kane249
Wibble
- Oct 27, 2006
- 52,156
- 68,898
- AFL Club
- West Coast
- Other Teams
- Sth Freo, Liverpool, Chicago Bulls
Certified Legendary Thread Race for the flag, in squiggly lines
- Thread starter Final Siren
- Start date
- Tagged users None
you didn't really just post that in the squiggle thread did you? Do you even have the smallest idea what this ENTIRE thread is about??
treefingers
Cancelled
- Jan 20, 2008
- 6,514
- 8,395
- AFL Club
- Hawthorn
So according to flagpole, Freo are the least likely side in the 8 to win the flag?
I think of the flagpole as an "At their best" rating - which would somewhat logically make sense. Fremantle haven't been exceptionally good at any one time this season, but their 'worst' has been significantly better than the other top-4 sides.
According to the squiggle Fremantle probably wont kick more than 11 goals in either final
Fremantle will have to kick 13-14 goals to defeat West Coast or Hawthorn in either PF or the GF decider
- Aug 18, 2009
- 4,229
- 17,495
- AFL Club
- Richmond
- Thread starter
- #4,707
Round 22, 2015
Animated!
The Hawthorn line has been intersected! It took more than three years but IT FINALLY HAPPENED.
Adelaide have put together four huge performances in a row, embarrassing Richmond, Essendon, Brisbane, and now West Coast by an average of 73 points. The Crows have really only had one outright bad game in more than two months, the 52-point loss to Sydney in Round 18.
The Swans had to fight for their seat at the top 4 table and did so by thrashing St Kilda.
Hawthorn won by 72 points, but allowed Brisbane to sneak through a couple more goals than expected, upon which the squiggle frowned.
Further down, the Pies dragged the Cats down to their level, and the Blues sank a little deeper.
If all squiggle tips are correct in the final round, the home & away ladder will look like the below. (It's tipping Port over Fremantle even without knowing Lyon is planning to rest half the team; that's how down it is on the minor premiers.)
Some impressive lengthening from Adelaide in the flagpole! This algorithm takes a lot of notice of exceptional high-scoring performances, especially against good teams, so it loved what the Crows delivered against the Eagles.
Live squiggle: fun for children of all ages
YippeeYiYeo
Norm Smith Medallist
- Jun 21, 2015
- 9,087
- 14,295
- AFL Club
- West Coast
- Banned
- #4,708
Live flagpole needs some work. How port can be higher than fremantle at this stage is a little absurd.
HoneyBadger35
The Big Kahuna
- Aug 11, 2011
- 26,515
- 72,102
- AFL Club
- West Coast
- Moderator
- #4,709
Not necessarily - as has been mentioned, if Freo win it will be a big blow to the flagpole, but if it's had success picking teams closest to a flag and Freo don't go close it looks very smart.
- May 4, 2013
- 4,664
- 9,680
- AFL Club
- Hawthorn
Final Siren You know how like, you automatically put +12 or whatever when a team travels. Did you do any research with this number, or did it just seem like a good easy to start with? Would be interesting to see if it should be more or less
Footypie32
especias secreto
It's a good year to be a bird!
Meeting down at Geelong to discuss who's going to dress up as what.
"And now we cross to our man on the spot Jay "Catman" Clark."
sproketman
Club Legend
- Aug 6, 2015
- 1,226
- 2,071
- AFL Club
- Fremantle
- Banned
- #4,716
Then it's useless.
Not really. Given 'Current form' of the teams, I'd be inclined to believe that Port is closer to winning a Premiership than Freo is.
Sure Port might not actually win until 2016/17 or even later, but I personally don't see Freo winning this year (on current form).
Then with Pav/Sandi/McPharlin either gone in 2016 or at the least slowing down, Port will be a better chance to finish top 4.
Will be interesting to see the flagpole after next week (even though Ross is gonna rest up as many as he can).
sproketman
Club Legend
- Aug 6, 2015
- 1,226
- 2,071
- AFL Club
- Fremantle
- Banned
- #4,718
Nonsense. This was said last year. People thought Port were a premiership favourite going into 2015 while Freo were a chance to miss finals. Instead Port missed finals.
