Certified Legendary Thread Race for the flag, in squiggly lines

Remove this Banner Ad

Swans supporter bragging about being successful is like someone that's on Centrelink bragging about how much money they've got.

I'm enjoying this thread too much for it to descend into merely flaming other clubs.

Oh, wait, you're paying out $ydney. My bad. Carry on.
 
Freo are way off to the right now. Ross Lyon hypermode has been activated.

(For a comparison, check out the Saints 2009 squiggle)

Great stuff.
I get the feeling conditions have played a part in that - both the Melbourne and Brisbane games were in trying conditions. I don't think this Freo side is better defensively than last year's (or even Saints '09).

Sydney have gone strongly to the right as well with their victory over West Coast, but that was in terrible conditions as well.
 

Log in to remove this ad.

Round 16, 2014

Hgkwa1d.png

Interactive squiggle (with predictor!)

A nightmare for Port Adelaide fans, as the Power sink while the Crows rise! Adelaide has moved fairly solidly into the top bracket of teams now, with good performances since Round 13 (36 points over North, a close loss away to Essendon, a 23-pt win over Port, and a 68-pt win away over the Giants). The squiggle predictor still has them in a battle for 8th spot with Gold Coast, another team that continues to head in the right direction in squiggle land.

Port's woes mean there's now real separation between the top 3 and the rest.

A few teams moved right this week (Western Bulldogs, Sydney, Fremantle, Brisbane, Essendon), which means improved defence, possibly aided by soggy conditions. Hopefully Freo and Sydney come back a little next week, because I don't want to have to change the graph axis.

Maybe the weather was to blame for Geelong's narrow victory at home over the Bulldogs, too, but they are now charting in their worst position of the year, sinking from premiership contender to roughly equal with Adelaide and Collingwood.

And North Melbourne are now officially the squiggle's least favourite team! Correct tips by team:
  • 15 West Coast
  • 13 St Kilda
  • 12 Western Bulldogs
  • 12 Geelong
  • 12 Greater Western Sydney
  • 12 Fremantle
  • 11 Melbourne
  • 11 Hawthorn
  • 10 Richmond
  • 10 Essendon
  • 10 Gold Coast
  • 10 Carlton
  • 9 Port Adelaide
  • 9 Sydney
  • 9 Brisbane Lions
  • 8 Collingwood
  • 8 Adelaide
  • 7 North Melbourne
In losing to Brisbane then defeating the Hawks, North showed how it can be easier to tip probabilistically. Before those games, you would have felt pretty safe saying North would win one of them. And you would have been right. But it wouldn't have been the game you expected.

Or maybe this proves that a good way to tip North is to toss a coin.

Fun With Algorithms

One algorithm I have is called BOUNCE. This one looks for teams that perform unexpectedly badly, then considers them a bit more likely to win the following week.

BOUNCE is not a great long-term performer, so its predictive value should not be overrated, but this year it has been going gangbusters and has 98 tips right, thanks to results like this:
  • Adelaide lost to Melbourne, then beat Collingwood
  • Essendon lost to Melbourne, then beat Richmond
  • Adelaide was thumped by Freo, then beat North Melbourne
  • Sydney lost to North Melbourne, then beat Fremantle
  • Carlton lost to Melbourne, then beat the Western Bulldogs
  • North Melbourne lost to Brisbane, then beat Hawthorn
 
It is indeed predicting a score of 65 for each team, and possibly the dullest Grand Final in living memory. But Freo's 65 is slightly better (65.1 vs 64.5), so it will bold them when it next updates.
Many people rate the drawn GF in 2010 as the best GF in recent years and that was 68-68.

As long as it's close I think people won't find it too dull.
 
With Freo playing GWS this weekend your axis may have to change after all.
Maybe it might but remember the squiggle will expect a massive win be Freo i would say in the order of 100 points so they would need to win by even more to make any kind of real change.

Teams move a lot when the crush other team the squiggle rate highly like the Sydney V Geelong game not when they rightly beat down weaklings.
 
Maybe it might but remember the squiggle will expect a massive win be Freo i would say in the order of 100 points so they would need to win by even more to make any kind of real change.

Teams move a lot when the crush other team the squiggle rate highly like the Sydney V Geelong game not when they rightly beat down weaklings.

The two large squiggle moves for Fremantle have been from their wins ove Brisbane and Melbourne.
 
Maybe it might but remember the squiggle will expect a massive win be Freo i would say in the order of 100 points so they would need to win by even more to make any kind of real change.

Teams move a lot when the crush other team the squiggle rate highly like the Sydney V Geelong game not when they rightly beat down weaklings.

Currently predicting a 122-49 win. Weather depending, I'd see that to be about right, so wouldn't expect too much movement in the squiggle this week.
 

