View previous topic :: View next topic |
Author |
Message |
Dan Rosenbaum
Joined: 03 Jan 2005 Posts: 541 Location: Greensboro, North Carolina
|
Posted: Mon Feb 14, 2005 5:46 pm Post subject: |
|
|
jkubatko wrote: | I did a small study using player-seasons from 1978-2004. To be included in the study, a player had to (a) see an increase of at least 50% in minutes per game from one season to the next and (b) play at least 41 games in each season. These criteria gave me 465 player-seasons. In 346 of these seasons (74.41%), the player's PER increased with an increase in playing time. The mean change in PER was 1.55 and the median was 1.58. The range of changes for the middle 50% was -0.05 to 3.18.
Because younger players are more likely to improve (and see an increase in playing time) than older players, I decided to add an age requirement to the criteria above. Now players had to be at least 30 years old in the target season to be inlcuded in the sample. The results were similar: in 42 of the 58 seasons (72.41%) the player's PER increased with an increase in playing time. The mean change in PER was 0.98 and the median was 1.19. The range of changes for the middle 50% was -0.21 to 2.32. |
Very nice work, Justin. Thank you!
If it is easy to do, what are the parallel statistics for decreases in playing time? |
|
Back to top |
|
|
jkubatko
Joined: 05 Jan 2005 Posts: 702 Location: Columbus, OH
|
Posted: Mon Feb 14, 2005 5:53 pm Post subject: |
|
|
Dan Rosenbaum wrote: | If it is easy to do, what are the parallel statistics for decreases in playing time? |
Decrease of at least 50% in minutes per game, no age requirement: 65 out of 91 (71.43%) saw a decrease in PER with a decrease in playing time.
Decrease of at least 50% in minutes per game, at least 30 years old in target season: 35 out of 41 (85.37%) saw a decrease in PER with a decrease in playing time. _________________ Regards,
Justin Kubatko
Basketball-Reference.com |
|
Back to top |
|
|
NickS
Joined: 30 Dec 2004 Posts: 384
|
Posted: Mon Feb 14, 2005 5:55 pm Post subject: |
|
|
Justin, about the study you described, even after you control for age you're going to have a selection bias for players that are recovering from injury. Look at AI this season, his minutes and productivity are up from last season but both can be explained by an increase in health.
You might mitigate this by comparing production in the season in which minutes increased with the average production from the previous 3 seasons.
On a related note, I wonder if there is any way to find people who's PER increased while their minutes decreased.
If one believes that production/minute should decrease as playing time increases it would seem logical that there should be players that would gain in efficiency when they lose minutes.
Obviously most players that lose minutes do so because they aren't playing as well. But there must be examples of players that have a back-up or starter at the same position that has a similar PER allowing the coach to adjust minutes.
An obvious example would be Jamison moving from a starter in GS to a bench player in Dallas but 1 sample does not a study make. Is there any way to isolate criteria to search for these players? |
|
Back to top |
|
|
NickS
Joined: 30 Dec 2004 Posts: 384
|
Posted: Mon Feb 14, 2005 5:58 pm Post subject: |
|
|
jkubatko wrote: | Dan Rosenbaum wrote: | If it is easy to do, what are the parallel statistics for decreases in playing time? |
Decrease of at least 50% in minutes per game, at least 30 years old in target season: 35 out of 41 (85.37%) saw a decrease in PER with a decrease in playing time. |
Shouldn't you make the age limit players under 30 (or 32 if you like)? You want to exclude players who have productivity and minutes declining due to age. |
|
Back to top |
|
|
jkubatko
Joined: 05 Jan 2005 Posts: 702 Location: Columbus, OH
|
Posted: Mon Feb 14, 2005 6:03 pm Post subject: |
|
|
NickS wrote: | Shouldn't you make the age limit players under 30 (or 32 if you like)? You want to exclude players who have productivity and minutes declining due to age. |
Players less than 30 years of age in target season: 30 out of 50 (60%) saw a decrease in PER with a decrease in playing time. _________________ Regards,
Justin Kubatko
Basketball-Reference.com |
|
Back to top |
|
|
jkubatko
Joined: 05 Jan 2005 Posts: 702 Location: Columbus, OH
|
Posted: Mon Feb 14, 2005 6:08 pm Post subject: |
|
|
NickS wrote: | Justin, about the study you described, even after you control for age you're going to have a selection bias for players that are recovering from injury. Look at AI this season, his minutes and productivity are up from last season but both can be explained by an increase in health. |
The players are listed below. Anyone fit that description?
