Certified Legendary Thread Race for the flag, in squiggly lines

[clogged]

Swans vs Dockers​

After a poor start, the Swans ignited in Round 5 and rocketed into good territory over the next few rounds. They've done everything right as far as the squiggle is concerned except kick big scores. Whether this matters remains to be seen; it's a defensive season, and finals are often low-scoring. Certainly the Swan's defence is amazing, with opponents almost always held to around 10 goals.

The Dockers have followed their 2013 script, hanging around in the 55-60/75-80 area for most of the season and heading northwards late. Despite a few extreme results (the Saints loss, keeping Brisbane to 22 points), it's been a consistent year for the Dockers, with no real trend in any direction.

Squiggle tip:

Sydney 79 - 63 Fremantle

Top 10 algorithms: 10 Swans (margin ranging from 8 to 19pts), 0 Dockers.

What did the algorithms say for last year's qualifying final between Freo and Geelong?

 

[The Swans Blog]

What did the algorithms say for last year's qualifying final between Freo and Geelong?

Clutching at straws mate...

 

[Engimal]

How many correct tips did the squiggle give in the H&A season?

 

[Clems Knee]

Compared to the last 20 years, this year looks even in the sense that there are currently 8 teams in good positions, with a big break to the rest. (Two of those "top 8" won't play finals: North Melbourne and Adelaide.) In past years, there are usually only 1-3 teams in the ideal zone, ahead of the pack.

I do think Geelong look very strong, though. It's not often that a team fails to win the flag from there, and when they do, it's usually to a team in an even better position - and there isn't one in 2013. That said, it does happen occasionally; it happened last year.

Finals predictions: Again, disclaimer: This isn't a world-beating tipping system. If it was right all the time, there would be no squiggles, because teams would stay right where they were. But the squiggle thinks the finals series will go like this:

Hawthorn 102 Sydney 85
Geelong 84 Fremantle 62
Collingwood 108 Port Adelaide 70
Richmond 97 Carlton 79

Sydney 86 Richmond 73
Fremantle 81 Collingwood 68

Hawthorn 92 Fremantle 78
Geelong 93 Sydney 69

Hawthorn 95 Geelong 93

It gives the Grand Final to Hawthorn due to home ground advantage, which I wouldn't really trust, since I don't think that's anywhere near as big a factor as usual in a Grand Final. In a 'neutral' venue it has Hawthorn 89 Geelong 99. And if they play at Simonds Stadium (you never know), it would be Hawthorn 83 Geelong 105.

This was the prediction of the 2013 finals based on the home and away. (On page 12 of this thread)

One out of four in the first week.

 

[The Speaker]

Clutching at straws mate...

Or, you know, he's just curious.

 

[Roobs321]

Interesting observations;

WCE certainly shifted to a more offensive side in the last couple weeks (might also largely be to do with their beneficial draw, had one of the better runs home).

Love how GWS has almost caught GC. Those two huge beltings the Giants copped mid-season seem to be holding them back slightly at the moment (in percentage more so than squiggle), but generally, have been very competitive and should be aiming for a fight for the 8 next year.

Sydney and Freo being close together now seems to promise an epic encounter this weekend. You get the sense that it has a prelim sort of relevance to it, who will be the defensive GF representative.

The most interesting thing for me about Bombers-North is how desperate both teams will be to have a finals win. Bombers haven't won a final in a decade, or a final beyond Week 1 since 2001. North this century also just have those SF wins over a young, rising Hawthorn which have been more than negated by a heap of beltings.

 

[clogged]

Clutching at straws mate...

Ahem.

I still find this thread to be pretty ******* funny. Richmond, 8 wins in a row, must be favourites, against a team that's won 16 of their last 17, narrowly going down to the Hawks at the MCG.

Yes of course, I see the logic, a team that was equal bottom 9 weeks ago, against the premiership favourites, and team that will most likely finish top. How obvious, Richmond must of course win this.

Sometimes people get things wrong. Even the squiggle.

 

[Crankyhawk]

145 tips for the season is pretty good IMO

 

[KissKiss]

145 tips for the season is pretty good IMO

According to the top of the page it would have been in the top 10 (equal 8th I think) on BF tipping

 

[Final Siren]

145 tips for the season is pretty good IMO

It is, but I have over 100 algorithms and every year some of them are going to fluke it. This year it's BOUNCE.

 

[Vintage]

Best thread on BF by far*.

*Hamish Mclachlan thread excluded.

 

[SterlingArcher]

Sorry Final Siren, you might have answered this - are the key indicators in this relative offence and defence? No other factors like Roby ?

 

[Illinois Nazi]

Sorry Final Siren, you might have answered this - are the key indicators in this relative offence and defence? No other factors like Roby ?

Nobody does other factors like Roby.

 

[AlibiMonday]

Nobody does other factors like Roby.

Are you meaning to tell me the comparative derivative of grass stain increases per quarter per player doesn't effect the result?

