That said, in my adjusted plus/minus work I have found that after accounting for traditional statisitics, minutes per game still helps explain why some players have higher adjusted plus/minus ratings than others.
Can you determine how much of this effect is on offense and how much on defense?
Bill James' theory about his similarity scores was that including playing time helped account for defense, in that players with abnormally low productivity for their minutes tend, on average, to be better defenders. It's a significant part of my rationale for including minutes in my similarity work.
Joined: 03 Jan 2005 Posts: 497 Location: Greensboro, North Carolina
Posted: Tue Feb 08, 2005 11:11 am Post subject:
admin wrote:
Dan Rosenbaum wrote:
That said, in my adjusted plus/minus work I have found that after accounting for traditional statisitics, minutes per game still helps explain why some players have higher adjusted plus/minus ratings than others.
Can you determine how much of this effect is on offense and how much on defense?
Bill James' theory about his similarity scores was that including playing time helped account for defense, in that players with abnormally low productivity for their minutes tend, on average, to be better defenders. It's a significant part of my rationale for including minutes in my similarity work.
Not easily and reliably, at least not yet. But I agree with your hypothesis. I would bet that what this is picking up is defense.
Posted: Tue Feb 08, 2005 2:01 pm Post subject: per minute scores and projections- part two
"this article doesnt present any information about a separate player specific usage/performance expectation / projection number with the Sonics. is their one?"
you didnt answer that question. i guess i'll assume that implies there isnt one unless i hear otherwise.
two different arguments. but if there is only one number, that seems like a shortcoming to me.
Joined: 30 Dec 2004 Posts: 533 Location: Near Philadelphia, PA
Posted: Tue Feb 08, 2005 2:21 pm Post subject: Re: per minute scores and projections- part two
jambalaya wrote:
"this article doesnt present any information about a separate player specific usage/performance expectation / projection number with the Sonics. is their one?"
you didnt answer that question. i guess i'll assume that implies there isnt one unless i hear otherwise.
There's more than 2 numbers. There is established ability, forecasted ability, and uncertainty in that forecast. At least in my work. And when I talk to the brass, they understand that the uncertainty is important even when I don't show them the numbers.
Either way, I know that we all take stats put up by 19-yr old LeBron as much better than the same numbers put up by a 27-yr old vet. Whether you do it qualitatively or quantitatively, there is an understanding of the growth curve. _________________ Dean Oliver
Author, Basketball on Paper
http://www.basketballonpaper.com
Current pts/48 minutes, with no minimum minutes. Please excuse the formatting:
NAME GP MPG PTS/48
1 J. O'Neal, IND 31 36.4 35.1
2 A. Stoudemire, PHO 49 36.4 34.4
3 ZhiZhi Wang, MIA 10 3.2 33.0
4 Allen Iverson, PHI 42 42.0 32.9
5 Dirk Nowitzki, DAL 45 39.5 32.6
6 Luke Jackson, CLE 10 4.3 32.4
7 S. O'Neal, MIA 49 35.1 31.7
8 Kobe Bryant, LAL 34 42.0 31.4
9 G. Arenas, WAS 45 39.9 29.7
10 Paul Pierce, BOS 48 36.3 29.6
11 LeBron James, CLE 44 41.5 29.3
12 Tim Duncan, SAS 48 34.9 29.3
13 Tracy McGrady, HOU 46 42.3 29.3
14 Dwyane Wade, MIA 46 38.7 29.2
15 Ben Gordon, CHI 44 22.7 29.0
16 Ray Allen, SEA 43 39.8 28.9
17 Chris Webber, SAC 39 36.0 28.8
18 Vince Carter, NJN/TOR 43 35.4 28.6
19 Ron Artest, IND 7 41.6 28.4
20 J. Richardson, GSW 39 38.0 28.1
21 Baron Davis, NOR 17 33.8 28.0
22 Michael Redd, MIL 45 38.2 27.9
23 Yao Ming, HOU 48 31.8 27.6
24 C. Maggette, LAC 38 36.7 27.6
25 Steve Francis, ORL 47 38.0 27.5
That's a pretty representative list of the perceived best scorers in the game. There's only two unusual names - Wang and Jackson - and both of these guys have very small samples. There aren't ANY players who play 15, 10, or even 5 minutes a night that project to star status if just given the minutes.
Also, keep in mind that Jackson, the 10th pick, may still prove to be a scorer, and Wang has averaged 23.5 pts/48 for his 1,196 career minutes. _________________ The Best Miami Heat Coverage
http://heat.mostvaluablenetwork.com/
That said, in my adjusted plus/minus work I have found that after accounting for traditional statisitics, minutes per game still helps explain why some players have higher adjusted plus/minus ratings than others.
admin wrote:
Can you determine how much of this effect is on offense and how much on defense?
Dan Rosenbaum wrote:
Not easily and reliably, at least not yet. But I agree with your hypothesis. I would bet that what this is picking up is defense.
I am pretty sure you're right. But I applied a naive regression model, and the results weren't exaclty inspiring.
Code:
MPG = -18.167 + 0.387 * ORTG
Here are the players who play the most MPG beyond their expectations (25 MPG cutoff):
Are you using Dean's offensive ratings? That list seems to be over-punishing low-efficiency, high-volume guys. LeBron James, Tracy McGrady and Kobe Bryant should not be on that list.