Code: |
-- MP/G -- -- PER ---
Name Year Yr-1 Yr Yr-1 Yr
+----------------------+----+-----+-----+-----+-----+
Greg Anderson 1997 7.5 20.2 7.8 11.4
Ron Anderson 1989 14.8 31.9 16.2 15.4
Greg Anthony 2002 14.8 25.2 12.0 12.6
Stacey Augmon 2001 11.7 17.9 12.2 12.3
Stacey Augmon 2004 12.3 20.5 9.3 10.0
Ron Boone 1980 19.3 29.5 9.3 12.1
Dudley Bradley 1988 13.2 22.1 10.3 11.4
Chucky Brown 1999 15.6 24.8 10.5 11.6
Mark Bryant 1996 13.4 22.4 13.2 13.1
Mark Bryant 1999 15.9 26.8 10.2 12.1
Jud Buechler 1999 8.2 21.1 14.1 12.2
Matt Bullard 2000 10.1 18.3 9.6 11.9
Don Buse 1982 18.9 30.8 11.7 14.8
Adrian Caldwell 1997 6.4 12.6 10.4 7.5
Tyrone Corbin 1997 18.1 32.9 12.6 11.0
Earl Cureton 1989 16.3 25.0 9.2 10.1
Emanual Davis 2001 13.0 20.8 10.2 11.8
Andrew DeClercq 2003 10.4 17.2 10.1 11.3
Greg Dreiling 1994 5.6 12.7 5.0 7.2
Blue Edwards 1996 16.6 33.8 11.6 11.9
James Edwards 1990 16.5 27.8 11.6 13.3
Duane Ferrell 1997 10.9 18.0 10.9 11.6
Vern Fleming 1996 12.5 22.7 14.6 11.3
Rick Fox 2001 18.0 27.9 11.7 13.8
Mike Gale 1982 14.4 23.9 13.7 10.9
Winston Garland 1995 15.2 26.5 12.2 10.1
Rickey Green 1991 13.4 28.5 11.0 13.1
Carl Herrera 1997 8.9 24.5 4.7 10.6
Fred Hoiberg 2004 12.4 22.8 10.3 13.6
Jaren Jackson 1998 15.1 27.1 10.1 9.7
Jim Jackson 2004 20.8 39.0 12.4 12.2
Phil Jackson 1979 10.4 18.1 8.1 10.7
Avery Johnson 2004 9.0 13.8 12.2 11.7
Popeye Jones 2002 14.2 24.3 13.1 14.4
Sam Lacey 1983 12.0 20.5 7.4 6.5
John Long 1990 13.5 21.5 9.5 10.7
John Lucas 1990 11.4 19.1 16.7 11.5
Tony Massenburg 1999 14.7 26.6 14.9 15.9
Sam Mitchell 1996 17.0 27.5 13.4 13.0
Will Perdue 1997 17.5 29.5 13.7 16.7
Wesley Person 2002 21.8 35.8 11.6 16.3
Ed Pinckney 1996 13.5 23.1 10.9 14.3
Kurt Rambis 1989 12.1 29.8 15.0 16.2
J.R. Reid 1999 14.0 25.1 11.3 13.7
Robert Reid 1989 15.8 26.2 10.9 12.3
Dave Robisch 1980 15.2 32.6 11.9 17.2
Sean Rooks 2003 11.9 19.2 7.6 7.7
Brad Sellers 1993 5.3 9.9 13.7 8.3
Purvis Short 1990 17.8 27.0 11.5 13.4
Rory Sparrow 1989 18.0 32.7 8.3 11.7
Bryant Stith 2001 15.4 32.1 13.8 11.7
Damon Stoudamire 2004 22.3 38.0 12.1 14.8
Darnell Valentine 1991 14.1 28.3 11.6 13.5
Loy Vaught 2001 6.8 10.8 8.5 13.3
Clarence Weatherspoon 2001 20.7 33.8 16.2 17.4
Mark West 1995 15.1 23.0 12.2 12.8
Joe Wolf 1996 9.3 16.6 6.8 8.8
Orlando Woolridge 1991 22.9 34.4 17.6 19.5
|
_________________ Regards,
Justin Kubatko
Basketball-Reference.com |
|
Back to top |
|
|
NickS
Joined: 30 Dec 2004 Posts: 384
|
Posted: Mon Feb 14, 2005 6:10 pm Post subject: |
|
|
jkubatko wrote: |
Players less than 30 years of age in target season: 30 out of 50 (60%) saw a decrease in PER with a decrease in playing time. |
With that small a sample size I still wonder if injuries (nagging or otherwise) are a significant factor (compare AK-47 this season to last season, for example).
