Certified Legendary Thread Race for the flag, in squiggly lines

[Hobbes]

Final Siren would be interesting to see a squiggle with only the interstate teams - with a 'home' squiggle that only includes their home games and 'away' squiggle that only includes their away games. Would be interesting to see how far apart each team is and where they sit on the squiggle in relation to each other.

If we're doing home/away squiggles, why exclude Victorian teams?

 

[iluvparis]

If we're doing home/away squiggles, why exclude Victorian teams?

Because interstate sides play are higher % of their games interstate.

Really I'm just interested in West Coast to be honest - want to see how big the disparity is. Also semi interested for tipping - maybe its more effective if you use the interstates side home squiggle when they are at home and away squiggle when they are away.

 

[Final Siren]

Hey Final Siren

Is there a way to look at the other squiggle models to see how they rate things? I know the current one (ISTATE-91:12) is the most accurate one but it'd still be cool to look at

What I've thought about doing is adding BUILD YOUR OWN MODEL to the live squiggle, so you tell it which factors to look at and how to weight them, then you see the rankings & tips it produces. That could be fun.

ISTATE-91:12 is doing all right this year, sitting at 71st out of my 182 algorithms for tips and 6th for margin error. Although that's 1 tip worse than just tipping the home team! Last year you would have gotten 53.9% right sticking with the home team, but this year you'd be on 71.1%.

The leader at the moment is HOMULT-70:10, which has 80% right (36 tips). It awards a +/-10% score bonus/penalty for home advantage every game and is more sensitive to recent form (weighting the most recent game at 30%). No doubt it's just having a lucky year, although it's not bad over the long run, either, with a 10-year accuracy of 69%.

 

[Pykie]

What I've thought about doing is adding BUILD YOUR OWN MODEL to the live squiggle, so you tell it which factors to look at and how to weight them, then you see the rankings & tips it produces. That could be fun.

ISTATE-91:12 is doing all right this year, sitting at 71st out of my 182 algorithms for tips and 6th for margin error. Although that's 1 tip worse than just tipping the home team! Last year you would have gotten 53.9% right sticking with the home team, but this year you'd be on 71.1%.

The leader at the moment is HOMULT-70:10, which has 80% right (36 tips). It awards a +/-10% score bonus/penalty for home advantage every game and is more sensitive to recent form (weighting the most recent game at 30%). No doubt it's just having a lucky year, although it's not bad over the long run, either, with a 10-year accuracy of 69%.

Whats your best algorithm so far for margin standard deviation? With the best error rating?

 

[beert]

What I've thought about doing is adding BUILD YOUR OWN MODEL to the live squiggle, so you tell it which factors to look at and how to weight them, then you see the rankings & tips it produces. That could be fun.

ISTATE-91:12 is doing all right this year, sitting at 71st out of my 182 algorithms for tips and 6th for margin error. Although that's 1 tip worse than just tipping the home team! Last year you would have gotten 53.9% right sticking with the home team, but this year you'd be on 71.1%.

The leader at the moment is HOMULT-70:10, which has 80% right (36 tips). It awards a +/-10% score bonus/penalty for home advantage every game and is more sensitive to recent form (weighting the most recent game at 30%). No doubt it's just having a lucky year, although it's not bad over the long run, either, with a 10-year accuracy of 69%.

I can't remember whether I've asked this before, but where do you get your algorithms from? Do you just sample out a range of possible parameter combinations or is it more sophisticated than that? Is the parameter space large enough that you could look at some sort of genetic approach for breeding / mutating algorithms to get the best ones?

 

[Iron]

I can't remember whether I've asked this before, but where do you get your algorithms from? Do you just sample out a range of possible parameter combinations or is it more sophisticated than that? Is the parameter space large enough that you could look at some sort of genetic approach for breeding / mutating algorithms to get the best ones?

I heard he goes to space and collects radioactive space dust and cyphons it through an everyday household sift, then mixes it with unicorns blood, then lathers the paste over the algorithm eggsack to let it sit for 3 days in room temperature.

It will then hatch into a baby algorithm and Mr Squiggle's motherly skills kick in and he raise it to be most accurate algorithm yet.

 

[Lsta062]

If we are in the situation this year where we can overtake GWS to finish 8th instead of 9th in Round 23, I'm starting to think that I'd rather not if Adelaide is the team that finishes 5th.

I can already imagine the belting that Adelaide would give us in the EF at Adelaide Oval. I'd say it would be like the 2014 EF against Port over there, but a 100+ margin in the end.

 

[WhiskeyTangoFoxtrot]

If we are in the situation this year where we can overtake GWS to finish 8th instead of 9th in Round 23, I'm starting to think that I'd rather not if Adelaide is the team that finishes 5th.

I can already imagine the belting that Adelaide would give us in the EF at Adelaide Oval. I'd say it would be like the 2014 EF against Port over there, but a 100+ margin in the end.

