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

Remove this Banner Ad

Classic response - two evidenced based data conclusions wiped away by someone who 'watched' his team and decided they were completely dominant.
Heh, but I reckon North were dominant, too. They kept pumping it inside 50 and their forwards were getting perfectly timed, unhindered leaps at the footy - plenty of which didn't quiiiite come off. Could very easily have been a bigger smashing.
 
Do you think Geelong's chart location is being negatively impacted by their accuracy? They seem to have an opposite style, with slower, high percentage entries inside 50 and a high conversion rate on the I50's and shots.

Our accuracy has seemed abnormally high in a lot of games this year but now that we're half way through the season, perhaps it's not abnormal at all and we're just accurate.
I haven't watched much of Geelong, so don't know how they're playing, but the Cats and Eagles are both a LOT more accurate than everyone else this season. Geelong are converting at 60% and West Coast at 59%, vs a league average of 51%.

That is almost certainly unsustainable - last year, for example, no team finished higher than 56%. You have to go back to St Kilda 2004 to find a team with better conversion over a season.

Which implies that it won't continue. Although that needn't be a bad thing - it may be that Geelong just start attempting more difficult shots, since they're converting so well, and wind up scoring even more heavily. But their conversion rate is highly likely to drop.

(edit: To answer your question, though: Yes, Geelong is being marked down by Squiggle because it assumes they've been fortunate to convert so well.)
 
Last edited:
I'm certainly not going to drag out the replay to have a look for myself, but did the game have an abnormally high proportion of shots in general play? Expected score is pretty handy with set shots because not a lot changes aside from the measured angle and distance, but general play stuff has a lot of other variables that it can't account for. Only way I can see such a massive difference from reality.
I too quickly deleted that game.

Champion Data is a black box, so who knows how they calculate it, but they do allegedly take into account all that kind of stuff, including whether it's a set shot or in general play, and how much pressure the kicker was under (using their new pressure metrics). I have no idea how well it works, but they do attempt to control for those things.
 

Log in to remove this ad.

What does the squiggle rate more - a big win against a bad side or a marginal win against a good side?

Asking for a friend.

It rates performance exceeding expectation. A team clearing the expected margin, scoring higher than expected or holding a team to a lower score than predicted will cause more movement. Against good or bad sides.
 
What does the squiggle rate more - a big win against a bad side or a marginal win against a good side?

Asking for a friend.

Compared to how humans rate them - it rates the flogging of a bad side a lot more than a marginal win against a good side (the latter could have just been down to luck/natural variance)
 
It rates performance exceeding expectation. A team clearing the expected margin, scoring higher than expected or holding a team to a lower score than predicted will cause more movement. Against good or bad sides.

Nods ... on the weekend we got about the expected score for us and kept the Hawks to a lower score than expected so our icon didn't move vertically but it did move to the right reflecting our defensive win.
 
I haven't watched much of Geelong, so don't know how they're playing, but the Cats and Eagles are both a LOT more accurate than everyone else this season. Geelong are converting at 60% and West Coast at 59%, vs a league average of 51%.

That is almost certainly unsustainable - last year, for example, no team finished higher than 56%. You have to go back to St Kilda 2004 to find a team with better conversion over a season.

Which implies that it won't continue. Although that needn't be a bad thing - it may be that Geelong just start attempting more difficult shots, since they're converting so well, and wind up scoring even more heavily. But their conversion rate is highly likely to drop.

(edit: To answer your question, though: Yes, Geelong is being marked down by Squiggle because it assumes they've been fortunate to convert so well.)
Conveniently enough, today there is a Fox Sports article on exactly this!

https://www.foxsports.com.au/afl/af...r/news-story/c20bfe594092e767cfd3b67162092796

The numbers are slightly different because Champion Data can consider all shots at goal, including complete misses, whereas I could only compare the ratio of goals to behinds. The conclusion is the same, though: Geelong and West Coast have been a lot more accurate so far.

Champion Data also can examine Expected Scores, which factors in the difficulty of shots taken. This should rule out the idea that the higher accuracy of the Cats & Eagles is simply because they're getting into better positions and taking easier shots. Which means there is probably a pretty big luck component.

It's also interesting to see Brisbane high on this table, since there's another report out today ranking them as the team least affected by injury this season:

https://www.foxsports.com.au/afl/af...r/news-story/c20bfe594092e767cfd3b67162092796

Together, these tables suggest that Brisbane are operating somewhere near the top of their potential at the moment, and more likely to get worse in the future than better.
 