(Log in to remove this ad.)

Fun With Algorithms

One algorithm I have is called BOUNCE. This one looks for teams that perform unexpectedly badly, then considers them a bit more likely to win the following week.

BOUNCE is not a great long-term performer, so its predictive value should not be overrated, but this year it has been going gangbusters and has 98 tips right, thanks to results like this:
  • Adelaide lost to Melbourne, then beat Collingwood
  • Essendon lost to Melbourne, then beat Richmond
  • Adelaide was thumped by Freo, then beat North Melbourne
  • Sydney lost to North Melbourne, then beat Fremantle
  • Carlton lost to Melbourne, then beat the Western Bulldogs
  • North Melbourne lost to Brisbane, then beat Hawthorn

Great work as always

I have a mate that puts a bit of time into Bounce backs and peaks.... Works very well. Do you look at the Peak side of the bounce? i.e. when a team hits a performance ceiling/unexpected result?
 
I'd really like to Sydney's squiggle start to climb north, but I guess it's been hard for the team to score consistently with Tippet out. It's too easy for defenders to get back and clog up the space in front of Franklin, especially when he's being tackled in the middle of a marking contest.

Holding the man, holding the ball, dropping the ball and throws are being ignored in the pursuit for a faster game.

I hate to say it, umpiring is as bad if not worse than when Goose was in charge.
 
I'd really like to Sydney's squiggle start to climb north, but I guess it's been hard for the team to score consistently with Tippet out. It's too easy for defenders to get back and clog up the space in front of Franklin, especially when he's being tackled in the middle of a marking contest.

Holding the man, holding the ball, dropping the ball and throws are being ignored in the pursuit for a faster game.

I hate to say it, umpiring is as bad if not worse than when Goose was in charge.

Bad call. At least Franklin gets away with head contact during bumps and doesn't get rubbed out ala Fyfe and co. Very well looked after is your 10 year star and your entire football club. You are not in a position to whinge until you play under the same rules as everyone else.
 
Bad call. At least Franklin gets away with head contact during bumps and doesn't get rubbed out ala Fyfe and co. Very well looked after is your 10 year star and your entire football club. You are not in a position to whinge until you play under the same rules as everyone else.

Haha. Oh god I ******g love this site.
 
Great work as always

I have a mate that puts a bit of time into Bounce backs and peaks.... Works very well. Do you look at the Peak side of the bounce? i.e. when a team hits a performance ceiling/unexpected result?
I've eyeballed it. I haven't analyzed it. It spits out things like this:
Code:
Round 16
North Melbourne (108) defeated Hawthorn (88) [North Melbourne by 20)
[bounce-91:12] Changing tip for bounce (was Hawthorn by 2.55793947504887)! GOOD IDEA!!!!!!!
UL => UW [-47 => 23] North Melbourne lost to Brisbane Lions by 4 (not -43), then beat Hawthorn by 20 (not -3)
EW => UL [2 => -23] Hawthorn beat Gold Coast by 53 (not 51), then lost to North Melbourne by 20 (not -3)
  bounce-91:12: +++RIGHT+++ (North Melbourne by 7.4) +/- 12.6
  istate-91:12: ---WRONG--- (Hawthorn by 16.4) +/- 36.4
Richmond (79) defeated Brisbane Lions (54) [Richmond by 25)
EW => EW [16 => 20] Richmond beat St Kilda by 44 (not 28), then beat Brisbane Lions by 25 (not 45)
UW => EL [47 => -20] Brisbane Lions beat North Melbourne by 4 (not -43), then lost to Richmond by 25 (not 45)
The above is not very readable, sorry. But it classifies results as Unexpected Win (UW), Expected Win (EW), Unexpected Loss (UL), or Expected Loss (EL), then looks to change tips when there's a close game expected between an UL and a EW.

The idea, of course, is that the UL team will be more motivated. Personally I believe motivation is a really big factor in football. Of all the factors I might include after scores and venue, motivation would be next in line. Teams that care more win more.

I see the evidence for this most clearly in the final round of the Home & Away season, when there can be big differences in the benefit of winning - e.g. for one team it won't affect them at all, but their opponent needs the win to make finals. I usually tip conservatively, but in the final round I go with motivation. Doesn't always work, of course, but it works more than it should. (I am still proud of correctly tipping three upsets in one round in 2011).

Unfortunately there is no objective way to score motivation. You can guess at it, and you can look at things that imply motivation, like assuming that a team that unexpectedly loses will have its pride stung. There might be more indicators, like milestone games. But you're moving further away from the thing you really want to measure, and less likely to get reliable numbers.
 

Remove this Banner Ad

Back
Top