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

[_DBF_]

when will it update?

 

[Seeds]

Up yours Squiggle!!!

Just kidding. Love your work and will be an avid follower again next year.

Why love the work of a backwards model that does not forecast.

 

[HTT]

Why love the work of a backwards model that does not forecast.

Would have ranked in the top 2000 site wide if it entered on footy tips with 117. Out of 500k total on that I reckon that's a pretty decent ability to forecast given squiggle can't actually watch the games.

 

[Gydafud]

*sad squiggle noises*

So uh.... sliding doors simulations? How many times does Cats win out of a 1000?

 

[Final Siren]

*sad squiggle noises*

So uh.... sliding doors simulations? How many times does Cats win out of a 1000?

There are a few different ways to approach this question. I'm not sure any are good for mental health, but anyway, let's do it.

The fairest objective way, imo, is to look at the actual result and see how hard you have to shake it to make it go the other way.

Before getting into that, though...

(1) You could say that what happened on the day was inexorable and immutable and there was never any chance of it happening differently; we just didn't know it yet. And, okay, fine, no-one can prove otherwise. If you subscribe to this theory, Geelong win 0 times out of 1,000. Obviously this is the only acceptable answer if you are a determinist or a Richmond supporter.

(2) You could look at pre-game predictions, like Squiggle giving Geelong a 61% win chance, and say the Cats would win 610 times out of 1,000. This is more reasonable than #1, but only a little, because it deliberately ignores the fact that we've gained a lot of relevant information since then, i.e. everything that actually happened on the day. But anyway, according to this theory, Geelong win 610 times out of 1,000 - or, if you'd rather trust the aggregate of all models, 560 times out of 1,000; or the markets, 480 times out of 1,000.

(3) You could try to adjust for that by feeding in the game result and ask the model to make another prediction, as if Saturday was a warm-up for the real GF this week. A bit more reasonable, and you get Squiggle still favouring Geelong, but not by as much, with a 53% win chance (530 wins out of 1,000). But this is still reality-defying, because Saturday wasn't a warm-up.

(4) Now for something that makes sense. One way to figure out how easily the result could have been different with a bit of luck or natural variation is to look at scoring shots. Richmond scored 12.9 to 7.8, and historically, when a team has 21 scoring shots to 15, they win 76.5% of the time. We know there's a lot of variability in scoring shot conversion, where teams have accurate and inaccurate days without warning and for no apparent reason, so this seems pretty fair. If, for example, Richmond had kicked the same score (12.9 81) and won by the same margin (31 pts) but Geelong had kicked 5.20, you would imagine the Cats must have had an unlucky day, and would do better in a re-run. (And indeed, a team with 21 scoring shots to 25 only wins 33.3% of the time.) So we can suppose there is a game in which everything is the same as the actual GF except that scoring shots convert into points at the usual rate, not what occurred on the day. Using this method, then, the GF result suggests that Geelong would win about 335 times out of 1,000. Richmond get slightly penalized here for being more accurate than usual in the GF and Geelong get a slight boost for being slightly less accurate.

(5) Similarly, you can calculate the standard deviation for footy results, assume that margins are normally distributed, and calculate the probability that it could have been at least 31 points less. This is the approach taken by Squiggle Sliding Doors and allows for a lot more luck/natural variation than #4, which only altered the results of scoring shots. So here you get alternate realities where Vlastuin isn't wiped out in the first quarter, or Ablett's shoulder holds up and he has a blinder, or the umpiring is heavily biased one way. Is this justified? Well, who knows. But there's quite a lot of variation in footy results, such that the standard deviation was 31.3 pts in 2020, so by this the chance of a 31 or 32-point shift in Geelong's favour is about 16%. Which gives you 160 wins out of every 1,000 games.

(Note: Here is where you might make an argument for momentum, or discounting the value of junk-time goals, but I don't really believe in either of those - even though the GF certainly presented a pretty good case for both. Those effects don't show up often enough to trust.)

 

[Dr Tigris]

Interesting that the Squiggle has the 2020 premiership out beyond the 2017 and 2019. I didn't feel that way about the team. They sort of scrapped and scrapped their way through this year. Hitting top gear on a few occasions only.

But a few years ago Final Siren was very clear that teams with great balance, but more inclined to attack won premierships - cause that's where they were on the chart. We've now got a large cluster of Premierships in a defensive zone.

So did things change over the last 5/6 years? Well I'd say yes. In an interview Lethal Matthews said he'd watch 40 to 50 games this year live and he really was impressed by how much better the defensive set ups were than they used to be. He said that the 3peat Hawks were nowhere near as good as this Tigers side because of this. I'll take his word on it. What we are seeing is a sea change where sophisticated defensive structures are being matched with certain skills becoming much more important to change how you win AFL footy games. IMHO only.

Any comment FS?

 

[Final Siren]

Interesting that the Squiggle has the 2020 premiership out beyond the 2017 and 2019. I didn't feel that way about the team. They sort of scrapped and scrapped their way through this year. Hitting top gear on a few occasions only.

