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

[the harry]

The Champion Data model is built upon player values and forecasts injury probabilities to take into account the values of replacement players.

Forecasting injuries would be nearly impossible. In a lot of instances, there would be a correlation across the team so it isn't enough to say "an average of x games are missed through injury per team" and then allocate it across the lists.

So much guess work involved surely makes it counterproductive other than sounding good?

 

[Happy Mastenator]

Forecasting injuries would be nearly impossible. In a lot of instances, there would be a correlation across the team so it isn't enough to say "an average of x games are missed through injury per team" and then allocate it across the lists.

So much guess work involved surely makes it counterproductive other than sounding good?

It is possible to account for known injuries though.

Say start of last year they could move Nic Nat out for the year. That would allow for greater accuracy. Models that didn't account for him missing would take time to catch up and realize performance was down.

Conversely at the start of the year those models that don't account for player injuries wouldn't factor in Nic Nat returning and the increase in output over vardy.

 

[Pessimistic]

So squiggle appears to be forecasting a massive form slump by the eagles

They are 10w 1L but the forecast is five losses in the second half. A slump from almost 90% to barely 50%

This team has a healthy home advantage and has 3 difficult fixtures at most

Gotta admire squiggles ballsy prediction

 

[LeverPuller]

Would love to know the last time the Dees were this close to the premiership zone - 2006?

 

[Hobbes]

Would love to know the last time the Dees were this close to the premiership zone - 2006?

I make it 1994, although 2000 managed a comparable spot (with more attack, less defence)

 

[HTT]

Good to see us return to our rightful position WAYYY over there

 

[Hobbes]

So squiggle appears to be forecasting a massive form slump by the eagles

They are 10w 1L but the forecast is five losses in the second half. A slump from almost 90% to barely 50%

This team has a healthy home advantage and has 3 difficult fixtures at most

Gotta admire squiggles ballsy prediction

They have a lot of wins, but not many impressive ones. They're a good team, but by rights they're probably more like an 8-3 team than a 10-1.

 

[Roos Addick]

Squiggle has North up to 4th!!! Yay!

Then being bundled out in straight sets! [emoji22]

 

[Final Siren]

So squiggle appears to be forecasting a massive form slump by the eagles

They are 10w 1L but the forecast is five losses in the second half. A slump from almost 90% to barely 50%

This team has a healthy home advantage and has 3 difficult fixtures at most

Gotta admire squiggles ballsy prediction

You'd always tip a team to "slump" from 10-1. To go from there to 20-2, you either need a really healthy dose of luck in close games or to be very dominant over the rest of the comp - and while the Eagles are definitely good, they're only 3rd on percentage, with Geelong and North not far behind.

It's also harder for non-Victorian teams to go unbeaten for a long time because more of their games involve significant amounts of home ground advantage, which pushes them towards the middle of the ladder. (They're more likely to win at least some games, but less likely to win all their games.)

This isn't really tipping a form slump; it's tipping a results slump, based on their form remaining about the same, given the fixture to come.

 

[HTT]

Squiggle has North up to 4th!!! Yay!

Then being bundled out in straight sets! [emoji22]

Squiggle thought we'd go out in straight sets last year, right up until we didn't

 

[Forward Press]

If we lost five in the second half of the year it wouldn't surprise. We have Crows, Swans, North, Pies away which we're a good chance to drop. And some tough home games in the run to September.

I don't think we can go 20-2 but of course happy to be proven wrong.

 

[Final Siren]

Champion Data uses a lot of subjectivity for a supposed objective analysis.

I'm pretty sure it's a proper model. And it gets to draw on a far richer data set than anyone else has access to. It's just annoying that you have to take everything it says on faith because they won't tell you anything about its internals.

 

[Final Siren]

If we lost five in the second half of the year it wouldn't surprise. We have Crows, Swans, North, Pies away which we're a good chance to drop. And some tough home games in the run to September.

I don't think we can go 20-2 but of course happy to be proven wrong.

It's tipping 16.45 wins for the Eagles at the moment. You can round that off to 16, but really it's saying 17 wins are equally likely. So 4-5 losses from here.

 

[richoatthedisco]

I'm pretty sure it's a proper model. And it gets to draw on a far richer data set than anyone else has access to. It's just annoying that you have to take everything it says on faith because they won't tell you anything about its internals.