Conveniently enough, today there is a Fox Sports article on exactly this!

https://www.foxsports.com.au/afl/af...r/news-story/c20bfe594092e767cfd3b67162092796

The numbers are slightly different because Champion Data can consider all shots at goal, including complete misses, whereas I could only compare the ratio of goals to behinds. The conclusion is the same, though: Geelong and West Coast have been a lot more accurate so far.

Champion Data also can examine Expected Scores, which factors in the difficulty of shots taken. This should rule out the idea that the higher accuracy of the Cats & Eagles is simply because they're getting into better positions and taking easier shots. Which means there is probably a pretty big luck component.

It's also interesting to see Brisbane high on this table, since there's another report out today ranking them as the team least affected by injury this season:

https://www.foxsports.com.au/afl/af...r/news-story/c20bfe594092e767cfd3b67162092796

Together, these tables suggest that Brisbane are operating somewhere near the top of their potential at the moment, and more likely to get worse in the future than better.
Does historical data back up that Geelong and West Coast can't sustain this? I only ask because one of the WA journos (Duffield or Quartermaine I think) is always on about how often the Eagles practice skills, including goal kicking. Perhaps they are just benefiting from spending more time on it?
 
Does historical data back up that Geelong and West Coast can't sustain this? I only ask because one of the WA journos (Duffield or Quartermaine I think) is always on about how often the Eagles practice skills, including goal kicking. Perhaps they are just benefiting from spending more time on it?
The data I have access to strongly imply that it's unsustainable, yeah.

It's an inherently random stat with a pretty tight cluster (low deviation) centered around 52% across a very large data set, and we can safely assume that all teams have been trying hard to raise their numbers.

So the fact that Geelong & West Coast's numbers are so rare in whole-season data sets means they will almost certainly move back toward the mean.

I'm sure West Coast have been practicing their skills, and are better than most - good teams usually have better accuracy numbers - but it's less likely that they've unlocked a secret that has eluded every other team for the last 15 years.

They and Geelong will probably finish around the top of the year's accuracy tables, but I'd guess with numbers like 55-56%, down from the current 59-60%.
 
Conveniently enough, today there is a Fox Sports article on exactly this!

https://www.foxsports.com.au/afl/af...r/news-story/c20bfe594092e767cfd3b67162092796

The numbers are slightly different because Champion Data can consider all shots at goal, including complete misses, whereas I could only compare the ratio of goals to behinds. The conclusion is the same, though: Geelong and West Coast have been a lot more accurate so far.

Champion Data also can examine Expected Scores, which factors in the difficulty of shots taken. This should rule out the idea that the higher accuracy of the Cats & Eagles is simply because they're getting into better positions and taking easier shots. Which means there is probably a pretty big luck component.

It's also interesting to see Brisbane high on this table, since there's another report out today ranking them as the team least affected by injury this season:

https://www.foxsports.com.au/afl/af...r/news-story/c20bfe594092e767cfd3b67162092796

Together, these tables suggest that Brisbane are operating somewhere near the top of their potential at the moment, and more likely to get worse in the future than better.


One of the things I noticed in Footballistics with the expected goals was that there is a chance Franklin suffers because he probably takes a large portion of all shots from 55m out near the boundary. So by its very nature he reverts to the mean.

It's a pretty fascinating aspect but would prefer them to only use set shots.
 
One of the things I noticed in Footballistics with the expected goals was that there is a chance Franklin suffers because he probably takes a large portion of all shots from 55m out near the boundary. So by its very nature he reverts to the mean.

It's a pretty fascinating aspect but would prefer them to only use set shots.
Yes, and the same effect seems to occur at a team level. It isn't often that a forward finds himself with an easy pass to a teammate in an even better spot, so what mostly happens is that players move the ball around until someone has a 50% or better chance of kicking a goal, then that person takes the shot. If it's an exceptionally skilled player, like Franklin or Betts, they're able to meet that criterion from more difficult chances, and if it's a less skilled player, they need easier chances, but either way, they're all usually looking for that 50%ish chance.

So even though there are quite large gaps between the goalkicking skills of players (and teams), they tend to wind up with similar conversion numbers.
 
Heh, but I reckon North were dominant, too. They kept pumping it inside 50 and their forwards were getting perfectly timed, unhindered leaps at the footy - plenty of which didn't quiiiite come off. Could very easily have been a bigger smashing.
It is a very good example of the statistics not matching the reality on the ground at the time. Because at the end of the day, the criteria that the AFL statistics are created from are supposed to be objective, they have the biases of the compiler built into them.

