Remove this Banner Ad

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

🥰 Love BigFooty? Join now for free.

Man, that Bulldogs movement.

Final Siren is the Flagpole section doing OK? Both the flagpoles and the predicted grand finals don't look quite right.

Sitting over in flag-land yet could still miss finals.

Squiggle needs a Bevo coefficient or something similar to make sense of this.
 

Log in to remove this Banner Ad

Squiggle expected a 5 goal win so Collingwood outperformed expectations.
Sydney were below predicted result.


True but it has the final score as Sydney losing 80 to 59 or something similar
 

Remove this Banner Ad

18th position with a projected 7.1 wins? :think:
Personally I have Richmond as a lock for the spoon with 5 wins (biggest off-season loss of talent since I've been tracking it!), but even that's optimistic, because projected ladders always give too many wins to the bottom team. From the FAQ:

Q. Why is the season tipped to be so close?

Most early-season projections will predict an unusually close year, tipping too few wins for the top team and too many for the wooden spooner. This isn't an error; it's because "How many wins will the top team have, whoever they turn out to be?" is a different question to "How many wins will Geelong have?" — even if we expect Geelong to finish on top.

It's highly likely that the minor premier will turn out to be a team that wins more games than people expected. But we can't know who this will be, since we can't predict who will be better than we predict.

It's the same as tossing a coin 10 times: If I'm predicting a ladder, I'd say heads and tails will each go 5-5, since this is their long-term average. I know that whichever one finishes on top will probably score 6 or more flips (75% chance), but I can't say which one that will be. I have to choose which question I want to get right: the number of wins for each team, or the number of wins of whoever finishes on top. If I choose to answer the first question, then my forecast for every individual team is more likely to be accurate, but the overall distribution of wins and losses will almost certainly be too spread out.
 
Personally I have Richmond as a lock for the spoon with 5 wins (biggest off-season loss of talent since I've been tracking it!), but even that's optimistic, because projected ladders always give too many wins to the bottom team. From the FAQ:

Q. Why is the season tipped to be so close?

Most early-season projections will predict an unusually close year, tipping too few wins for the top team and too many for the wooden spooner. This isn't an error; it's because "How many wins will the top team have, whoever they turn out to be?" is a different question to "How many wins will Geelong have?" — even if we expect Geelong to finish on top.

It's highly likely that the minor premier will turn out to be a team that wins more games than people expected. But we can't know who this will be, since we can't predict who will be better than we predict.

It's the same as tossing a coin 10 times: If I'm predicting a ladder, I'd say heads and tails will each go 5-5, since this is their long-term average. I know that whichever one finishes on top will probably score 6 or more flips (75% chance), but I can't say which one that will be. I have to choose which question I want to get right: the number of wins for each team, or the number of wins of whoever finishes on top. If I choose to answer the first question, then my forecast for every individual team is more likely to be accurate, but the overall distribution of wins and losses will almost certainly be too spread out.

If Richmond are a lock for the spoon with five wins, North fans are in for our best season in half a decade even if we come 17th again. I’ll take it…
 
Personally I have Richmond as a lock for the spoon with 5 wins (biggest off-season loss of talent since I've been tracking it!), but even that's optimistic, because projected ladders always give too many wins to the bottom team. From the FAQ:

Q. Why is the season tipped to be so close?

Most early-season projections will predict an unusually close year, tipping too few wins for the top team and too many for the wooden spooner. This isn't an error; it's because "How many wins will the top team have, whoever they turn out to be?" is a different question to "How many wins will Geelong have?" — even if we expect Geelong to finish on top.

It's highly likely that the minor premier will turn out to be a team that wins more games than people expected. But we can't know who this will be, since we can't predict who will be better than we predict.

It's the same as tossing a coin 10 times: If I'm predicting a ladder, I'd say heads and tails will each go 5-5, since this is their long-term average. I know that whichever one finishes on top will probably score 6 or more flips (75% chance), but I can't say which one that will be. I have to choose which question I want to get right: the number of wins for each team, or the number of wins of whoever finishes on top. If I choose to answer the first question, then my forecast for every individual team is more likely to be accurate, but the overall distribution of wins and losses will almost certainly be too spread out.
Ryan Reynolds Reaction GIF


Why wouldn't a prediction take into account the historical data of past minor premier/wooden spoon total wins though?

Wouldn't that all but guarantee a more accurate projection?
 

🥰 Love BigFooty? Join now for free.

Ryan Reynolds Reaction GIF


Why wouldn't a prediction take into account the historical data of past minor premier/wooden spoon total wins though?

Wouldn't that all but guarantee a more accurate projection?

Because the saints winning 27 wooden spoons in the last 150 years doesn't tell you anything about the likelihood of it happening in 2025.

Past performance doesn't indicate future returns.

Feeding the saints 1897 spoon etc etc into a probability model (even in a minor way) is doing more harm than helping when it comes to accuracy.
 
Because the saints winning 27 wooden spoons in the last 150 years doesn't tell you anything about the likelihood of it happening in 2025.

Past performance doesn't indicate future returns.

Feeding the saints 1897 spoon etc etc into a probability model (even in a minor way) is doing more harm than helping when it comes to accuracy.
Not sure what relevance saints number of spoons has to do with total win averages for the 18th side....

You could almost guarantee this years wooden spooner will end up with closer to the average wins for 18th sides over the last 10 seasons than they will ending up on 7(.1) wins. So why not take that into account?

If you have to clarify that your predictions are always too harsh on the top team and too generous to the bottom team, surely you adjust the algorithm to suit? Doesn't make sense to me.
 
If you have to clarify that your predictions are always too harsh on the top team and too generous to the bottom team, surely you adjust the algorithm to suit? Doesn't make sense to me.

Because the thought that the predictions are always too harsh on top and too generous on bottom teams is just an in progress hypothetical atm.

It looks that way atm, but maybe (with more data)- that hypothetical is incorrect.

If you keep changing your prediction model everytime you see something that looks like something you need to correct for- you will constantly be changing the whole thing.
Any never actually improving accuracy- because you can measure and isolate the effectiveness of each change
 
Because why 10 years?
That's just a pluck based on feels.

Why not 5 years or 1 year or 100 years or 10.1 years or 5.643 years?
10 was just a random point with a decent sample size (plus I couldnt be bothered looking up exactly how many years weve had 18 sides for), and it'd still be more accurate than 7.1 will be.

Go 15 or 20 years if you like, it'd be much the same I'd imagine.
 
Because the thought that the predictions are always too harsh on top and too generous on bottom teams is just an in progress hypothetical atm.

It looks that way atm, but maybe (with more data)- that hypothetical is incorrect.

If you keep changing your prediction model everytime you see something that looks like something you need to correct for- you will constantly be changing the whole thing.
Any never actually improving accuracy- because you can measure and isolate the effectiveness of each change
Ill bet you $100 that the spooner doesn't win 7 games. Or even 6 games. And I'm happy to lock that bet in every year for the next decade. Or 2 decades if you want.

It looks that way, because it is.
 

Remove this Banner Ad

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

🥰 Love BigFooty? Join now for free.

Back
Top