What if you searched for players that had a drop in playing time of 25% or more, but excluded players who missed more than 10-12 games in the second season?
That won't exclude Elton Brand playing with a knee injury, but it might be a better filter. |
|
Back to top |
|
|
NickS
Joined: 30 Dec 2004 Posts: 384
|
Posted: Mon Feb 14, 2005 6:17 pm Post subject: |
|
|
jkubatko wrote: | NickS wrote: | Justin, about the study you described, even after you control for age you're going to have a selection bias for players that are recovering from injury. Look at AI this season, his minutes and productivity are up from last season but both can be explained by an increase in health. |
The players are listed below. Anyone fit that description?
|
I can think of a few, but not many. That list convinces me that injuries are not playing a major cause. I'm not sure that it specifically proves anything but it is convincing. |
|
Back to top |
|
|
Mike G
Joined: 14 Jan 2005 Posts: 3628 Location: Hendersonville, NC
|
Posted: Tue Feb 15, 2005 10:27 am Post subject: |
|
|
Between '66 and '71, player minutes increased by an average of 8% per player per year. There are 86 instances of players whose minutes increased by 50% from one year to the next, and 42 cases of players' minutes dropping below 2/3 of the previous year.
Of the big minute-losers, 28/42 (.667) also had reduced PAR rates (simple pts+reb+ast, per minute). Their average age was 29.4, while the average age for the 14 other guys (whose minutes were reduced while PAR increased) was 27.5 .
Meanwhile, of the 86 MPG big-gainers, we have 47 whose PAR improved (.547). These win-win guys' average age was 26.2. The 39 players who responded negatively to big minute gains avg age was also 26.2 .
There was about a 1.1% average annual decline in league-wide PAR in this interval. This could be factored into the analysis. However, I still think the enormous dilution of talent that goes with expansion should swamp such a small effect. By 1971, almost half the league was expansion teams. |
|
Back to top |
|
|
NickS
Joined: 30 Dec 2004 Posts: 384
|
Posted: Tue Feb 15, 2005 12:08 pm Post subject: |
|
|
Another query for Justin.
I was thinking about the categories of people that have their minutes decreased or increased by 50% and that tells us something (I'm becoming more and more convinced that players don't, as a rule, lose efficiency by adding minutes) but doesn't really describe manu ginobili.
Can you do a query for all players that played between 26-31 mpg and had a win % over 65%? I'd just be curious to see a list and see if there are any comparisons to be made.
I can think of a couple examples.
Danny Fortson (98-99) had foul problems limit his minutes
Reggie Miller (02-03,03-04) was on the tail end of his career
Michael Redd (02-03) was playing behind ray allen
I don't know why Adrian Dantley (87-8 didn't play more.
Anyway, it would be interesting to see a list of names. |
|
Back to top |
|
|
gabefarkas
Joined: 31 Dec 2004 Posts: 1313 Location: Durham, NC
|
Posted: Tue Feb 15, 2005 3:19 pm Post subject: |
|
|
regarding the PER differences for the increased minutes guys, i'm curious to see if there's a pattern by the amount or % of minutes by which they were increased.
for example, if someone went from 5 mpg to 15 mpg (ie, Mike Sweetney), was their PER increase different than if they went from 10 mpg to 35 mpg (ie, Zach Randolph)? |
|
Back to top |
|
|
jkubatko
Joined: 05 Jan 2005 Posts: 702 Location: Columbus, OH
|
Posted: Tue Feb 15, 2005 4:54 pm Post subject: |
|
|
gabefarkas wrote: | regarding the PER differences for the increased minutes guys, i'm curious to see if there's a pattern by the amount or % of minutes by which they were increased. |
From 1978-2004, there were 5600 cases where a player played at least 41 games in back-to-back seasons. I calcuated the change in minutes played per game from one season to the next and rounded to the nearest minute. In the table below, the first column is the change in minutes played per game from year n-1 to year n, and the second column is the mean change in PER from year n-1 to year n.
