Certified Legendary Thread Race for the flag, in squiggly lines

[Byrons Firen]

You are stealing Byrons Firen thunder. Or are you he?

Nay lads t'is not I

 

[Roobs321]

Pre-season Sydney-North GF prediction starting to look a little bit more achievable now, according to the squiggle.

 

[Roobs321]

Also, whilst the ladder predictor might look a little silly with 7 teams at 17-5 or better (I'd pay to see that), I think the ladder positions mostly look spot on.

 

[Final Siren]

Round 11, 2014

Interactive squiggle

Hawthorn disappoints, but the squiggle likes the work of Sydney and North Melbourne. Both those sides have interesting trajectories, with the Roos completing a process of switching from an attacking team into a defensive one, and the Swans just blasting onwards from Round 5.

And Melbourne are now officially less bad than St. Kilda! *pop* *champagne*

 

[Dangas]

This chart is interesting, clearly the Paul Roos influence on Ross Lyon depicted with Freo most defensive and doing circle work.

 

[Spoken4]

This chart is interesting, clearly the Paul Roos influence on Ross Lyon depicted with Freo most defensive and doing circle work.

Photo of Ross Lyon arriving at Freo oval:

 

[Shane Hird]

Bombers have done two loops. That can't be good.

 

[The Swans Blog]

Lol the demons drew boobs

 

[Augustine]

We moved forward ever so slightly after the Port game which was in retrospect a great win. Does the squiggle take these things into account later on?

 

[KillaKanga]

Round 11, 2014

Interactive squiggle

Hawthorn disappoints, but the squiggle likes the work of Sydney and North Melbourne. Both those sides have interesting trajectories, with the Roos completing a process of switching from an attacking team into a defensive one, and the Swans just blasting onwards from Round 5.

And Melbourne are now officially less bad than St. Kilda! *pop* *champagne*

Going by this the top 7 is almost a lock with the Suns being the 'will they, won't they team. The other 7 in the 8 at the moment will most probably play finals.

 

[TennisPlayerAndy]

We're heading to Freo territory.....

 

[SJ]

Love this thread guys.

Just as we can calculate a team specific modifier for individual grounds, have we tried team specific modifiers vs other teams? Eg Cats better and Hawks worse head to head? Does this help the model at all, especially when combined with team specific ground modifiers?

Have you tried this Final Siren?

For simplicity, say there was a four team competition where:

Every time they played, Team A would beat Team B 100-90.
Every time they played, Team B would beat Team C 100-90.
Every time they played, Team C would beat Team D 100-90.
Every time they played, Team D would beat Team A 100-90.

Can you apply a match-up multiplier?

 

[Membling]

We're heading to Freo territory.....

It's going to get interesting....

 

[tazaa]

It's going to get interesting....

Norths 2014 is starting to mirror Fremantles 2012. Lost some unthinkable games early on but became a defensive unit that beat Geelong at the G by years end.

 

[mattis117]

I love this thread - the predictor is surprisingly accurate.

The ladder predictor is kinda pointless in it's current format. A bit of thought and it could be a really cool addition. What I mean by that is, as Final Siren said it just picks a winner rather than basing the ladder on probabilities. The results just seem very unlikely.

 

[KillaKanga]

It's going to get interesting....

nice hat

 

[Final Siren]

Have you tried this Final Siren?

For simplicity, say there was a four team competition where:

Every time they played, Team A would beat Team B 100-90.
Every time they played, Team B would beat Team C 100-90.
Every time they played, Team C would beat Team D 100-90.
Every time they played, Team D would beat Team A 100-90.

Can you apply a match-up multiplier?

I haven't tried it, no. Just from my own feeling, it's not that common for a team to consistently have the wood over another, regardless of venue and ladder position. I suspect what would happen is the fact that it doesn't happen very often would cause it to be a tiny factor, which kind of negates the basic idea of being able to tip a big upset based on one team being another's bunny.

