Warden
March 21st, 2004, 09:35 PM
GOALIES: http://www.shawnstewart.com/StatsPrelimG.htm
note: Goalie result aren't as robust because of less sample size but still interesting.
PLAYERS: http://www.shawnstewart.com/StatsPrelim.htm
As some of you know during the day I work as a statistican/economist etc.
Given the current state of the NHL and upcoming CBA expiry I've decided to do some "fun" analysis
I've compiled a database of player performance statistics and 2003/2004 salary data and have begun constructing statistical models to understand the relationship between performance and salary (among other factors such as player demographics, team data etc.)
I'm in the VERY early stages of constructing the regression model but have some early results I thought might be interesting. I'm not presenting the model at the moment but the values of "predicted" salary come from an early incarnation.
For example, for Jaromir Jagr, his current salary is 11 million. Using his performance indicators, team information and demographics the model predicts that he should make around 5.3 million. This model is based on the last 3 years of regular season data for the entire NHL. "Overpaid/Underpaid" is used loosely here because I still haven't accounted for intangible variations based on the player's star status and owner's spending habits etc etc. Some will argue that Forsberg is NOT overpaid but in the frame of performance and salary parity he is.
However you will notice that the validity of the model begins to be confirmed by some of the names that appear (and have been argued about in the past.. ahem.. Holik)
Enjoy. I'll continue to do a full analysis including "What If" scenarios and will present the full statistical methodology in due time.
P.S. Please no statistics questions or suggestions! Feel free to argue over player worth though!
edit:
Here's a quick look at top 10 PIM leaders over the last 3 years.. just to look at non-scorers:
PETER WORRELL,COL,800000 ,1447040.62,-647041 - UNDERPAID
JODY SHELLEY,CLB,600000,624783.3717,-24783.4 - UNDERPAID (barely)
DONALD BRASHEAR,PHI,2250000,3176896.434,-926896 - UNDERPAID
CHRIS NEIL,OTT,575000,134502.1852,440497.8 - UNDERPAID
MATT JOHNSON,MIN,1165000,-314914.299,1479914 - OVERPAID
TIE DOMI,TOR,2000000,2220649.138,-220649 - UNDERPAID (barely)
REED LOW,STL,700000,1049334.131,-349334 - UNDERPAID
KRZYSZTOF OLIWA,CAL,500000,663314.0903,-163314 - UNDERPAID (barely)
CHRIS SIMON,CAL,1500000,1980604.994,-480605 - UNDERPAID
MATTHEW BARNABY,COL,1587500,2644932.022 ,-1057432 - UNDERPAID
As you can see the model doesn't favour pure scorers. It also does a fairly good job of representing non-scorers. In fact Shelley, Domi, Oliwa are almost dead on.
whirled
note: Goalie result aren't as robust because of less sample size but still interesting.
PLAYERS: http://www.shawnstewart.com/StatsPrelim.htm
As some of you know during the day I work as a statistican/economist etc.
Given the current state of the NHL and upcoming CBA expiry I've decided to do some "fun" analysis
I've compiled a database of player performance statistics and 2003/2004 salary data and have begun constructing statistical models to understand the relationship between performance and salary (among other factors such as player demographics, team data etc.)
I'm in the VERY early stages of constructing the regression model but have some early results I thought might be interesting. I'm not presenting the model at the moment but the values of "predicted" salary come from an early incarnation.
For example, for Jaromir Jagr, his current salary is 11 million. Using his performance indicators, team information and demographics the model predicts that he should make around 5.3 million. This model is based on the last 3 years of regular season data for the entire NHL. "Overpaid/Underpaid" is used loosely here because I still haven't accounted for intangible variations based on the player's star status and owner's spending habits etc etc. Some will argue that Forsberg is NOT overpaid but in the frame of performance and salary parity he is.
However you will notice that the validity of the model begins to be confirmed by some of the names that appear (and have been argued about in the past.. ahem.. Holik)
Enjoy. I'll continue to do a full analysis including "What If" scenarios and will present the full statistical methodology in due time.
P.S. Please no statistics questions or suggestions! Feel free to argue over player worth though!
edit:
Here's a quick look at top 10 PIM leaders over the last 3 years.. just to look at non-scorers:
PETER WORRELL,COL,800000 ,1447040.62,-647041 - UNDERPAID
JODY SHELLEY,CLB,600000,624783.3717,-24783.4 - UNDERPAID (barely)
DONALD BRASHEAR,PHI,2250000,3176896.434,-926896 - UNDERPAID
CHRIS NEIL,OTT,575000,134502.1852,440497.8 - UNDERPAID
MATT JOHNSON,MIN,1165000,-314914.299,1479914 - OVERPAID
TIE DOMI,TOR,2000000,2220649.138,-220649 - UNDERPAID (barely)
REED LOW,STL,700000,1049334.131,-349334 - UNDERPAID
KRZYSZTOF OLIWA,CAL,500000,663314.0903,-163314 - UNDERPAID (barely)
CHRIS SIMON,CAL,1500000,1980604.994,-480605 - UNDERPAID
MATTHEW BARNABY,COL,1587500,2644932.022 ,-1057432 - UNDERPAID
As you can see the model doesn't favour pure scorers. It also does a fairly good job of representing non-scorers. In fact Shelley, Domi, Oliwa are almost dead on.
whirled