Hey guys, JamieBennKenobi here.
Just stumbled upon an old stat project I did in college and thought I'd share with you guys! This was done in 2009 before I really knew anything about CORSI and the like so it is certainly limited in its usefulness, but maybe ya'll will find it interesting. I've pasted the abstract below and linked to the PDF. Let me know what you think!
ABSTRACT: Of all major American sports, hockey is the least statistically analyzed. Using
data about a team’s scoring, special teams, shots, and goaltending, the following report
seeks to determine what the most significant, yet non-obvious, predictors of regular
season points in the NHL are after the lockout in 2004. Eight variables were selected as
potential predictor variables. A multiple linear regression analysis shows that five of
these variables are significant: power play goals (ppg), shots on goal per game (sogg),
shots against on goal per game (sagg), penalty minutes (pim), and save percentage
(savep). Save percentage and power play goals were found to be the most statistically
significant predictors (P<0.001) while the remaining three variables were marginally
predictive. Power play goals, shots per game, and save percentage positively associated
with higher point totals while shots against on goal per game and penalty minutes are
negatively associated with point totals. The results suggest that the most important
factors in winning in the post lockout NHL are goaltending and special teams, while
general control of possession (shots on goal) is still vital.