Joined: 14 Jan 2005 Posts: 1506 Location: Delphi, Indiana
Posted: Wed Feb 09, 2005 7:47 am Post subject:
A totally out-of-context snippet, followed by a question or 3:
"... Fortson is the quintessential example, but fellow Sixth Man Award contender Antonio Daniels ..."
Can 2 guys from the same team be "Sixth Men"?
Daniels is indeed #6 in minutes for the Sonics; Fortson is 8th-man, by that measure. Fellow reserve Radmanovic plays more than either. If James is the nominal starter at C, then Fortson is the #9 man.
Posted: Wed Feb 09, 2005 3:53 pm Post subject: regarding per minute standard choice
only about a dozen player average 40 minutes a game or within a small fraction of it. that is an average of .4 players at that playing time level per team.
less than 50 players average 36 minutes a game or close enough to it.
that is an average of 1.7 per team.
so league wide 4.6 of the five players with the most playing time on a team fall short of a 40 minute standard and 3.3 of each 5 fall short of a 36 minute standard. those are large majorities falling short of the standard.
a 36 minute standard is better in my mind than 48 or 40 and not bad, but this data makes me inclined to use a 30 minute per minute standard in my own analysis, and just reward the few players who can and do play more with both their extra time and extra production. and not project the rest further beyond what is realistic (if their role changed and got bigger) given this playing time data.
110 players play 30 minute a night. that is 3.75 per team. that is a strong majority of top five minutes played guys.
Posted: Wed Feb 09, 2005 8:53 pm Post subject: stepping up for around 20 to 30 minutes /gm comparison
i prepared this comparison of basically all the guys in the western conference who stepped up from around 20 minutes a night to around 30 minutes at some point in their careers. i found that a lot of the guys playing 30 minutes or more a night started out that way their first year- they are not included. i'd guess between half and two-thirds of the remaining guys getting to the 30 minute a night plateau did it in their second year. in a few cases, this transiton took place over 2 years rather than one. a few other guys who made the journey in a meadering or non-traditional pattern were not included.
last<20min seas min FGA pt reb ast / first>30 min seas min FGA pt reb ast
looking at the data standardized to 30 minutes per game, the changes by stat category are fairly modest, generally less than 10% and mostly up in their first major role averaging around 30 minutes a night compared to a previous season at around 20 minutes (in most cases from one year to the next, but in some cases over a two year period where the minute increase was more gradual).
inside players improved across the board including in the number of shots and points per 30 minutes.
small forwards showed the most improvement overall, though less so on the skill of rebounding.
shooting guards actually declined on rebounding and only gained modestly on assists. they increased in shots and points, with the efficency improving slightly.
point guards declined modestly across the board, using the 30 minute standardization.
i didnt analyze how much further the players improved (or not) in future years after stepping up to around 30 minutes a night, or when they plateaued or declined.
I don't see a pattern there, as far as defense is concerned. I think offensive production is overwhelming defense.
How are you calculating these numbers? They don't seem to pass the "laugh test". Is there anyone who actually thinks that T-Mac should only be playing 23 mpg, AI 22mpg, or Ben Wallace 20 mpg? Out of curiousity if you sort by expMPG, who shows up at the top?
It's in my post somewhere above: MPG = something + something else * Offensive Rating. The underlying idea, a good one, is that players get exactly the playing time they deserve. If we can isolate how much playing time a player's offense is worth, and subtract that from his actual playing time, we'll get a number that shows how much that player's defense is worth.
Unfortunately, it doesn't work. Alas.
NickS wrote:
They don't seem to pass the "laugh test". Is there anyone who actually thinks that T-Mac should only be playing 23 mpg, AI 22mpg, or Ben Wallace 20 mpg?
That's not what that number means. The expected number of minutes is based only on the player's offensive contributions.
Quote:
Out of curiousity if you sort by expMPG, who shows up at the top?
I'm afriad all that data is gone. _________________ ed
It's in my post somewhere above: MPG = something + something else * Offensive Rating. The underlying idea, a good one, is that players get exactly the playing time they deserve. If we can isolate how much playing time a player's offense is worth, and subtract that from his actual playing time, we'll get a number that shows how much that player's defense is worth.
Unfortunately, it doesn't work. Alas.
NickS wrote:
They don't seem to pass the "laugh test". Is there anyone who actually thinks that T-Mac should only be playing 23 mpg, AI 22mpg, or Ben Wallace 20 mpg?
That's not what that number means. The expected number of minutes is based only on the player's offensive contributions.
Thanks for the clarification. I actually mostly figured that out after I posted, but the explanation helps. So the theory is dependent on assuming (1) that players get exactly the playing time they deserve and (2) that we can measure their offensive contributions.
You would think, with all the work that's been done that (2) could, at least be true but as KP pointed out the ORTG that you were using seems to undervalue high output / low efficiency scorers. I know that in BoP DeanO mentions a rule-of-thumb adjustement to increase ORTG credit for players who use more possessions (I don't remember the rule, however). I wonder how the list would change if you made that modification.
All times are GMT - 5 Hours Goto page Previous1, 2
Page 2 of 2
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
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
