Chasing Europe - Everton vs. History

[ToffeeDan]

Would be great if Suarez can just get a booking in the next two matches...

Mbia of QPR is also on 9 and would miss our game if booked next weekend. Suarez would not miss the Derby (barring any straight red card of course).

 

[RobotCrow]

On a serious note though, how much predictive accuracy do historical results count in something like football with all its season-to-season variables.
Could a statistician of Robotcrows stature (you're like the sexy rockstars of statisticians right?) actually come up with a more-or-less accurate predictor of results (aside from real outside variables)? Do betting companies do something like this or do they thumbsuck? Is stats actually magic?

This isn't meant to be a predictive model, so it's not likely to be very useful for that. There are other models used for predicting sports outcomes, of course, and some are much more accurate than others. One of the best around is a system put together by John Hollinger (formerly) of ESPN; however, it only relates to the NBA.

One of the most difficult aspects of predictive modeling with soccer is simply that due to the existence of draws and the very discrete, low-scoring nature of matches, and the need to rotate squads, the long-run best team very frequently doesn't win any given match. That is, it's fairly easy to build a model based simply on spending that will suggest ManU will win the league and that Norwich will finish near the bottom. However, that same model is virtually useless in predicting whether or not ManU will actually beat Norwich in a hypothetical fixture at Carrow Road in January, because the margin between winning and losing (or drawing) in a one-off competition is razor-thin (hence, domestic cups don't always feature the "best" teams in the finals and semi-finals).

All of that said, I COULD (if I had more time and energy...maybe next season) easily build a model that accounts for additional elements, such as strength of schedule, records against common opponents, recent form, etc... and it would (in the long run) be a more useful tool for prediction, even if it would prove fairly useless on a match-by-match basis.

 

[RobotCrow]

I wanted to thumb you right in your excellent First Post, but GOT won't let me.
Then this one came along too, here's another thumber.

haha. thanks mate!

 

[RobotCrow]

Mbia of QPR is also on 9 and would miss our game if booked next weekend. Suarez would not miss the Derby (barring any straight red card of course).

Right--I must have looked at the schedule screwy or something. I thought we had Liverpool in 3 weeks, but that's obviously not the case. Hopefully Luis can hold off on the ankle-stomping then until it's useful for us.

 

[Philw]

This isn't meant to be a predictive model, so it's not likely to be very useful for that. There are other models used for predicting sports outcomes, of course, and some are much more accurate than others. One of the best around is a system put together by John Hollinger (formerly) of ESPN; however, it only relates to the NBA.

One of the most difficult aspects of predictive modeling with soccer is simply that due to the existence of draws and the very discrete, low-scoring nature of matches, and the need to rotate squads, the long-run best team very frequently doesn't win any given match. That is, it's fairly easy to build a model based simply on spending that will suggest ManU will win the league and that Norwich will finish near the bottom. However, that same model is virtually useless in predicting whether or not ManU will actually beat Norwich in a hypothetical fixture at Carrow Road in January, because the margin between winning and losing (or drawing) in a one-off competition is razor-thin (hence, domestic cups don't always feature the "best" teams in the finals and semi-finals).

All of that said, I COULD (if I had more time and energy...maybe next season) easily build a model that accounts for additional elements, such as strength of schedule, records against common opponents, recent form, etc... and it would (in the long run) be a more useful tool for prediction, even if it would prove fairly useless on a match-by-match basis.

Big NBA fan, his stuff is pretty good. Great thread btw, would Thumb you if i could!

 

[Royal]

It was a lengthy selection process, but I knew what I was getting into and I chose correctly. I was thinking of writing about that process sometime, but I'm afraid some tool will just tell me that choosing a club is done on grass or something.

I'd be interested in hearing about it, especially if it's as interesting as your thread which I really enjoyed.
It's true that football's not a sport you can predict with any kind of accuracy but hey it's still interesting for comparisons sake.

 

[RobotCrow]

I'd be interested in hearing about it, especially if it's as interesting as your thread which I really enjoyed.
It's true that football's not a sport you can predict with any kind of accuracy but hey it's still interesting for comparisons sake.

Yeah, thanks. I'll work on a write-up. I think (though I'm a bit biased) that it's interesting, if only as a commentary on how we come to align ourselves with clubs and organizations that we see as extensions of ourselves in ways--even if it's just perception and not reality.

 

[RobotCrow]

Big NBA fan, his stuff is pretty good. Great thread btw, would Thumb you if i could!

Thanks--btw, what is the deal with Thumbs? Why can/can't people give Thumbs?

 

[Brennan]

Thanks--btw, what is the deal with Thumbs? Why can/can't people give Thumbs?

At the moment only upgraded members can give them, the thumb system is a trial after we got rid of the old rep system.