Yeah, I mean more about their player stats: deciding what is and isn't an effective disposal, contested possession, clanger, etc.

 

[Forward Press]

Yeah, I mean more about their player stats: deciding what is and isn't an effective disposal, contested possession, clanger, etc.

I'm not sure how they define 'clanger' (I believe all out on the fulls and free kicks against are counted, includes turnovers but not sure how they define that), but effective disposal and contested possession do have definitions.

Effective disposal is one that ends with a teammate disposing of the ball (thus ensuring the player isn't penalised for the receiver dropping the mark or handballing or kicking into the right space for a teammate to run onto), or a kick of 40m or greater to a 50:50 contest or better (long kicks are fine to an evenly matched contest even if the opponent wins the ball).

I also believe the criteria for contested possession relates to whether the opponent has an opportunity to get the ball but you win it instead. Marks on a lead are not contested possessions. Possessions from free kicks are. The first possession from a stoppage is also always a contested possession.

 

[HTT]

I'm pretty sure it's a proper model. And it gets to draw on a far richer data set than anyone else has access to. It's just annoying that you have to take everything it says on faith because they won't tell you anything about its internals.

Is there any way to access that data, even for a fee? Or do you have to be 'someone'?

 

[Miguel Sanchez]

If we lost five in the second half of the year it wouldn't surprise. We have Crows, Swans, North, Pies away which we're a good chance to drop. And some tough home games in the run to September.

I don't think we can go 20-2 but of course happy to be proven wrong.

If you'd told us at the start of the season we'd finish 16-6 I think our fans would, to a man, be over the ******* moon.

 

[rippersnipper]

I'm not convinced that we'll play finals yet, let alone finish top 4. We still have to play Geelong, west coast, Sydney again, Collingwood, Adelaide, Essendon (who knows which bombers side will show up). None of those are at all easy games and depending on how we play we could either win all of them or lose all of them. I think at best we'll finish on 12 wins.

 

[Matele]

I’m pegging is to finish with 17 wins. So the squiggle is on the money for me.

 

[Roovolution]

You'd always tip a team to "slump" from 10-1. To go from there to 20-2, you either need a really healthy dose of luck in close games or to be very dominant over the rest of the comp - and while the Eagles are definitely good, they're only 3rd on percentage, with Geelong and North not far behind.

It's also harder for non-Victorian teams to go unbeaten for a long time because more of their games involve significant amounts of home ground advantage, which pushes them towards the middle of the ladder. (They're more likely to win at least some games, but less likely to win all their games.)

This isn't really tipping a form slump; it's tipping a results slump, based on their form remaining about the same, given the fixture to come.

Bit of a ‘regression to the mean’ as us gamblers know all to well after a really hot streak comes to an end.

 

[Eastern Rangers]

So squiggle appears to be forecasting a massive form slump by the eagles

They are 10w 1L but the forecast is five losses in the second half. A slump from almost 90% to barely 50%

This team has a healthy home advantage and has 3 difficult fixtures at most

Gotta admire squiggles ballsy prediction

It’s not a prediction it’s just punishing us for burying sides at the moment.


Sent from my iPhone using Tapatalk

 

[Pessimistic]

10-1 to 6-5 is a fair jump with any explanation whatsoever

 

[Seadog]

10-1 to 6-5 is a fair jump with any explanation whatsoever

I'll take it though.

 

[HTT]

10-1 to 6-5 is a fair jump with any explanation whatsoever

It is a bit but the squiggle is paying more attention to margins than the results - look at the fairly close games vs Freo (8pts), Carlton (10), St Kilda (13), Hawthorn (15) and Geelong (15) and asserting that they are probably more like an 8-3 side than a 10-1. 8-3 to 6-5 isn't so unrealistic.
I think this is a bit of a chink in the squiggle's armour, if there is one, I think it underestimates the importance of banking wins a little bit. But, it is based on hell of a lot more data than could possibly fit in my head, so I'll take its word for it

 

[Jiggyman]

They have a lot of wins, but not many impressive ones. They're a good team, but by rights they're probably more like an 8-3 team than a 10-1.

By rights we're a 10-1 team, as we've won 10 games...

7 impressive wins. The others (Blues, Hawks, Saints) were tight contests, and 2 of those were in Melbourne. However, I agree that we are not the 'stand-out' side as there are multiple contenders.

Squiggle takes into account the previous season, so sometimes it's standings have to be taken into context.