Code: |
-22 | 0.58 1 out of 1 (100.0%) improved
-20 | -3.64 0 out of 2 ( 0.0%) improved
-19 | -3.67 0 out of 7 ( 0.0%) improved
-18 | -2.33 0 out of 5 ( 0.0%) improved
-17 | -2.74 2 out of 14 ( 14.3%) improved
-16 | -3.07 1 out of 13 ( 7.7%) improved
-15 | -1.61 6 out of 17 ( 35.3%) improved
-14 | -1.79 9 out of 32 ( 28.1%) improved
-13 | -1.51 10 out of 35 ( 28.6%) improved
-12 | -1.99 9 out of 50 ( 18.0%) improved
-11 | -1.47 19 out of 74 ( 25.7%) improved
-10 | -1.99 16 out of 78 ( 20.5%) improved
-9 | -1.02 32 out of 108 ( 29.6%) improved
-8 | -1.29 40 out of 129 ( 31.0%) improved
-7 | -1.26 42 out of 152 ( 27.6%) improved
-6 | -1.08 61 out of 206 ( 29.6%) improved
-5 | -1.17 72 out of 239 ( 30.1%) improved
-4 | -0.75 101 out of 294 ( 34.4%) improved
-3 | -0.49 129 out of 310 ( 41.6%) improved
-2 | -0.62 159 out of 410 ( 38.8%) improved
-1 | -0.29 198 out of 473 ( 41.9%) improved
0 | -0.12 229 out of 468 ( 48.9%) improved
1 | 0.16 224 out of 418 ( 53.6%) improved
2 | 0.34 222 out of 394 ( 56.3%) improved
3 | 0.37 164 out of 298 ( 55.0%) improved
4 | 0.80 165 out of 265 ( 62.3%) improved
5 | 0.76 153 out of 236 ( 64.8%) improved
6 | 0.79 101 out of 158 ( 63.9%) improved
7 | 0.93 89 out of 134 ( 66.4%) improved
8 | 1.22 79 out of 110 ( 71.8%) improved
9 | 1.17 73 out of 104 ( 70.2%) improved
10 | 1.32 74 out of 95 ( 77.9%) improved
11 | 1.09 36 out of 52 ( 69.2%) improved
12 | 1.63 37 out of 47 ( 78.7%) improved
13 | 1.81 40 out of 52 ( 76.9%) improved
14 | 1.73 17 out of 23 ( 73.9%) improved
15 | 2.50 18 out of 22 ( 81.8%) improved
16 | 2.63 15 out of 18 ( 83.3%) improved
17 | 1.68 9 out of 13 ( 69.2%) improved
18 | 1.79 7 out of 9 ( 77.8%) improved
19 | 1.71 5 out of 7 ( 71.4%) improved
20 | 2.95 11 out of 13 ( 84.6%) improved
21 | 3.13 6 out of 7 ( 85.7%) improved
22 | 3.99 3 out of 3 (100.0%) improved
23 | 3.78 3 out of 3 (100.0%) improved
25 | 3.36 1 out of 1 (100.0%) improved
26 | 4.59 1 out of 1 (100.0%) improved
|
For example, players with an increase of 10 minutes played per game saw an average increase of 1.32 in PER. _________________ Regards,
Justin Kubatko
Basketball-Reference.com |
|
Back to top |
|
|
gabefarkas
Joined: 31 Dec 2004 Posts: 1313 Location: Durham, NC
|
Posted: Tue Feb 15, 2005 10:10 pm Post subject: |
|
|
Justin - is there any chance I can get the raw data for that? the specific per difference for the 1 with -22, for both with -20, for all 7 with -18, etc?
i am curious to run some correlative stuff with the data.... |
|
Back to top |
|
|
Kevin Pelton Site Admin
Joined: 30 Dec 2004 Posts: 979 Location: Seattle
|
Posted: Tue Feb 15, 2005 11:00 pm Post subject: |
|
|
That's pretty remarkably linear. |
|
Back to top |
|
|
HoopStudies
Joined: 30 Dec 2004 Posts: 706 Location: Near Philadelphia, PA
|
Posted: Tue Feb 15, 2005 11:18 pm Post subject: |
|
|
admin wrote: | That's pretty remarkably linear. |
And that actually makes me skeptical.
I think the question of causality in all this is not clear.
If guys get more minutes because they're playing better, that's a vindication of a coach's ability to recognize a guy's ability.
If guys are playing better because they're getting more minutes, why is that? Does that just mean that they don't get used to the system in short time? Shouldn't there be some threshold past which they shouldn't be playing better with more minutes? _________________ Dean Oliver
Author, Basketball on Paper
The postings are my own & don't necess represent positions, strategies or opinions of employers. |
|
Back to top |
|
|
|