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 

Does PER underestimate 3pt shooting?
Goto page 1, 2, 3  Next
 
Post new topic   Reply to topic    APBRmetrics Forum Index -> General discussion
View previous topic :: View next topic  
Author Message
gkrndija



Joined: 20 Feb 2005
Posts: 64

PostPosted: Wed Nov 09, 2005 3:45 pm    Post subject: Does PER underestimate 3pt shooting? Reply with quote

This is part of the calculation of PER.

http://www.basketball-reference.com/about/per.html

Code:
    uPER = (1/MP)*
         [ 3P {1}
         + (2/3)*AST {2}
         + (2 - factor*(tmAST/tmFG))*FG {3}
         + (FT*0.5*(1 + (1 - (tmAST/tmFG)) + (2/3)*(tmAST/tmFG))) {4}
         - VOP*TO {5}
         - VOP*DRBP*(FGA - FG) {6}
         - VOP*0.44*(0.44 + (0.56*DRBP))*(FTA - FT) {7}
         + VOP*(1 - DRBP)*(TRB - ORB) {8}
         + VOP*DRBP*ORB {9}
         + VOP*STL {10}
         + VOP*DRBP*BLK {11}
         - (PF*((lgFT/lgPF) - 0.44*(lgFTA/lgPF)*VOP))) {12} ]

Most of the terms in the formula above should be clear, but let me define the less obvious ones:

    factor = (2/3) - (0.5*(lgAST / lgFG)) / (2*(lgFG / lgFT))
    VOP = lgPTS / (lgFGA - lgORB + lgTO + 0.44*lgFTA)
    DRBP = (lgTRB - lgORB) / lgTRB


Now part of the problem I notice with the equation is for example:

Player A shoots 4-12 in 3 pointers for 12 points
Player B shoots 6-12 in 2 pointers for 12 points

According to the equation, Player B will have a higher PER roughly equal to ~7/10ths of an additional steal. But Player A should be higher because they both created the same amount of points, shot the same eFG%, but Player A created 2 more chances for offensive rebounds.

They should both have the same PER after step 3(steps are in {}). But Player A is overcredited getting the same amoung of points.

Step 1 should read:

3P*0.5*(2 - factor*(tmAST/tmFG))

Players appear to get 50% credit for hitting a FG and get another ~30% credit for creating their own shot from step 3. So they get ~80% credit for the FG. The same should apply to the extra point from a 3FG. However, I don't think the remaining ~20% credit should be given to the passer because he already got credit from the FG portion of the equation.

Player A is then even more overpenalized in step 6 for missing more shots, when it should be the other way around.

Step 6 should offset the advantage Player B has in missed shots and should read:


- VOP*DRBP*(1-eFG%)*FGA

an additional step should give Player A more credit for more chances in offensive rebounds.

+ VOP*ORBP*(MissFGA-((1-eFG%)*FGA))

The final equation for uPER

Code:
    uPER = (1/MP)*
         [ 3P*0.5*(2 - factor*(tmAST/tmFG)) {1}
         + (2/3)*AST {2}
         + (2 - factor*(tmAST/tmFG))*FG {3}
         + (FT*0.5*(1 + (1 - (tmAST/tmFG)) + (2/3)*(tmAST/tmFG))) {4}
         - VOP*TO {5}
         - VOP*DRBP*(1-eFG%)*FGA {6}
         + VOP*ORBP*(MissFGA-((1-eFG%)*FGA)) {7}
         - VOP*0.44*(0.44 + (0.56*DRBP))*(FTA - FT) {8}
         + VOP*(1 - DRBP)*(TRB - ORB) {9}
         + VOP*DRBP*ORB {10}
         + VOP*STL {11}
         + VOP*DRBP*BLK {12}
         - (PF*((lgFT/lgPF) - 0.44*(lgFTA/lgPF)*VOP))) {13} ]
Back to top
View user's profile Send private message
John Hollinger



Joined: 14 Feb 2005
Posts: 77

PostPosted: Wed Nov 09, 2005 4:19 pm    Post subject: Reply with quote

I don't know if this has come up already on this board, but I've had discussions about the 3-point quandary with some people here before. It's been five years now and the PER could probably use some fine-tuning, and this is as good place as any to start. There's a few other minor tweaks I''m looking at too -- seeing if there's more detail on assisted baskets, shooting fouls and things of that nature.

