the game played that week.

 

[WhiskeyTangoFoxtrot]

Given the Crows are ahead of us on the flagpole , it's a big game for us too for that reason
Can't wait

Even moreso is the fact that a win to the Crows could see West Coast and the Crows basically swap places on the final ladder. So, for both teams, it could be the difference between finishing 2nd and getting a home qualifying final and finishing 5th and missing out on the double chance.

 

[Tippett2Goal]

Interesting Sydney and West Coast appear to be on exactly the same line, which says the Squiggle rates West Coasts thrashing of lower teams quiet highly? Is it possible to re-run the season with a cap on wins at 10 goals (ie any win that was by more than 10goals/60 points is treated as a 60pt win)?

Or would that be too disadvantageous for top teams with large expected lines?

 

[MWPP]

Interesting Sydney and West Coast appear to be on exactly the same line, which says the Squiggle rates West Coasts thrashing of lower teams quiet highly? Is it possible to re-run the season with a cap on wins at 10 goals (ie any win that was by more than 10goals/60 points is treated as a 60pt win)?

Or would that be too disadvantageous for top teams with large expected lines?

I also suspect that Squiggle also wouldn't have looked too kindly on Sydney's loss to Richmond/very narrow win over Brisbane

 

[Tippett2Goal]

I also suspect that Squiggle also wouldn't have looked too kindly on Sydney's loss to Richmond/very narrow win over Brisbane

Interestingly our Richmond loss we moved back 3 in def but up 1 in attack, so fairly marginal difference, also lost 4 def and may 1 atk in the Brisbane game
Presumedly helped by the away disadvantage, in both of those.

So its possible, the net result of those 2 games was approx -7def. No real impact on Atk as we scored roughly 100pts in both. (we got back +10def by beating the hawks)

But I guess looking at West Coast it isnt really the winning margin that sees you on that line anyway, its the defense of holding StK to 29pts

 

[MWPP]

Interestingly our Richmond loss we moved back 3 in def but up 1 in attack, so fairly marginal difference, also lost 4 def and may 1 atk in the Brisbane game
Presumedly helped by the away disadvantage, in both of those.

So its possible, the net result of those 2 games was approx -7def. No real impact on Atk as we scored roughly 100pts in both. (we got back +10def by beating the hawks)

But I guess looking at West Coast it isnt really the winning margin that sees you on that line anyway, its the defense of holding StK to 29pts

Yep. My point was that after our saints game, we didn't gain much offensively (1-2 points), but gained a lot defensively (6-7)- which happens to be about the same that Sydney lost (defensively) after Richmond/Brisbane

Final Siren posted a summary earlier, which highlighted how we have been the most predictable team in terms of tips. Squiggle is now 11/11 when it comes to the Eagles, which is better than its record for any other side. An average punter would have had a similar result by simply picking the favourite, too
ie we haven't had any upsets (wins or losses) this year
Sydney on the other hand have had at least one upset in each direction (Tigers, Swans)

 

[Komrade_Tiger]

This year is unusual in that the top 8 looks fairly set (with Port probably a sneaky chance), but most years there are a number of teams that rise or fall from the 8 as the year goes on, so results would drop in and out of the squiggles on a week-to-week basis, resulting in a lot of movement in the squiggles that didn't have anything to do with the game played that week.

not really, after a certain number of rounds, i believe its something like 8 or 9, the top 8 is practically set and anyone who changes it is breaking the trend (movement during the year notwithstanding)

 

[Final Siren]

Hi Final Siren,

The Crows finally did a squiggly move to the right on the back of a soft game at home (our first one for the year!) But the 'soft' teams won't actually be there in September so play no part in winning the flag.

So I would love to see a squiggle starting at 50/50 just showing games against other top 8 teams. Does it change the positions much?

Interesting Sydney and West Coast appear to be on exactly the same line, which says the Squiggle rates West Coasts thrashing of lower teams quiet highly? Is it possible to re-run the season with a cap on wins at 10 goals (ie any win that was by more than 10goals/60 points is treated as a 60pt win)?

Or would that be too disadvantageous for top teams with large expected lines?

These are the kinds of things human meat brains do: apply gut-feel rules to exclude factors from the data set. We can't juggle every result from every game of the season in our heads, so we run it through a few rules of thumb to separate signal from noise... or what we think is noise.

The problem is this isn't a good strategy so much as a good workaround for the fact that our brains are made of meat. Computers have no trouble considering large data sets all at once, and if you make them follow the same strategy, it just means throwing away useful data.

Since I have a meat brain, in the past I have tried strategies like capping winning margins. They very reliably do worse. Here is tip accuracy of the regular squiggle plus two clones that are only different in one way: they cap the maximum movement a team can have from a single game. Cap-10 is a fairly gentle cap, only affecting large movements, while Cap-5 is a harsher cap:

In fact, it's a strength of the squiggle that it remains interested in gigantic thrashings and games against weak teams. Because when you go back and look, you see that the big movements are very, very often indicators that a team has suddenly gotten a lot better.