But a few years ago Final Siren was very clear that teams with great balance, but more inclined to attack won premierships - cause that's where they were on the chart. We've now got a large cluster of Premierships in a defensive zone.

So did things change over the last 5/6 years? Well I'd say yes. In an interview Lethal Matthews said he'd watch 40 to 50 games this year live and he really was impressed by how much better the defensive set ups were than they used to be. He said that the 3peat Hawks were nowhere near as good as this Tigers side because of this. I'll take his word on it. What we are seeing is a sea change where sophisticated defensive structures are being matched with certain skills becoming much more important to change how you win AFL footy games. IMHO only.

Any comment FS?

100%. Strong defence-first teams used to wind up in the Ross Lyon Dead Zone where they never won premierships. But the landscape has hugely changed in the last 5 years.

Not only has every post-Hawks premier been a strong defence-first team, practically all the contenders are, too.

Compare for example 2012, which used to be one of only two years since 2000 when a defensive team won the flag:



The Swans won, but they were the only defence-minded team in the comp apart from Fremantle.

Since then, seasons increasingly look like this instead, where the premier is defence-oriented and so are almost all the good teams:



In 2020, all the Top 4 except Brisbane were defence-oriented; in 2018, Melbourne was the only exception; in 2017, Adelaide were.

So it's totally flipped around, and now it's the defensive teams getting solid finals results, while the attack-minded teams flame out.

 

[Dr Tigris]

Thanks FS Amazing graphics. 2012 much more spread out, but also much more attacking. I reckon it's the evolution of coaching where teams have worked out how the defend and attack. Good old Ross just did the defend bit.

I have a personal theory about what the Tigers are doing that I haven't seen from anyone else. I have to finish a write up and post it at some point. But when teams realised that contested ball wasn't correlated with scoring, and that good defence can be a springboard to attack the game changed. Once you view the world differently then things that didn't use to happen become what should happen.

 

[Gydafud]

Thanks for the amazingly detailed response to my somewhat rhetorical question. I certainly didn’t expect that level of detail. I think my mental health is recovering thanks to that answer.

Tigers are inevitable. Accept your Tigers.

 

[Joao]

So what you're saying is bring back Ross and Roos?

Ahead of their time they were.

 

[Dr Tigris]

So what you're saying is bring back Ross and Roos?

Ahead of their time they were.

Apart from the bit of scoring as well as being defensive.

 

[Joao]

Apart from the bit of scoring as well as being defensive.

Not sure about Roos but Saints in 09 and us in 13 could score just fine. Think they occupy a similar spot on the squiggle to some of the recent premiers.

 

[Dr Tigris]

Not sure about Roos but Saints in 09 and us in 13 could score just fine. Think they occupy a similar spot on the squiggle to some of the recent premiers.

I was partly tongue in cheek. However I think that modern teams have a really solid understanding of how to score from defence. Much better than it used to be.

The Squiggle hides as much as it illuminates. But it does illuminate a lot.

 

[Final Siren]

Interesting that the Squiggle has the 2020 premiership out beyond the 2017 and 2019. I didn't feel that way about the team. They sort of scrapped and scrapped their way through this year. Hitting top gear on a few occasions only.

Just on this part, I'll probably take a better look at Richmond's years in the future, but at first glance, I think it's reasonable to say 2020 was their best year.

In 2017, Richmond were coming off a 13th-placed finish and reached the finals with a percentage of 118%, including two losses by over 10 goals. The finals series was outstanding, of course, but before that, there wasn't a whole lot to suggest Richmond were a flag team.

In 2019, they started from a much higher base, but were outside the Top 8 as late as Round 14, and wobbling so badly that you saw these kinds of articles:


Home & away percentage was 114%, which is even worse than 2017. Another huge finals series, but there weren't any really strong flag rivals (unlike 2017 and 2020), and they got GWS (who'd finished 6th) in the GF.

In 2020, although the Tigers only finished the H&A season 3rd again, they did so with a percentage of 130% and without the benefit of a single game with home advantage. Their 12.5 wins from 17 games is a better win ratio than their 15 or 16 wins from 22 in 2017 and 2019 respectively. And although the finals were a tougher slog, they comprehensively defeated a highly-rated GF opponent.

So Richmond were probably more of a force throughout the entire year in 2020 than in either 2017 or 2019, and while their finals performance wasn't as dominant, it's good enough to hold up.

 

[Dr Tigris]

Just on this part, I'll probably take a better look at Richmond's years in the future, but at first glance, I think it's reasonable to say 2020 was their best year.

In 2017, Richmond were coming off a 13th-placed finish and reached the finals with a percentage of 118%, including two losses by over 10 goals. The finals series was outstanding, of course, but before that, there wasn't a whole lot to suggest Richmond were a flag team.

In 2019, they started from a much higher base, but were outside the Top 8 as late as Round 14, and wobbling so badly that you saw these kinds of articles:

View attachment 997138
Home & away percentage was 114%, which is even worse than 2017. Another huge finals series, but there weren't any really strong flag rivals (unlike 2017 and 2020), and they got GWS (who'd finished 6th) in the GF.

