This is Google's cache of viewtopic.php?p=22708&sid=4b09c64ce68220d5a474543e96d8989a. It is a snapshot of the page as it appeared on Mar 30, 2011 06:18:10 GMT. The current page could have changed in the meantime. Learn more

Text-only version
These search terms are highlighted: eli w  
APBRmetrics :: View topic - Calculating PER - Yet Again
APBRmetrics Forum Index APBRmetrics
The statistical revolution will not be televised.
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Calculating PER - Yet Again
Goto page Previous  1, 2
 
Post new topic   Reply to topic    APBRmetrics Forum Index -> General discussion
View previous topic :: View next topic  
Author Message
KnickerBlogger



Joined: 30 Dec 2004
Posts: 180

PostPosted: Tue Sep 30, 2008 12:34 pm    Post subject: Reply with quote

Eli W wrote:
One issue could be something that's come up in past years - in calculating lgPER, team turnovers should NOT be included in the league turnover total. Instead of 34719 (lgTO with team turnovers included), try using 33283 (lgTO without team turnovers included). The pace calculations should still include team turnovers. The lgPER that I'm getting is 0.2909676318.


With TO set to 33283, my lgPER is 0.290694586 (a difference of -0.0002730458). However now the player's values for PER is lower than B-R:

Joe Johnson,17.12
Josh Smith,18.93
Marvin Williams,14.38

I also tried dividing my turnovers for both the league PER calculation and the team by 1.04314515 (34719 / 33283) and the values are closer, but still off.

Joe Johnson,17.23
Josh Smith,19.04
Marvin Williams,14.47

Interesting thing is that Joe Johnson is low by at least .02 (rounding) but if you round Josh Smith up that much - he'll be too high.
_________________
KnickerBlogger.Net - now indispensable!
Back to top
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger
Eli W



Joined: 01 Feb 2005
Posts: 402

PostPosted: Tue Sep 30, 2008 2:03 pm    Post subject: Reply with quote

KnickerBlogger wrote:
With TO set to 33283, my lgPER is 0.290694586 (a difference of -0.0002730458). However now the player's values for PER is lower than B-R:

Joe Johnson,17.12
Josh Smith,18.93
Marvin Williams,14.38


I don't see how you're getting those numbers. Using all the values you listed I'm getting 17.27808644 for Johnson. That's from summing your one through eleven and 3PM (to get 1105.421671), dividing by minutes (to get 0.330667565), multiplying by lgPace (94.7996849), dividing by team pace (93.61752707), multiplying by 15, and dividing by lgPER (0.290694586).
Back to top
View user's profile Send private message Send e-mail Visit poster's website
KnickerBlogger



Joined: 30 Dec 2004
Posts: 180

PostPosted: Tue Sep 30, 2008 3:04 pm    Post subject: Reply with quote

Eli W wrote:
KnickerBlogger wrote:
With TO set to 33283, my lgPER is 0.290694586 (a difference of -0.0002730458). However now the player's values for PER is lower than B-R:

Joe Johnson,17.12
Josh Smith,18.93
Marvin Williams,14.38


I don't see how you're getting those numbers. Using all the values you listed I'm getting 17.27808644 for Johnson. That's from summing your one through eleven and 3PM (to get 1105.421671), dividing by minutes (to get 0.330667565), multiplying by lgPace (94.7996849), dividing by team pace (93.61752707), multiplying by 15, and dividing by lgPER (0.290694586).


Yeah but now my one-eleven changed because I changed the # of turnovers in the league calculations. Hence why my PER for Johnson has changed.

Looking at Justin's comments on his values for the league averages, I've tried to modify my league averages to be closer to his:

Justin
lgPace 94.798877
lgVOP 1.0475
lgFact 0.592886798
lgPER 0.290967632

My Original
lgPace 94.7996849
lgVOP 1.047480891
lgFact 0.592886798
lgPER 0.288


Changing my TO -> 33283
lgPace 94.21958054
lgVOP 1.053930169
lgFact 0.592886798
lgPER 0.290694586

So in this case lgPER is correct, but Pace & VOP is out of whack. So going back if I combine the two (and adjust for league turnovers after calculating the first three...)

lgPace: 94.7996849
lgVOP 1.047480891
lgFact 0.592886798
lgPER 0.290964438

Now to calculate:

Joe Johnson: 17.37 - 0.290964438165423
Joe Johnson,17.37
multPER = (lgPace/tmPace)*15/lgPER*(1/MIN)
lgPace: 94.7996849015317
tmpace: 93.0310833364483
VOP: 1.04748089136038
factor: 0.59288679849619

1.57142546081918E-02 = (94.7996849015317/93.0310833364483)*15/0.290964438165423*
(1/3343)

one: 316
two: 1061.39147678549
three: 284.063641456583
four: -233.588238773365
five: -652.825721932941
six: -24.6982022848142
seven: 79.3639789946358
eight: 63.7465116710989
nine: 87.9883948742719
ten: 13.8245446997564
eleven: -58.6997998470369


Still a tad high.
_________________
KnickerBlogger.Net - now indispensable!
Back to top
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger
Eli W



Joined: 01 Feb 2005
Posts: 402

PostPosted: Tue Sep 30, 2008 3:11 pm    Post subject: Reply with quote

Sorry if I wasn't clear, but team turnovers should be excluded only for the calculation of lgPER (the calculation of VOP and lgPace should still include team turnovers). This shouldn't affect your one through eleven values.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
KnickerBlogger



Joined: 30 Dec 2004
Posts: 180

PostPosted: Tue Sep 30, 2008 3:29 pm    Post subject: Reply with quote

So now my one + two + ... + eleven + 3PM = 1105.56658564368
which is a tad higher than your 1105.421671

Dividing by minutes = 0.330710914042381
which is a tad higher than your 0.330667565

hmmmm......
_________________
KnickerBlogger.Net - now indispensable!
Back to top
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger
KnickerBlogger



Joined: 30 Dec 2004
Posts: 180

PostPosted: Tue Sep 30, 2008 3:37 pm    Post subject: Reply with quote

OK looking at my original calcs & the most recent ones:

two +0.2
four -0.47
five +1.0
six -0.3
seven +0.25
eight +0.1
nine +0.18
eleven +0.14

(the others are either very close or identical)

Back to being stumped...
_________________
KnickerBlogger.Net - now indispensable!
Back to top
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger
Eli W



Joined: 01 Feb 2005
Posts: 402

PostPosted: Tue Sep 30, 2008 3:38 pm    Post subject: Reply with quote

I just used your values from this post:

viewtopic.php?p=22694#22694

It didn't have eleven listed, so I used the value you listed in the first post of the thread, which is probably why things are off slightly.

Other than that it looks like you should match up with B-R now.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
KnickerBlogger



Joined: 30 Dec 2004
Posts: 180

PostPosted: Tue Sep 30, 2008 3:42 pm    Post subject: Reply with quote

Got it (I think). Was still using the Hollinger version for the team pace (and the Oliver one for the lgPace).

THANKS!

Joe Johnson: 17.26 - 0.290964438165423
Joe Johnson,17.26
multPER = (lgPace/tmPace)*15/lgPER*(1/MIN)
lgPace: 94.7996849015317
tmpace: 93.64
VOP: 1.04748089136038
factor: 0.59288679849619

1.56120688810856E-02 = (94.7996849015317/93.64)*15/0.290964438165423*(1/3343)

one: 316
two: 1061.39147678549
three: 284.063641456583
four: -233.588238773365
five: -652.825721932941
six: -24.6982022848142
seven: 79.3639789946358
eight: 63.7465116710989
nine: 87.9883948742719
ten: 13.8245446997564
eleven: -58.6997998470369
_________________
KnickerBlogger.Net - now indispensable!
Back to top
View user's profile Send private message Visit poster's website AIM Address Yahoo Messenger
Eli W



Joined: 01 Feb 2005
Posts: 402

PostPosted: Tue Sep 30, 2008 5:11 pm    Post subject: Reply with quote

It's not useless, but I agree that anything beyond the first decimal probably isn't adding anything.

Actually, if anyone is looking for a project, figuring out standard errors/confidence intervals for PER could be interesting.
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Display posts from previous:   
Post new topic   Reply to topic    APBRmetrics Forum Index -> General discussion All times are GMT - 5 Hours
Goto page Previous  1, 2
Page 2 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group