|
XUL
OTD Janitor



Registered: 03/16/05
Posts: 28,261
Loc: America
Last seen: 4 years, 2 months
|
Multiple regression prediction models for NHL (A good stats website?)
#23959481 - 12/26/16 10:37 PM (7 years, 1 month ago) |
|
|
I have been interested in using prediction models for the Penguins and their foes.
What is the best website to collect detailed statistics for every game played this season?
I am trying to find: shots on goal; power play goals; power play opportunities; and so forth.
I am looking for a directory for every game played and every action recorded.

--------------------
TRUMP 2020
Edited by XUL (12/26/16 10:42 PM)
|
Brian Jones
Club 27



Registered: 12/18/12
Posts: 12,342
Loc: attending Snake Church
Last seen: 13 hours, 35 minutes
|
Re: Multiple regression prediction models for NHL (A good stats website?) [Re: XUL]
#23963832 - 12/28/16 08:32 PM (7 years, 1 month ago) |
|
|
I don't see how it would work. In a good offensive possession the team brings it down ice, passes and hopefully gets some shots on goal. These could be measurable. But half of a hockey game is at center ice where a possession can last half a second to two seconds, and teams can trade possessions 10 times in 15 seconds. How do you factor that in?.
-------------------- "The Rolling Stones will break up over Brian Jones' dead body" John Lennon I don't want no commies in my car. No Christians either. The worst thing about corruption is that it works so well,
|
Brian Jones
Club 27



Registered: 12/18/12
Posts: 12,342
Loc: attending Snake Church
Last seen: 13 hours, 35 minutes
|
Re: Multiple regression prediction models for NHL (A good stats website?) [Re: Brian Jones]
#23977989 - 01/03/17 05:01 AM (7 years, 27 days ago) |
|
|
I'm not sure, but you have to simply it I (don't think multiple regression is the way to go). You can do powerplays, you can do penalty killing. you can do over time and shootouts, but how do you quantify regular hockey. The reason we have stupid stats like the corsi is because is because possession is impossible to quantify even if we start coiunting half seconds. I am not unimpressed by your idea, but I think it can only apply to certain situations. IDK Heavy duty stats in NHL in its infancy. Maybe you can advance it
-------------------- "The Rolling Stones will break up over Brian Jones' dead body" John Lennon I don't want no commies in my car. No Christians either. The worst thing about corruption is that it works so well,
|
XUL
OTD Janitor



Registered: 03/16/05
Posts: 28,261
Loc: America
Last seen: 4 years, 2 months
|
Re: Multiple regression prediction models for NHL (A good stats website?) [Re: Brian Jones]
#23978745 - 01/03/17 12:48 PM (7 years, 27 days ago) |
|
|
Quote:
Brian Jones said: I'm not sure, but you have to simply it I (don't think multiple regression is the way to go). You can do powerplays, you can do penalty killing. you can do over time and shootouts, but how do you quantify regular hockey. The reason we have stupid stats like the corsi is because is because possession is impossible to quantify even if we start coiunting half seconds. I am not unimpressed by your idea, but I think it can only apply to certain situations. IDK Heavy duty stats in NHL in its infancy. Maybe you can advance it
You have a good point. I will keep that in mind.
--------------------
TRUMP 2020
|
|