While I'm at it, are there any other suggestions from the crowd on improvements?
Back to top
View user's profile Send private message AIM Address
davis21wylie2121



Joined: 13 Oct 2005
Posts: 373
Location: Atlanta, GA

PostPosted: Wed Nov 09, 2005 4:56 pm    Post subject: Reply with quote

While Player A created 2 more ORb opportunities, he also created 2 more DRb opportunities for the opponent by missing those threes. If Player A misses those 2 additional shots and the defense grabs both rebounds, then scores on both possessions, that's a clear disadvantage to Player A's actions vs. Player B's, even though they created the same amount of points. Now, the defense might not necessarily grab both (or any) of the misses, and even if they do, they might not score on both possessions... but we know DRb's are easier to come by than ORb's. To say that Player A has "created" chances for ORb's is true, but the odds are greater that he has just handed over 2 free possessions to the opponent.
Back to top
View user's profile Send private message
Eli W



Joined: 01 Feb 2005
Posts: 286

PostPosted: Wed Nov 09, 2005 5:19 pm    Post subject: Reply with quote

This is a pretty big change, but have you ever considered completely dropping the defensive aspects of PER and just having it be a measure of a player's offensive contributions? When I am describing PER to someone for the first time I struggle to sum up what it measures - I say it combines all the ways a player statistically contributes to a game, but it fails to capture a lot of what happens on defense. You candidly acknowledge its limits regarding defense, but does it make sense to do this and yet still include blocks and steals (and defensive rebounds) in the calculation - especially when you also have a separate defensive PER?
Back to top
View user's profile Send private message Send e-mail Visit poster's website
gkrndija



Joined: 20 Feb 2005
Posts: 64

PostPosted: Wed Nov 09, 2005 8:24 pm    Post subject: Reply with quote

Davis21wylie2121 wrote:
While Player A created 2 more ORb opportunities, he also created 2 more DRb opportunities for the opponent by missing those threes. If Player A misses those 2 additional shots and the defense grabs both rebounds, then scores on both possessions, that's a clear disadvantage to Player A's actions vs. Player B's, even though they created the same amount of points. Now, the defense might not necessarily grab both (or any) of the misses, and even if they do, they might not score on both possessions... but we know DRb's are easier to come by than ORb's. To say that Player A has "created" chances for ORb's is true, but the odds are greater that he has just handed over 2 free possessions to the opponent.


Those opponents are going to get these poessessions regardless of whether Player A's team turn it over, scores or misses a shot. They're not really free poessessions. However, Team A get to save their poessession with more offensive rebounds. It doesn't produce points as much as it saves points. Something that 3 point shots produce more of when their percentages are adjusted to 2 pt. FGs.

Quote:

I don't know if this has come up already on this board, but I've had discussions about the 3-point quandary with some people here before. It's been five years now and the PER could probably use some fine-tuning, and this is as good place as any to start. There's a few other minor tweaks I''m looking at too -- seeing if there's more detail on assisted baskets, shooting fouls and things of that nature.

While I'm at it, are there any other suggestions from the crowd on improvements?


I would suggest modifying the PF part of the equation and adding something like Technical/Flagrant fouls and And1s. Maybe focus more on shooting fouls than other Personal fouls.

Maybe upping the DRBP in the FT part of the equation to around 86%
http://82games.com/comm13.htm

Adding goaltending would help account for poor defence, but I don't know how well the league tracks the stat in the PBPs.
Back to top
View user's profile Send private message
John Hollinger



Joined: 14 Feb 2005
Posts: 77

PostPosted: Thu Nov 10, 2005 9:19 am    Post subject: Reply with quote

Interesting point on the defense, because we've all seen cases where player's blocks and steals numbers are hugely deceiving. I think that's definitely where I want to be with this at some point -- that there's an offensive rating and a defensive rating and an overall rating that is the sum of those two. I'm doing some research on D this year that hopefully moves us closer.