That said, it is worth trying just for the amusing Geelong/Hawthorn chart that would result.

I love this thread - the predictor is surprisingly accurate.

The ladder predictor is kinda pointless in it's current format. A bit of thought and it could be a really cool addition. What I mean by that is, as Final Siren said it just picks a winner rather than basing the ladder on probabilities. The results just seem very unlikely.

It actually does that as of about 45 minutes ago. But it's still very much a work in progress.

I haven't checked its accuracy in much depth yet, but I see that this time in 2012, i.e. after Round 11, it did a terrific job of predicting the finals, getting the matchups almost exactly right all the way through to the Grand Final (where it thought Hawthorn would beat Sydney). Which is nifty.

This time last year, it wouldn't have been so great, getting the preliminary finalists right but the results wrong (expecting a Geelong v Sydney Grand Final).

 

[mattis117]

I haven't tried it, no. Just from my own feeling, it's not that common for a team to consistently have the wood over another, regardless of venue and ladder position. I suspect what would happen is the fact that it doesn't happen very often would cause it to be a tiny factor, which kind of negates the basic idea of being able to tip a big upset based on one team being another's bunny.

That said, it is worth trying just for the amusing Geelong/Hawthorn chart that would result.


It actually does that as of about 45 minutes ago. But it's still very much a work in progress.

I haven't checked its accuracy in much depth yet, but I see that this time in 2012, i.e. after Round 11, it did a terrific job of predicting the finals, getting the matchups almost exactly right all the way through to the Grand Final (where it thought Hawthorn would beat Sydney). Which is nifty.

This time last year, it wouldn't have been so great, getting the preliminary finalists right but the results wrong (expecting a Geelong v Sydney Grand Final).

Maybe this will just be an odd year - but a three win difference between 7th and 8th? That would surprise me.

 

[Final Siren]

We moved forward ever so slightly after the Port game which was in retrospect a great win. Does the squiggle take these things into account later on?

Nope. No retrospective credit.

 

[Smiling Buddha]

Amazing work. Kudos to 'max barry'.

 

[Chrizzt]

And Melbourne are now officially less bad than St. Kilda! *pop* *champagne*

I follow this thread loosely, so apologies if you've defined this, but how do you define them being 'less bad'? Melbourne are better defensively but Saints are better offensively, and if anything the Saints are still closer to the Off/Def bisector line.

 

[mattis117]

I follow this thread loosely, so apologies if you've defined this, but how do you define them being 'less bad'? Melbourne are better defensively but Saints are better offensively, and if anything the Saints are still closer to the Off/Def bisector line.

The best way to compare is to draw diagonal lines across the graph (y=-x+c basically). So the combination of defense and offence for Melbourne is slightly greater than for St Kilda.

 

[Chrizzt]

The best way to compare is to draw diagonal lines across the graph (y=-x+c basically). So the combination of defense and offence for Melbourne is slightly greater than for St Kilda.

Yeah I thought this might be the best way also, at a quick glance it looked like the Saints were clearly closer, but after careful numerical evaluation (holding a ruler against my monitor) I can see that the Dees are a bit ahead.

 

[im_a_lazy_sod]

Lol the demons drew boobs

Now we need two clubs to drop down badly enough to make nipples

 

[SimonH]

Yeah, consider this a fun version for now.

The major defect right now is that it simply tips every game individually and compiles the results. But this is not the right way to do it. If we have two teams playing each other 10 times, and Team A is expected to beat Team B 60% of the time, we should predict 6 wins to Team A and 4 wins to Team B. Not, as this predictor currently does, tip Team A to beat Team B every time, for 10 wins.

I love that last night's version (since watered-down) predicted that 7 teams could wind up on 17 or more wins! It's amazing that that's even technically feasible from here.

The squiggle ladder predictor might be a work in progress, but it quite accurately brings into focus something that I hadn't thought of before—that on the evidence so far, there are 7 sides with the ability to make some serious waves this year, and the rest are making up the numbers.