People are placing far to much emphasis on the supposed numbers, and not enough on what their eyes witnessed. Otherwise every tipping program would get 9/9 every round.
 

(Log in to remove this ad.)

It is a very good example of the statistics not matching the reality on the ground at the time. Because at the end of the day, the criteria that the AFL statistics are created from are supposed to be objective, they have the biases of the compiler built into them.

People are placing far to much emphasis on the supposed numbers, and not enough on what their eyes witnessed. Otherwise every tipping program would get 9/9 every round.
Yes, although I notice the top 3 models tracked by Squiggle (20% of the field) are all currently beating the best "Expert" from the Herald-Sun's Supertipping panel of 30 people.

Ideally, I think, you want to be a human who understands the limits of both mathematical models and your own psychology. Because although models have their limitations, they are clearly better than us at some things, such as managing large amounts of data and rating probability.

If you can take the best from both worlds, you can tip well. In most cases, though - the default case - humans will do worse than models, even humans with a lot of expertise, so I tend trust a model over my own perception unless I can think of a good reason why the model would be wrong.
 
Yes, although I notice the top 3 models tracked by Squiggle (20% of the field) are all currently beating the best "Expert" from the Herald-Sun's Supertipping panel of 30 people.

Ideally, I think, you want to be a human who understands the limits of both mathematical models and your own psychology. Because although models have their limitations, they are clearly better than us at some things, such as managing large amounts of data and rating probability.

If you can take the best from both worlds, you can tip well. In most cases, though - the default case - humans will do worse than models, even humans with a lot of expertise, so I tend trust a model over my own perception unless I can think of a good reason why the model would be wrong.

Reminded me of Roby's Power Rankings back in the day which was a sound model in itself and did okay - until he started altering the results based on his perceived umpiring injustices in games which promptly resulted in it all falling in a heap.

People like to think they know better than these models but none of them have proven this by posting evidence of their superior tipping results.
 
Geelong did punch a long way right and landed ahead of the Giant's fist after the Tiger beating last night. Could be because of the low points tally the Tigers were kept to. But indeed the previous 4-goal ish wins over perceived also-rans like North, Sydney, Dogs and Suns just wouldn't have moved the needle much. The Tigers being a better side i think gave the Cats much more of a movement.
 
What would be the cats rating under the old squiggle where it just counted total points and didn't care about accuracy?

On SM-G960F using BigFooty.com mobile app

I'd be interested in this as well. I completely understand and agree with Final Siren that our accuracy will revert to the mean at some point but I also feel like we're clearly superior to all other sides, at this moment in time, which may also revert to the mean.
 
Yep. Also been interesting that not only have the cats been accurate. But their opponents have been inaccurate.
I'd be interested in this as well. I completely understand and agree with Final Siren that our accuracy will revert to the mean at some point but I also feel like we're clearly superior to all other sides, at this moment in time, which may also revert to the mean.

On SM-G960F using BigFooty.com mobile app
 
I went through squiggles predictions from 2018 .A few teams were overrated by sqiggle over the season a few teams under rated.
If sqiggle predicts team A to beat team B by 12 points and team A wins by 8 points then team A was overrated by 4 points and team B underrated by 4 points.The numerical mean being zero and over the coarse of a season you would expect it to be close to evening out and that was true for most teams except.
-Carlton overrated by 12 points per game
-Gold Coast overrated by 8 points per game.
Both of those sides were bad had bad seasons some big blowouts and squiggle didnt quite get how bad they were.
-Brisbane underrated by 8 points per game
That was understandable they finished with an abnormally high percentage for a side that rarelly won and when they did win they won by a big margain a few times.
-West Coast Eagles underrated by 14 points a game .No explanation for that at all.
The only other notable side being Adelaide overrated by 5 points over the whole season .The 2017 grand finalists who had a poor season overall.
The rest were within a few points either side of the mean.
 
I'd be interested in this as well. I completely understand and agree with Final Siren that our accuracy will revert to the mean at some point but I also feel like we're clearly superior to all other sides, at this moment in time, which may also revert to the mean.
Yep, in Hawthorn's threepeat years 2013-15 they were converting scoring shots at 57.3%, 58.9% and 58.2%. Across the three seasons they kicked at 58.1%, so definitely not impossible for a good team with accurate forwards/game plan that results in quality i50s to stay near 60% over a season or longer.
 

Remove this Banner Ad

Back
Top