View Single Post
09-20-2013, 07:06 PM
Boom Boom Geoffrion*
Join Date: Jan 2009
Location: NYC
Country: Greece
Posts: 7,553
vCash: 500
Originally Posted by Blue Blooded View Post
Over a large enough sample, these things even out. One time your partner is nicked up, another one of the opponents are. These are obviously big factors in the particular moment, and we like to think they have as big effect on the whole. But that is just not how it is.

Re. Hitting and winning. There was a study made this summer in lieu of Leafs management talking about what they were considering important factors for winning which were pretty much the opposite views from the hockey analytics community. A Leaf fan took it upon himself to examine which statistics actually correlated with winning. The chosen data set was the past six seasons as most modern stats have only been counted as far back as that.

Road hits (to avoid biased counters) had a correlation with P% of 1.5% and with W% of 0.3%.

Compare that to say 5v5 Close CF% which had a correlation of 58.4% to P% and 55.1% to W%. Just to show the difference in magnitude. 5v5 CF% is also repeatable, which means that a team can sustain it over several seasons.

For example 5v5 GF% is most closely correlated with winning, which is hardly surprising, but it isn't nearly as repeatable as 5v5 CF%. This meaning that GF% is likely a lot more based on random chance than CF%.
I'd like to compare every single players Corsi ratings one day, just for fun. I'm curious to see where the elite, great, good, average, and terrible players rank. The results will probably shock the hell out of us and provide the forum with LOL's for a long time.

Originally Posted by Dactyl View Post
my eyes have told me that if G isnt paired with an elite dman he is crap
You're eyes are deceiving you.

Cruel, cruel eyes. Can't trust 'em.

Boom Boom Geoffrion* is offline