Adjusting the rebound rate on missed free throws makes sense as well.
Back to top
View user's profile Send private message AIM Address
Chicago76



Joined: 06 Nov 2005
Posts: 37

PostPosted: Thu Nov 10, 2005 11:22 pm    Post subject: Reply with quote

gkrndija wrote:

I would suggest modifying the PF part of the equation and adding something like Technical/Flagrant fouls and And1s. Maybe focus more on shooting fouls than other Personal fouls.

Maybe upping the DRBP in the FT part of the equation to around 86%
http://82games.com/comm13.htm

Adding goaltending would help account for poor defence, but I don't know how well the league tracks the stat in the PBPs.


Something else I've been thinking about regarding PER/Hollinger stats in the past couple of years regards RBR and another PER adjustment:

1-Rebounding rate may be marginally understated for players who go to the line often. I don't know if 82games has done a study or not, but the odds of a player getting his own rebound on a missed FT is pretty low. This can slightly understate the rebounding rate of certain players (good rebounders with high FTA like Moses Malone 79-80, Barkley, and Garnett). It is very small, but enough to alter RBR by 0.3-0.5 in extreme cases, when you contrast this with someone like Rodman (low FTA). You could argue guys positioned high on the lane have a disadvantage as well, but in my mind a FTA is a synthetic rebounding opportunity for the FTA. For everyone else in the lane, it simulates game conditions (other than the obvious block out factor for ORB). Better rebounders are stationed closer to the basket and worse rebounders, further away, as typically found in live ball situations. Whether you would include missed FTA for the shooter would depend upon if you were trying to measure: A) pure rebounding ability over the course of play or B) rebounding ability knowing that player X would go to the line often.

2-Technical foul shooters. Having a Hoiberg, Miller, Kerr, Price, etc. on the court has its advantages in these situations. It's a very slight adjustment (especially since illegal D is no longer in effect), but the value attributed to the shooter is too high. The value should be divided between the player drawing the T on the opposing team (if applicable/available) and the shooter. With the shooter being given credit only for the increased odds he has of hitting the shot vs. the next best FT shooter on the floor (or a proxy for the league average 2nd shooter).
Back to top
View user's profile Send private message
Mike G



Joined: 14 Jan 2005
Posts: 1506
Location: Delphi, Indiana

PostPosted: Fri Nov 11, 2005 7:39 am    Post subject: Reply with quote

Chicago76 wrote:


1-Rebounding rate may be marginally understated for players who go to the line often. I don't know if 82games has done a study or not, but the odds of a player getting his own rebound on a missed FT is pretty low. This can slightly understate the rebounding rate of certain players (good rebounders with high FTA like Moses Malone 79-80, Barkley, and Garnett). It is very small, but enough to alter RBR by 0.3-0.5 in extreme cases, ...


How about high FTA players who also miss a large % ?, Rebounds are created which are essentially not available to the FT shooter, except rarely.

Last year, Shaq missed 412 FTA. If he'd missed 300 fewer, the way I calculate his rebound rate, it would deprive the opponent of all those rebounds, and his rate jumps just about 5%. This transforms Shaq from the 10th-best rebounder in the league, to the 7th-best.

If no one has thought of this before, I'm amazed. Less amazing that I never thought of it. (Of course, if Shaq is an 82% FT shooter, he probably doesn't get hacked as often.)
Back to top
View user's profile Send private message Send e-mail
Mike G



Joined: 14 Jan 2005
Posts: 1506
Location: Delphi, Indiana

PostPosted: Fri Nov 11, 2005 8:37 am    Post subject: Upon further review Reply with quote

I guess I was overenthusiastic about this potential refinement. Even if Shaq misses 400 FT, only perhaps 60% of those are 'reboundable' misses (2nd of 2, and and-1). So his RebRt is only depressed by perhaps 3%, and he stands to gain only one place in the rankings of rebounders.