- Aug 18, 2009
- 4,229
- 17,495
- AFL Club
- Richmond
- Thread starter
- #4,719
Oh yeah, that number is the result of brute-force analysis of every possible combination of variables. 12 is what works best over the last 20 years.
But for particular years, other numbers (and algorithms) do better, of course. This year, interstate home advantage doesn't seem to have been that strong, and none of the top 15 algorithms by 2015 correct tips use it. (ISTATE-91:12 has 136 tips at 72.3%; the best algorithm so far, which is basically flagpole, has 143 tips at 76.1%. Although ISTATE-91:12 has had a good year by error margin; it's just had a lot of 1, 2 and 3 point tips go the wrong way. I think it's something like 2-10 from those.)
Last edited:
Data from the Flagpole:
Attack Defence Combine
7.
Fremantle 58.7 69.8 64.3
8.
Port Adelaide 65.0 62.8 63.9
Going from the figures, and those of the other teams, it is clear that the Flagpole depends more on attack then defence (as shown by WC being higher than Sydney on the 'pole, despite having a lower 'combined' total).
On current form Port has a far better attack, meaning that the flagpole feels they are a better chance for a premiership.
It doesn't matter if you don't agree with statistics and probabilities, it won't change them.
Attack Defence Combine
7.
8.
Going from the figures, and those of the other teams, it is clear that the Flagpole depends more on attack then defence (as shown by WC being higher than Sydney on the 'pole, despite having a lower 'combined' total).
On current form Port has a far better attack, meaning that the flagpole feels they are a better chance for a premiership.
It doesn't matter if you don't agree with statistics and probabilities, it won't change them.
- Aug 18, 2009
- 4,229
- 17,495
- AFL Club
- Richmond
- Thread starter
- #4,721
I am thinking about having it drop down flags to zero once a team can't make the finals. That would look pretty neat. Like, TAKE THAT, PORT.
I dunno about scaling it based on a team's chance of making finals, though, or the benefit of having home finals. As I mentioned before, I think that's only worthwhile if the algorithm can do a better-than-human job of it. Otherwise the graphic becomes less useful to you, because you lose the ability to read it for what it is--a conversion of match results into a form rating--and add your own assessment of things like the benefit of home finals, travel, finishing 5th rather than 4th, etc.
I just don't know of any great way to sensibly calculate that kind of thing, so I leave it to you.
- Aug 18, 2009
- 4,229
- 17,495
- AFL Club
- Richmond
- Thread starter
- #4,722
Correct! And this is highly influenced by the fact that attacking teams have won plenty of flags. If St. Kilda and Fremantle had saluted -- which you might think could easily have happened, given a little luck -- then I'd be using a different algorithm, since the current one wouldn't have done so well at predicting premiers.
sproketman
Club Legend
- Aug 6, 2015
- 1,226
- 2,071
- AFL Club
- Fremantle
- Banned
- #4,723
These are not statistics and probabilities. It's a number of heuristic models that change depending on their best fit for the result. Individual games are picked by standard Squiggle algorithm, the grand final is pick by its own separate algorithm than rates recent form higher, and flagpole is picked by a third unrelated one that rates scoring more favourably.Data from the Flagpole:
Attack Defence Combine
7.Fremantle 58.7 69.8 64.3
8.Port Adelaide 65.0 62.8 63.9
Going from the figures, and those of the other teams, it is clear that the Flagpole depends more on attack then defence (as shown by WC being higher than Sydney on the 'pole, despite having a lower 'combined' total).
On current form Port has a far better attack, meaning that the flagpole feels they are a better chance for a premiership.
It doesn't matter if you don't agree with statistics and probabilities, it won't change them.
The only algorithm that would make sense in terms of 'who is closest to the flag' would be one based on a Markov chain. Based on the Squiggle predictions for games, Hawthorn and West Coast are reasonable chances of not making the GF, if they didn't, would they still lead in the flag pole?
All hail Campo, Woosha and Squiggle!
The statistics of points for/against are used in the squiggle. The flagpole algorithm i'm sure uses a somewhat similar mechanism based on game stats to generate its data.
You can argue the semantics of the terms I used if you want. In any case my point remains - just because YOU don't agree with the specific algorithm used because it denigrates your team in some way doesn't make it 'useless'.
I personally find it quite useful, and would do so regardless of where Hawthorn is placed.
Similar threads