For example, if you apply a movement cap to 2015, here are the first two games that get capped:

• R2 Port Adelaide 44 - 92 Sydney
• R8 Port Adelaide 43 - 76 Richmond

Richmond's season turned around at precisely R8, while Port were on the slide, so they're not good results to ignore.

There are a ton of these games throughout history. Probably the best example is R6 2007, when Geelong (10th) defeated Richmond (15th) 222 to 65. Any model that caps that result, or discounts it because it was only against a weak team, misses the Cats' meteoric rise that year.

Edit: Fixed an error where I had a West Coast game as one of the two biggest 2015 movements. That was actually in 2014.

 

[Hearts to hearts]

These are the kinds of things human meat brains do: apply gut-feel rules to exclude factors from the data set. We can't juggle every result from every game of the season in our heads, so we run it through a few rules of thumb to separate signal from noise... or what we think is noise.

The problem is this isn't a good strategy so much as a good workaround for the fact that our brains are made of meat. Computers have no trouble considering large data sets all at once, and if you make them follow the same strategy, it just means throwing away useful data.

Since I have a meat brain, in the past I have tried strategies like capping winning margins. They very reliably do worse. Here is tip accuracy of the regular squiggle plus two clones that are only different in one way: they cap the maximum movement a team can have from a single game. Cap-10 is a fairly gentle cap, only affecting large movements, while Cap-5 is a harsher cap:

In fact, it's a strength of the squiggle that it remains interested in gigantic thrashings and games against weak teams. Because when you go back and look, you see that the big movements are very, very often indicators that a team has suddenly gotten a lot better.

For example, if you apply a movement cap to 2015, here are the first two games that get capped:

• R2 Melbourne 30 - 123 West Coast
• R8 Port Adelaide 43 - 76 Richmond

Now we know that West Coast went on to make the Grand Final, which seemed pretty unthinkable in R2. And Richmond's season turned around at precisely R8.

There are a ton of these games throughout history. Probably the best example is R6 2007, when Geelong (10th) defeated Richmond (15th) 222 to 65. Any model that caps that result, or discounts it because it was only against a weak team, misses the Cats' meteoric rise that year.

It's fantastic how you explain all this so even a non-maths meat brain can (mostly) understand it.

 

[Bradesmaen]

Is anyone else amazed at the Giants line being so consistent? Every other team is all over the place but the Giants line is pretty straight in comparison.

 

[Final Siren]

Round 11, 2016

​

Animated!

Three clear winners this week!

1. North Melbourne took Richmond apart, like they always do, only more convincingly.

2. Adelaide held St Kilda to 6 goals while kicking 19: exactly what they needed, since their scoring power hadn't been accompanied by enough defensive strength.

3. Port Adelaide delivered a thumping away win over Collingwood. As the Pies and Tigers fell away, this put Port a lot closer to the top 8 than the bottom 9.

But all the top teams had a reasonable week, with the exception of Hawthorn, who should have beaten Melbourne by more than 18 points.

Good period for squiggle tipping, too, with 17/18 from the last two weeks! The line-ball decisions fell the right way.

Looking at the weather, I thought Gold Coast vs Sydney might be one of those games that breaks the squiggle, sending teams shooting off to the right. But they managed to score enough to keep it sensible.

The ever-changing Ladder Predictor! The nice thing about Port's surge is it means there might be a ridiculously small gap between top 4 and 9th. Or maybe less "nice" and more "terrifying," if you're the supporter of a team in contention.

And on Flagpole, after Hawthorn lost top spot last week, now the Crows squeak past them and the Eagles! That's an unusual call, but Adelaide have really only had one bad game this year: the 26-point home loss to Geelong. Which wasn't that bad. Aside from that, they've had honourable away losses to Hawthorn (by 3), North Melbourne (by 10), and the Bulldogs (by 15), and 7 wins with an average winning margin of 46 points.

West Coast vs Adelaide coming up this week. In fact, there are a few games between top sides, which offer the opportunity for lots of movement if anyone really takes charge.

Live squiggling for all ages!

 

[Taranaki AFL]

Sydney heading back to the future (2005) - they can't win a flag playing like this...

 

[Millky95]

Sydney heading back to the future (2005) - they can't win a flag playing like this...

Was mostly due to the weather this round, you'd think. We will go up and left between now and the end of the season. Doubt we'll be this far right again

 

[Sirlemsip]

Percentage is going to have such a huge impact at the end of this year that it isn't funny...

I look at Hawthorns and think that it could mean the difference between a double chance, a home elimination final, or an interstate away final.

Whilst the top 8 is pretty much 'set' in terms of the teams in there, the placings therein will probably not be solid until after round 23. Crazy.

