Pythagorean exponent - hockey

[Waz]

For those of you that use models for the NHL, what exponent do you use to convert the projected final score to a probability? For baseball I use something that takes into account the run environment for that particular game, but I'm not sure that will work for hockey.

[Justin7]

Do you have a database? If yes, I could give you some suggestions on how to develop one.

[Pokerjoe]

Did you google it, Waz? I did once and came up with a site where a guy had done a lot of math on it. I may have even posted a link here.

[Pokerjoe]

http://www.hockeyanalytics.com/Research_files/Win_Probabilities.pdf

Pretty good starting place.

[Salmon Steak]

http://www.hockeyanalytics.com/Resea...babilities.pdf

Pretty good starting place.

That is good stuff PokerJoe. I will look into it in more depth when I get home from the lab.

[jgilmartin]

Great site that is. Part of me kinda wishes you hadn't posted that

[Poogs]

Is that site still relevant now that there are no more ties?

[TomG]

Run a regression just like everything else in life

[Waz]

http://www.hockeyanalytics.com/Resea...babilities.pdf

Pretty good starting place.

I actually came across this research when I googled it. It seemed fairly sound, but I was surprised how little I found on the web.

[Peregrine Stoop]

I'm surprised how little there is as well. Most Canadians, who are about the only people that watch hockey, are pale nerds stuck in the indoors during 3/4 of the year due to it being too cold. I figured there would be analysis out the Moose's rear on this topic.

[Waz]

I'm surprised how little there is as well. Most Canadians, who are about the only people that watch hockey, are pale nerds stuck in the indoors during 3/4 of the year due to it being too cold. I figured there would be analysis out the Moose's rear on this topic.

I think it might be because there aren't as many stat guys who are also hockey fans. For baseball, it seems like everyone loves statistics and analyzing everything you can think of. For hockey, I think people fall into the mentality that stats are not as important, and it's about who can beat the piss out of there opponent and who can outskate who.

[Peregrine Stoop]

baseball also breaks down into a series of nearly independent trials, while other team sports would require much more difficult stat usage

[Waz]

This is very true.

[Waz]

Another issue I am having involves totals. If I put an average total of 5.87 on a game, and the posted total line for under 6 is -115. How can I figure out what the probability of the game going over 5.5? I started to use Poisson, but realized it won't work because certain games cannot land on 6 due to a 3-3 tie that forces an extra goal in overtime and making the total 7. How can I adjust for this in calculating the probability?

[jetsjets1028]

here we go bruins ere we go

[Pokerjoe]

Great site that is. Part of me kinda wishes you hadn't posted that

Well, it was already public. But I know what you mean.

My rule is basically, if it's out there, I'll talk about it here. But if it's mine alone, I keep it to myself. Fine line we draw sometimes.

[Pokerjoe]

For example--and not to be a dick--I'm not answering Waz's Q about totals, LOL.

[Pokerjoe]

baseball also breaks down into a series of nearly independent trials, while other team sports would require much more difficult stat usage

Independent trials by independent operators at that. Nothing breaks down for stats like baseball.

[jgilmartin]

Well, it was already public. But I know what you mean.

My rule is basically, if it's out there, I'll talk about it here. But if it's mine alone, I keep it to myself. Fine line we draw sometimes.

Yep, that is a sensible approach.