Additionally, there are other uncertainties. If Shaq were a great FT shooter, he might in fact go for more contact, and his FTA does not necessarily fall. And, a stat that punishes him for bad FT% -- even in the seemingly unrelated category of Rebounding -- isn't necessarily inappropriate. He gave up possessions when he missed those FT. Missing FTA is worse than missing FGA, in that they are less reboundable.

Most likely, this has been discussed hereabouts, already.
Back to top
View user's profile Send private message Send e-mail
dlirag



Joined: 30 Dec 2004
Posts: 28

PostPosted: Fri Nov 11, 2005 9:12 am    Post subject: Reply with quote

The discussion in "Value of a steal" suggests that one could justifiably assign more value to a steal than 1 VOP. As mentioned in that thread, its positive effect seems to be slightly bigger than the negative effect of a turnover.
Back to top
View user's profile Send private message
kbche



Joined: 19 Jul 2005
Posts: 51
Location: washington d.c.

PostPosted: Fri Nov 11, 2005 8:35 pm    Post subject: per ratings Reply with quote

Hi John,

When ranking players, a career average of a players' PER values might be better than just using one season. The most valuable players perform well with different teams for long periods of time.

Kimberly
Back to top
View user's profile Send private message Send e-mail
gabefarkas



Joined: 31 Dec 2004
Posts: 879
Location: Durham, NC

PostPosted: Tue Nov 15, 2005 1:27 pm    Post subject: Reply with quote

John Hollinger wrote:
I don't know if this has come up already on this board, but I've had discussions about the 3-point quandary with some people here before. It's been five years now and the PER could probably use some fine-tuning, and this is as good place as any to start. There's a few other minor tweaks I''m looking at too -- seeing if there's more detail on assisted baskets, shooting fouls and things of that nature.

While I'm at it, are there any other suggestions from the crowd on improvements?


One big thought that's been lingering inside me for a while: PER is essentially a per-minute stat (or per 40 or whatever). How about a gross, overall assessment of production, normalized for the overall production of all players?
Back to top
View user's profile Send private message Send e-mail AIM Address
Nikos



Joined: 16 Jan 2005
Posts: 339

PostPosted: Tue Nov 15, 2005 1:55 pm    Post subject: Reply with quote

I agree with Gabe, but I think that the MPG should be factored relative to the standard deviation or the MPG of the player who plays the most. (Sort of a relative team MPG factor). That way if a team runs a platoon system or whatever, they get credit for being a major player in the platton system despite possibly playing 31-32mpg (even though most players on the team play less than 26-27mpg etc....
Back to top
View user's profile Send private message
Mountain



Joined: 13 Mar 2007
Posts: 216

PostPosted: Tue Aug 14, 2007 11:43 am    Post subject: Reply with quote

John Hollinger wrote:
I don't know if this has come up already on this board, but I've had discussions about the 3-point quandary with some people here before. It's been five years now and the PER could probably use some fine-tuning, and this is as good place as any to start. There's a few other minor tweaks I''m looking at too -- seeing if there's more detail on assisted baskets, shooting fouls and things of that nature.

While I'm at it, are there any other suggestions from the crowd on improvements?


Looking for something else, I came across this. Where do things stand regarding possible updating of PER?
Door closed or still could happen?


Last edited by Mountain on Thu Aug 16, 2007 2:12 pm; edited 1 time in total
Back to top
View user's profile Send private message
gabefarkas



Joined: 31 Dec 2004
Posts: 879
Location: Durham, NC

PostPosted: Tue Aug 14, 2007 1:34 pm    Post subject: Reply with quote

With PbP data now, there's the possibility of determining exactly how many fouls are drawn, rather than using that pesky 0.44 multiplier. That might be worth investigating...
Back to top
View user's profile Send private message Send e-mail AIM Address
Display posts from previous:   
Post new topic   Reply to topic    APBRmetrics Forum Index -> General discussion All times are GMT - 5 Hours
Goto page 1, 2, 3  Next
Page 1 of 3

 
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