 

[MC_9]

Percentage is going to have such a huge impact at the end of this year that it isn't funny...

I look at Hawthorns and think that it could mean the difference between a double chance, a home elimination final, or an interstate away final.

Whilst the top 8 is pretty much 'set' in terms of the teams in there, the placings therein will probably not be solid until after round 23. Crazy.

The predictor is showing only 1 win difference between 2nd and 8th. Also I imagine it would be the first time that a team could finish with 15 wins, and only come in 8th. There have been plenty of occasions that 15 wins will get you in the top 4.

Very even season at the top.

 

[The Swans Blog]

Interesting Sydney and West Coast appear to be on exactly the same line, which says the Squiggle rates West Coasts thrashing of lower teams quiet highly? Is it possible to re-run the season with a cap on wins at 10 goals (ie any win that was by more than 10goals/60 points is treated as a 60pt win)?

Or would that be too disadvantageous for top teams with large expected lines?

Squiggle loves a thrashing since it's more indicative of attacking teams and attacking teams win grand finals, not defensive teams *cough* 2005, 2012 *cough*

 

[A Swallow]

Percentage is going to have such a huge impact at the end of this year that it isn't funny...

I look at Hawthorns and think that it could mean the difference between a double chance, a home elimination final, or an interstate away final.

Whilst the top 8 is pretty much 'set' in terms of the teams in there, the placings therein will probably not be solid until after round 23. Crazy.

Yep. Very happy that North have managed to go from quite an ordinary percentage a few weeks ago, to 130.8% now. Only Sydney and to a lesser extent Geelong have significantly superior percentages than North now. Will be important.

 

[Sirlemsip]

Yep. Very happy that North have managed to go from quite an ordinary percentage a few weeks ago, to 130.8% now. Only Sydney and to a lesser extent Geelong have significantly superior percentages than North now. Will be important.

Interesting conundrum it leaves for Clarko & the Hawks this week:

option 1: Rest a few of the older guys against Essendon to keep them fresh for Kangas the following week

option 2: Minimal changes with a view to get a big percentage boost this week.

Will be an interesting chat around the selection table this week.

 

[Bruton77]

As measured by average margin error (difference between predicted and actual), the Bullies have been the easiest for the squiggle to tip, and Geelong the hardest:

Tips|Team|Average Margin Error\9|Western Bulldogs|15.29\9|North Melbourne|19.49\8|Sydney|21.19\7|Adelaide|21.52\10|West Coast|22.49\9|Essendon|23.00\8|Brisbane Lions|24.48\7|Fremantle|25.23\6|Carlton|30.53\7|Hawthorn|31.55\4|Richmond|32.95\7|Greater Western Sydney|33.55\5|Melbourne|35.30\8|St Kilda|36.73\4|Collingwood|38.62\5|Port Adelaide|39.37\7|Gold Coast|39.76\4|Geelong|40.01

This isn't the same thing as who's travelled the least distance on the chart, though, since teams can move even when the predicted margin is right, by being more attacking or defensive. The Bulldogs are a great example: They've become much lower-scoring this year, while also holding their opposition to lower scores. So on the weekend they were tipped to beat Collingwood by 21pts, and did, but in a low-scoring match, with both teams moving rightwards and downwards.

The Dogs have done quite a bit of this, so in terms of chart distance travelled, they've covered almost the most ground! But they haven't gotten much better or worse: they've just become less attacking and more defensive.

Squiggle Distance Travelled in 2016

​

But Geelong are still difficult. They've moved quite a lot almost every week.

And GWS are surprisingly average. Their squiggle movement looks bigger because it's all in the same direction.

More of this please (if there is any).

What years did the squiggle miss and vice versa?

Which individual teams moved furthest and least over a single season?

Thanks

 

[MWPP]

Interesting conundrum it leaves for Clarko & the Hawks this week:

option 1: Rest a few of the older guys against Essendon to keep them fresh for Kangas the following week

option 2: Minimal changes with a view to get a big percentage boost this week.

Will be an interesting chat around the selection table this week.

Surely they'll opt for option 2, with the bye coming up soon anyway

 

[Sirlemsip]

Surely they'll opt for option 2, with the bye coming up soon anyway

I'm certainly hoping for option 2.
Most of the older guys have already had a rest anyway. I think Burgoyne is the only one of the 30+ guys to play every game.

 

[The Dice Man]

Tigers squiggle has now drawn an image to show exactly where they are heading.

 

[GreyCrow]

Interesting conundrum it leaves for Clarko & the Hawks this week:

option 1: Rest a few of the older guys against Essendon to keep them fresh for Kangas the following week

option 2: Minimal changes with a view to get a big percentage boost this week.

Will be an interesting chat around the selection table this week.

Crows rested Thompson against Saints. Not sure how much more we would have scored. It was our kicking at goal that let us down.

I think you can rest 1 or 2 and it wont/shouldn't hurt Hawthorn