[gwnn]

10 more days or so.

[helene_t]

Why is Netherlands refered to as "someone else"?

[whereagles]

helene_t, on May 28 2008, 12:45 PM, said:

Why is Netherlands refered to as "someone else"?

Well, duh! they're outsiders (ok, poll choice limitations)

Winner: POR

[han]

My vote gave my choice the plurality.

[TimG]

helene_t, on May 28 2008, 07:45 AM, said:

Why is Netherlands refered to as "someone else"?

Norway and Netherlands got about 25% of the vote when this poll appeared in "offline bridge".

[helene_t]

Yes but unfortunately (?) Bertens/Bakkeren don't play footie.

[han]

Do you know this for a fact Helene?

[whereagles]

yes. their beer belly doesn't let them see the ball when it's on the floor

[Aberlour10]

someone else (cryptic) = Oranje boven!

[helene_t]

han, on May 28 2008, 03:25 PM, said:

Do you know this for a fact Helene?

Oeps, http://www.do.nl/nieuwsklok/w0506.htm

But as for Huub Bertens I could find no footie related links.

[helene_t]

Btw, I was thinking about footie yesterday. My B.Sc. (math) supervisor wrote a book about the Poisson process, in which one of his examples was the 74 world champship. He claimed that the number of goals per match followed a Poisson distribution which he thought had some negative consequences for the seriousness of the game. My Ph.D. supervisor (who used to play a lot of footie) disagreed strongly with my B.Sc. supervisor.

Since one of the main themes for my own research is random effect GLMs with a Poisson link, I would like to settle the issue by applying something similar to my modeling of last years Bermuda Bowl to footie. I have been somewhat pesimistic with respect to finding suitable data. But yesterday while I was looking at the advertisements of a bookmaker shop it struck me that it might be possible to use the odds to infer the stochastic model which the bookmakers subconsciously rely on. (Of course there is the danger that the bookmakers might use some statistical tool for computing the odds, in which case I would be reverse-engineering that tool which would not be so interesting).

[gwnn]

I know that "equal matches" are usually 2.5-3-2.5 in terms of odds at bookmakers. You can just say that they are intelligent and they have a perfect probability model - if they say Germany-Poland 1.20-2.4-3.6 (just inventing 3 numbers) then it's probably right to accept that Germany are twice as likely to beat then than draw them and thrice as likely to win than lose. Of course you probably knew all this already, or possibly even said these things in your post, but I can't read scientific texts.

[Gerben42]

This poll is too early. There are too many variables to consider. I'll just say "Netherlands" but of the 16, at least 8 teams could win and it wouldn't be a big surprise.

[helene_t]

gwnn, on May 30 2008, 06:24 AM, said:

if they say Germany-Poland 1.20-2.4-3.6 (just inventing 3 numbers) then it's probably right to accept that Germany are twice as likely to beat then than draw them and thrice as likely to win than lose.

Yes, agree. But the bookmakers also offer odds for Germany winning by 2-1, Poland by 1-0 etc.

Now my idea is this: suppose Germany's number of goals is Poisson distributed with parameter lambdaG and Polands with lambdaP. Then the probability of that Germany winns by 2-1 is Poisson(2¦lambdaG) Poisson(1¦lambdaP). Since there are only two parameters in the model and more than two data points (say the bookmakers offer odds for some 8 different outcomes for each macth) then I could test the hypothesis that the bookmakers implicitly use my model. Alternatively they may use some other model, say a negative binomial model or a bivariate Poisson model (with correlation between the number of P-goals and G-goals, say if the wind blows heavily towards Poland's goal it increases the number of G-goals and decreases the number of P-goals. OTOH if the referee is obsessed with off-side cancellations of goals it will reduce both).

Next step is to find out how they estimate lambdaP and lamdaG. Say we know then for this particular match and also for Germany-BurkinoFaso. Is it possible on the basis of those data points to predict the lambdas for Poland-BurkinoFaso?

[Rossoneri]

Then what about bogey teams?

[gwnn]

What's with all the injuries?

[mike777]

Why are the Dutch and The English horrible, really horrible at penalty shots? This has been going on for years now.

Hopefully they have trained and improved in this area.

[Aberlour10]

mike777, on Jun 6 2008, 02:37 PM, said:

Why are the Dutch and The English horrible, really horrible at penalty shots? This has been going on for years now.

Hopefully they have trained and improved in this area.

Team England dont need it at the moment, they did not qualify for the Europeans.
It's strange, english club teams dominate Europe but the national team plays total crap.

[1eyedjack]

Aberlour10, on Jun 6 2008, 09:08 PM, said:

It's strange, english club teams dominate Europe but the national team plays total crap.

maybe the english club teams are just full of brazilians, italians <insert country> etc?

[gwnn]

1eyedjack, on Jun 6 2008, 10:45 PM, said:

Aberlour10, on Jun 6 2008, 09:08 PM, said:

It's strange, english club teams dominate Europe but the national team plays total crap.

maybe the english club teams are just full of brazilians, italians <insert country> etc?

Of course that's true to some extent, but they do have top notch players up there, especially ManU and Chelsea. They have at least one top10 player in the world for every position on the field (except of course goalies). They can be very smug though - auto proclaiming themselves as top favorites before every tournament they qualify to.