There's a long way to go before we get to that point. If Richmond do finish 8th then they'd obviously be playing a lot better than they are currently.

 

[#1_pigeonracer]

Hi FinalSiren. Any chance we can sneak a peak at HOMULT-70:10's tips?

 

[YippeeYiYeo]

Surprised the dockers squiggle hasn't nose dived more.

 

[Final Siren]

Hi FinalSiren. Any chance we can sneak a peak at HOMULT-70:10's tips?

The only difference to regular squiggle is it thinks GWS will beat the Hawks.

Worth bearing in mind that over the last 5 years, HOMULT-70:10 has done 8 tips worse per season than regular squiggle.

 

[Benno_900]

Woah - massive shift towards the premiership zone

 

[Kane McGoodwin]

Surprised the dockers squiggle hasn't nose dived more.

Perhaps after tomorrow....

 

[Tas]

Perhaps after tomorrow....

Always a worry playing the Crows when a team is on struggle street, there isn't a bigger coach killer side than the Crows.

 

[Smedts_2]

I can't remember whether I've asked this before, but where do you get your algorithms from?

Roby.

 

[tazaa]

North needed this. Weve basically wiped out all the shooutouts in one game whilst keeping our attack gains

 

[Dirty Bird]

Last nights dour game knocked the attack out of both sides that the Crows are top 4 again.

Thankyou

 

[GoBlues13]

There's a long way to go before we get to that point. If Richmond do finish 8th then they'd obviously be playing a lot better than they are currently.

Yes. Let's face it, Richmond are playing as shit as hell at the moment. I don't get how they'll ever end up with a top 8 finish. I think even Melbourne have a better chance of making finals that Richmond and that hasn't happened in a while

 

[Lsta062]

There's a long way to go before we get to that point. If Richmond do finish 8th then they'd obviously be playing a lot better than they are currently.

Yeah, but Adelaide smashed us there last year a week after we defeated Hawthorn. We were playing some good footy around then too.

There is something about a team with a good ruck department and a strong forward line that we can't handle (North Melbourne, West Coast, Adelaide).

Yes. Let's face it, Richmond are playing as shit at the moment. I don't get how they'll ever end up with a top 8 finish. I think even Melbourne have a better chance of making finals that Richmond and that hasn't happened in a while

I can think of at least 8-9 sides better than Richmond right now, and that is being optimistic. I wouldn't take a bottom 4 finish out of the question for Richmond this year

 

[GoBlues13]

Yeah, but Adelaide smashed us there last year a week after we defeated Hawthorn. We were playing some good footy around then too.

There is something about a team with a good ruck department and a strong forward line that we can't handle (North Melbourne, West Coast, Adelaide).


I can think of at least 8-9 sides better than Richmond right now, and that is being optimistic. I wouldn't take a bottom 4 finish out of the question for Richmond this year

You never know, with Deledio back this week it could help them but if they keep on playing shit then I dunno, Carlton will probably finish in front of them on the ladder.

 

[Hawkinspired]

North needed this. Weve basically wiped out all the shooutouts in one game whilst keeping our attack gains

Still not sold on North. Clearly they were more polished than the Bulldogs but to concede 80 more disposals and 10 more inside 50's it was more the lack of skill on the Bulldogs part, than excellent defence from NM.

Bulldogs were as bad as I've seen them and the injury's look like their finally taking a toll. The squiggle still rates the Bulldogs as an excellent side hence the large squiggle movement.

Not to take anything away from North, they are showing consistency which they have struggled with for years.

 

[nowhereneargeelong]

Dogs forward entries were appalling.

They should have won that game.

 

[GoBlues13]

Still not sold on North. Clearly they were more polished than the Bulldogs but to concede 80 more disposals and 10 more inside 50's it was more the lack of skill on the Bulldogs part, than excellent defence from NM.

Bulldogs were as bad as I've seen them and the injury's look like their finally taking a toll. The squiggle still rates the Bulldogs as an excellent side hence the large squiggle movement.

Not to take anything away from North, they are showing consistency which they have struggled with for years.

Yes, I agree, the injuries are really kicking in for WB. JJ, Bob, Suckling, North have barely got an injury list. North's defensive pressure was amazing, and Jarrad Waite and Daniel Wells - they played really good. The way North played, they deserved the win

 

[Pykie]

Dogs forward entries were appalling.

They should have won that game.

We turned over a lot of their turnovers, should of belted them imo. We had around 30-40 more tackles.

They had no direct avenues to score.

Thompson killed Stringer.

 

[nowhereneargeelong]

We turned over a lot of their turnovers, should of belted them imo. We had around 30-40 more tackles.

They had no direct avenues to score.

Thompson killed Stringer.

They smashed you in possession and inside 50's. Yet they had a shocker.

I reckon Swans will belt you in a few weeks. Saints may even come close next week.