In 2020, although the Tigers only finished the H&A season 3rd again, they did so with a percentage of 130% and without the benefit of a single game with home advantage. Their 12.5 wins from 17 games is a better win ratio than their 15 or 16 wins from 22 in 2017 and 2019 respectively. And although the finals were a tougher slog, they comprehensively defeated a highly-rated GF opponent.

So Richmond were probably more of a force throughout the entire year in 2020 than in either 2017 or 2019, and while their finals performance wasn't as dominant, it's good enough to hold up.

That's the problem with using your intuition for analysis versus a model. You tell yourself stories that may be biased (by dominance in finals) and aren't properly reflected in the reality of the data.

 

[KissKiss]

Just from memory wasnt the 2018 Tigers probably their best H&A season?

 

[Dr Tigris]

Just from memory wasnt the 2018 Tigers probably their best H&A season?

Absolutely.

But in 2017, 19 and 20 we were accelerating into finals. In 2018 we we decelerating. I'd rather hit the finals in 3rd hitting top performance than be 1st and slowing down.

 

[threenewpadlocks]

In 2020, although the Tigers only finished the H&A season 3rd again, they did so with a percentage of 130% and without the benefit of a single game with home advantage. Their 12.5 wins from 17 games is a better win ratio than their 15 or 16 wins from 22 in 2017 and 2019 respectively. And although the finals were a tougher slog, they comprehensively defeated a highly-rated GF opponent.

So Richmond were probably more of a force throughout the entire year in 2020 than in either 2017 or 2019, and while their finals performance wasn't as dominant, it's good enough to hold up.

Also worth noting that the Tigers team was more injury-hit throughout the year than any of the last three years, which makes those wins even more impressive.

 

[HTT]

Also worth noting that the Tigers team was more injury-hit throughout the year than any of the last three years, which makes those wins even more impressive.

2019 was definitely worse than 2020 on an injury front, though come finals we were probably better off in 2019 only really missing Rance and having solid game time in other returning players.

 

[Final Siren]

Absolutely.

But in 2017, 19 and 20 we were accelerating into finals. In 2018 we we decelerating. I'd rather hit the finals in 3rd hitting top performance than be 1st and slowing down.

Good H&A seasons are quickly forgotten when the team falls over in finals. But yeah, Richmond's 2018 was exceptional. Finished on top with a buffer of 2 wins and 15 percentage points over 2nd, and 3 wins & 15 percentage points over the rest of the top 4.

There were two challengers capable of bursts of dazzling play but also plagued by inconsistency - Melbourne (5th) and Geelong (8th) - and no clear standouts from the rest of the top 8.

I'm not sure if I agree with the popular thinking that the rot had already set in late-season and finally caught the Tigers in the prelim... as opposed to just running into a red-hot Collingwood on the day, and being otherwise nicely set up for a flag tilt. It is obviously a very good thing to have a late form surge, but teams have won the flag by just being good all year and staying that way into September (Hawthorn 2014) or even sliding a bit from a high peak (Hawthorn 2015, Geelong 2009). And, conversely, plenty of teams have displayed late-season form surges that then abruptly hit a wall (Melbourne 2018, Sydney 2017, even Geelong 2008).

 

[Dr Tigris]

Good H&A seasons are quickly forgotten when the team falls over in finals. But yeah, Richmond's 2018 was exceptional. Finished on top with a buffer of 2 wins and 15 percentage points over 2nd, and 3 wins & 15 percentage points over the rest of the top 4.

There were two challengers capable of bursts of dazzling play but also plagued by inconsistency - Melbourne (5th) and Geelong (8th) - and no clear standouts from the rest of the top 8.

I'm not sure if I agree with the popular thinking that the rot had already set in late-season and finally caught the Tigers in the prelim... as opposed to just running into a red-hot Collingwood on the day, and being otherwise nicely set up for a flag tilt. It is obviously a very good thing to have a late form surge, but teams have won the flag by just being good all year and staying that way into September (Hawthorn 2014) or even sliding a bit from a high peak (Hawthorn 2015, Geelong 2009). And, conversely, plenty of teams have displayed late-season form surges that then abruptly hit a wall (Melbourne 2018, Sydney 2017, even Geelong 2008).

Very true.

Personally I reckon that Collingwood played about the best football in the last decade in that first half. Simply superb. Yes we had sickness etc. But the Pies were awesome that day. I guess that's the tail of the distribution coming into play.

 

[LicoriceAllsorts]

What is North Melbourne's attack rating atm Final Siren , I can't read it due to their position

 

[Dirty Bird]

What is North Melbourne's attack rating atm Final Siren , I can't read it due to their position

39.476897

 

[HTT]

What is North Melbourne's attack rating atm Final Siren , I can't read it due to their position

Obviously already been answered, but as an FYI you can see each team’s attack and defence listed in a table on the ‘flagpole’ tab

 

[Dirty Bird]

Obviously already been answered, but as an FYI you can see each team’s attack and defence listed in a table on the ‘flagpole’ tab

the way I actually did it was deselect every single team so it placed North right in the middle of the graph and allowed the rest to become visible.
Wish I knew the flagpole route though, would've made the job 15seconds quicker for myself