Certified Legendary Thread The Squiggle is back in 2023 (and other analytics)

[Forward Press]

??? Well then what is it meant to do? Thought it was a ‘predictive tool’???

Just an interesting different way to approach things. Final Siren has some metrics that he uses to churn out results and the teams move accordingly based on whether they exceeded or failed to meet those expected results.

I'm not asking for or expecting a model that's right "all of the time". I asked if it predicted last year that Richmond would win the flag. Negative. I then asked on what occasions it has predicted the premier. So far, crickets.

I believe it correctly predicted the Hawks threepeat and most flags before it. Like most other pundits, it didn't rate the Dogs and only started backing Richmond late in the season.

 

[Rod Stroker]

Just an interesting different way to approach things. Final Siren has some metrics that he uses to churn out results and the teams move accordingly based on whether they exceeded or failed to meet those expected results.

Based on that explanation it sounds like more of a lagging indicator than a leading indicator

 

[the harry]

I'm not asking for or expecting a model that's right "all of the time". I asked if it predicted last year that Richmond would win the flag. Negative. I then asked on what occasions it has predicted the premier. So far, crickets.

The model of every bookie in the country said that Crows should win.

Sample sizes of one are very, very bad to draw any conclusions from.

 

[Forward Press]

Based on that explanation it sounds like more of a lagging indicator than a leading indicator

It's both. It uses results from previous games to predict how the next round and the rest of the season would go. It doesn't profess to be comprehensive or authoritative, it's just a bit of fun (ie don't plonk down your hard earned cash at Sportsbet based on squiggle). If you want better analytical tools ask the gambling agencies, I'm sure they have better number crunchers.

 

[the harry]

When looking at anything that predicts, you need to remember that it uses probabilities and that even when in a two horse race, it predicts a 65%/35%, the 35% can still win, and by some margin, and the prediction isn't necessarily incorrect.

 

[Isaac Cumming No 1]

If we had a model that was right all of the time Final Siren would never share it and he'd be living with his harem on his yacht in Monaco.

It is merely a predictive tool but a neat one nonetheless and some interesting trends have been proven by it. And he posts it for free on BigFooty.

And any model based on probabilities wont be right all the time. The key word is probability.

 

[iluvparis]

Right. So why are people putting so much credence into what squiggle says when it didn't predict last years premier?

Because if we had a model that could pick 100% of games AFL would be pretty boring

 

[Final Siren]

I'm not asking for or expecting a model that's right "all of the time". I asked if it predicted last year that Richmond would win the flag. Negative. I then asked on what occasions it has predicted the premier. So far, crickets.

You can inspect Squiggle's accuracy via https://live.squiggle.com.au - every historical tip is there for past years under "Tips". The older model used in past years is at https://live.squiggle.com.au/v1/.

For Grand Finals in particular I've usually employed a specialist algorithm, which has not gone well. It missed Richmond last year (would have been correct under Squiggle 2.0), missed the Bulldogs in 2016, got Hawthorn in 2015. I think in 2014 I was using OFFDEF-75 for Grand Finals and missed the Hawks, even though both generations of algorithm since would have tipped them.

But if you're interested in statistical modeling, you'll know the best way to determine accuracy is to look at a score like Mean Absolute Average or Bits, not number of correct tips (and not number of correct tips in one game per year). There's a great article on why this is the case here: Tipping Accuracy vs MAE as a Footy Forecaster Metric.

 

[Final Siren]

??? Well then what is it meant to do? Thought it was a ‘predictive tool’???

Yeah, it is. It started out as a visualization tool, i.e. how have teams tracked over time, but once you have a decent model, you might as well predict what will happen. And they're intertwined: if your model makes good predictions, that's a sign that it's measuring the right things, and its rankings are reasonably trustworthy.

 

[Rod Stroker]

You can inspect Squiggle's accuracy via https://live.squiggle.com.au - every historical tip is there for past years under "Tips". The older model used in past years is at https://live.squiggle.com.au/v1/.

For Grand Finals in particular I've usually employed a specialist algorithm, which has not gone well. It missed Richmond last year (would have been correct under Squiggle 2.0), missed the Bulldogs in 2016, got Hawthorn in 2015. I think in 2014 I was using OFFDEF-75 for Grand Finals and missed the Hawks, even though both generations of algorithm since would have tipped them.

But if you're interested in statistical modeling, you'll know the best way to determine accuracy is to look at a score like Mean Absolute Average or Bits, not number of correct tips (and not number of correct tips in one game per year). There's a great article on why this is the case here: Tipping Accuracy vs MAE as a Footy Forecaster Metric.

Thanks.

 

[Pessimistic]

The tower is the best graphic of squiggle. it shows where your team could finish if it goes better, worse or stays the same (which is the most probable)

It occasionally has a freak out which makes very entertaining graphics. Picasso watch out

 

[Roovolution]

Because if we had a model that could pick 100% of games AFL would be pretty boring

But extremely profitable....

 

[spinynorman]

I'm not asking for or expecting a model that's right "all of the time". I asked if it predicted last year that Richmond would win the flag. Negative. I then asked on what occasions it has predicted the premier. So far, crickets.

Pretty sure it was predicting Hawthorn beating West Coast in the 2015 Grand Final from around Round 9 of that year, when they were 6th and 2nd on the table respectively (and most people thought West Coast were pretenders having a good early run that would fall apart). Also didn't much rate Fremantle in 2015 and your boys in 2016 when they were top of the ladder (and each team's supporters were calling the model clearly flawed for not having them in its Grand Final predictions). But, like everyone else in the country, it didn't see either of the last two premiers coming.