 

[cryptor]

Are you meaning to tell me the comparative derivative of grass stain increases per quarter per player doesn't effect the result?

#calculus #youwouldntunderstand

 

[Final Siren]

This was the prediction of the 2013 finals based on the home and away. (On page 12 of this thread)

One out of four in the first week.

Oh yeah, that was a crazy week! But even this week you should expect there to be at least one incorrect tip, since if each game has a 60/40 favourite, it is 87% likely that at least one favourite will lose. (1 - 0.6 x 0.6 x 0.6 x 0.6.) We just don't know which one(s)!

 

[clogged]

Oh yeah, that was a crazy week! But even this week you should expect there to be at least one incorrect tip, since if each game has a 60/40 favourite, it is 87% likely that at least one favourite will lose. (1 - 0.6 x 0.6 x 0.6 x 0.6.) We just don't know which one(s)!

North

 

[B4Bear]

An interesting version would be to have the relative positions of where each of the Grand Final participants were rated in relation to each other, with a tie line to the cup for that year. That way we can see how often the favourtie won/lost.

 

[Smedts_2]

I haven't read all 70 pages (or even a good portion of them) so excuse me if this has already been done, but do you have an output graph which shows the end of season position of each cup winner compared to the position of each losing Grand Finalist?

 

[Final Siren]

Sorry Final Siren, you might have answered this - are the key indicators in this relative offence and defence? No other factors like Roby ?

By indicators you mean inputs? The only inputs are the scores and the venue (state). If there are any other factors you think are relevant, like ins & outs, which team is more motivated, the weather, etc, you should weigh those yourself.

The squiggle doesn't consider those things for two reasons: I haven't found much long-term evidence for their importance, and I think the squiggle is more useful when it's simple. You can use it as a pretty good visualization of where a team is at, based purely on its scores, and make adjustments as you think necessary using your superior human brain.

 

[Final Siren]

Does the squiggle go into finals?

Yep. But the interactive squiggle hasn't been through a finals campaign before, so may break down a few times before I figure out what's wrong and fix it. The predictor and tips are a bit screwy at the moment because it doesn't know who's playing this weekend.

 

[Final Siren]

By the way, I'm taking suggestions on how to improve/modify that thing over the off-season.

The LIVE Squiggle

I was actually thinking about this: the interactive squiggle could extrapolate from the current scores. This would produce some fun results if you checked it when, say, Richmond were leading Sydney 33-0 at the 15 minute mark of the first quarter. Wouldn't be too hard to do, either, since it already updates by itself using live score data.

It'd be cool to have a pre-season prediction squiggle. Have a squiggle essentially "predicting" the entire year before round 1 then compare at the end of the year?

The season predictor will be back in 2015, and will indeed make what will inevitably prove to be a wildly inaccurate prediction of the year to come. It won't be that exciting, because the squiggle ignores everything that happens in the off-season, though, and so will predict a 2015 that looks very similar to 2014.

I'm not sure if I can make it so that you can go back in time and see what the season predictor was saying after Round <whatever>... it's a little tricky to implement. Past tips are a bit easier; I think I can do that.

I think the squiggle's largely fine as it is. The predictor is the killer app for me - it's elegant and based off sound fundamentals.

If you're looking for a couple of suggestions, here's some:

• I think a nice supplement to the squiggle would be a table which adds defence and offence together. This would essentially symbolise a form ladder/strength ranking.
• A tweak of the algorithm to progressively dilute the defence explosion that can happen when a team keeps it opposition to a minimal score. While mathematically winning 50pts:10pts is the same ratio as 250pts:50pts. The former seems to have a radical impact on the chart compared to the latter.
• I'm a big fan of Champion Data's 3300 rule, which essentially weighs every game evenly. If there was a way to adjust the algorithm to avoid point inflation and take it back to a zero-sum, that would be good. I'm assuming this would enable even greater comparisons between seasons.
• Some exotic selections would be interesting - largest movement per week/YTD (for/against), most stagnant, quarter-by-quarter sets,
• Personally, I would love ISTATE 91-12 (and any other models) tip data available for the year. Unfortunately, we lose access to it at the end of the week

I'm sure I could dream up loads more suggestions, but that will do for now.

Thanks for being the best poster on BF.

Thanks for these; they are great ideas!

 

[Final Siren]

I haven't read all 70 pages (or even a good portion of them) so excuse me if this has already been done, but do you have an output graph which shows the end of season position of each cup winner compared to the position of each losing Grand Finalist?

I don't, but you can see that using the interactive squiggle to click back through the years.

 

[Final Siren]

Squiggle PSA: I'm off to Adelaide this weekend to watch the Tigers prove the squiggle wrong! The interactive squiggle may have trouble doing its first ever live updates during finals, so if something goes wrong I'll fix it next week.

 

[Paul Kersey]

Squiggle PSA: I'm off to Adelaide this weekend to watch the Tigers prove the squiggle wrong! The interactive squiggle may have trouble doing its first ever live updates during finals, so if something goes wrong I'll fix it next week.

All the best!