 

[Rod Stroker]

Pretty sure it was predicting Hawthorn beating West Coast in the 2015 Grand Final from around Round 9 of that year, when they were 6th and 2nd on the table respectively (and most people thought West Coast were pretenders having a good early run that would fall apart). Also didn't much rate Fremantle in 2015 and your boys in 2016 when they were top of the ladder (and each team's supporters were calling the model clearly flawed for not having them in its Grand Final predictions). But, like everyone else in the country, it didn't see either of the last two premiers coming.

I didn't really rate us either in 2016 when we won the first 9 straight, as we hadn't really beaten anyone strong in that stretch.
As many on here said at the time, we were the worst looking team to ever go 9-0 (and was probably proven as we only won 2 of the last 12 or something piss poor like that).

 

[Forward Press]

I believe Squiggle (accurately as it turned out) downplayed Freo 2015 and North 2016.

It hasn't downplayed us so far in 2018 so all hail the Squiggle.

 

[Komrade_Tiger]

If we had a model that was right all of the time Final Siren would never share it and he'd be living with his harem on his yacht in Monaco.

It is merely a predictive tool but a neat one nonetheless and some interesting trends have been proven by it. And he posts it for free on BigFooty.

you mean he doesn't?

 

[Komrade_Tiger]

??? Well then what is it meant to do? Thought it was a ‘predictive tool’???

it's a modelling tool,

It's usage extends in so far as helping to quantify the why a team is favourite in a match,

the historical data of the "premiership zone" is a good yardstick of teams, of course history can be defied as that is how history tends to be created but it just helps to quantify our opinions. As Final Siren has said in the past and i'm paraphrasing here, but humans can't handle huge amounts of data like squiggle can so it's a tool insofar as that it can spot trends and factors that aren't easily seen by mere mortals but it can't make logical leaps like we can with regard to injuries, conditions et al

tl;dr Squiggle should be used to counter opinions, support them or act as the basis, but you still need to take your own 'leap of faith' on the factors it can't quantify

 

[flamingEMBERS]

You can inspect Squiggle's accuracy via https://live.squiggle.com.au - every historical tip is there for past years under "Tips". The older model used in past years is at https://live.squiggle.com.au/v1/.

For Grand Finals in particular I've usually employed a specialist algorithm, which has not gone well. It missed Richmond last year (would have been correct under Squiggle 2.0), missed the Bulldogs in 2016, got Hawthorn in 2015. I think in 2014 I was using OFFDEF-75 for Grand Finals and missed the Hawks, even though both generations of algorithm since would have tipped them.

But if you're interested in statistical modeling, you'll know the best way to determine accuracy is to look at a score like Mean Absolute Average or Bits, not number of correct tips (and not number of correct tips in one game per year). There's a great article on why this is the case here: Tipping Accuracy vs MAE as a Footy Forecaster Metric.

do you apply the home ground advantage to the grand final for MCG Tenants?

 

[Final Siren]

do you apply the home ground advantage to the grand final for MCG Tenants?

I haven't in the past but will start doing so this year.

The modeling community has been split on this issue, with half of us using normal HGA and the other half dropping it for the Grand Final, since until recently, non-Victorian teams regularly overpowered Victorian ones for the flag, which lead to the idea that more neutral crowds and the greater relative importance of form-based factors meant HGA was reduced or eliminated. But we've had a spate of surprisingly strong GF performances from Victorian teams over the last five years, so the tide is turning.

In the absence of good evidence to the contrary, we should assume that Grand Finals are like any other game, so I'll be using the usual method of calculating home ground advantage for the 2018 GF.

 

[lamby]

it's a modelling tool,

As the saying goes: all models are wrong, but some are useful.

 

[HTT]

The tower is the best graphic of squiggle. it shows where your team could finish if it goes better, worse or stays the same (which is the most probable)

It occasionally has a freak out which makes very entertaining graphics. Picasso watch out

I love the tower late in seasons. Last year Richmond was basically predicted to finish 3rd or 6th, which created an interesting shape

 

[B4Bear]

It appears that there is a bug in the Squiggle.

It says that we will finish a minimum of 12th. I have been reliably informed by the media and BF that the wooden spoon was ours. How can this be?

Surely I shouldn’t believe The Squiggle (all hail) over the media and BF pundits?

 

[Miguel Sanchez]

It appears that there is a bug in the Squiggle.

It says that we will finish a minimum of 12th. I have been reliably informed by the media and BF that the wooden spoon was ours. How can this be?

Surely I shouldn’t believe The Squiggle (all hail) over the media and BF pundits?

Go back to the start of the season and Squiggle was saying that you were bugger all chance of finishing any higher than 11th.

So it's changed its prediction based on observations over the first 10 weeks of the season. As have the media, and mug punters.

Everyone got you guys wrong. And us. It's a nice feeling, enjoy the ride.

 

[Suma Magic]

Final Siren

May have been noted/explained before, but the finals have Stadium Australia in there for the Swans.

 

[B4Bear]

Go back to the start of the season and Squiggle was saying that you were bugger all chance of finishing any higher than 11th.

So it's changed its prediction based on observations over the first 10 weeks of the season. As have the media, and mug punters.

Everyone got you guys wrong. And us. It's a nice feeling, enjoy the ride.

Oh I agree. You too.

Wonder what Squiggle would do if it was a North/Eagles GF?