Starter stats and sub stats

[Mike G]

Subs are non-starters who 'sub in' off the bench.
Using pages like this one: http://bkref.com/tiny/M8Kn2

In a single season, from 1999-00 to 2010-11, in the regular season, player did not start game, sorted by descending Minutes Played.

I sorted by Games, then copied every player season of fewer than 64 games (as a substitute), from almost 2000 minutes down to 640.
This selection allows for significant games as a starter; and I found 470 players in this century who also had at least 480 minutes as a starter in the season; all at least 14 games.

The theory: A given player is essentially the same player at all points during a given season; and then we can see whether he's better suited to starting or to coming off the bench. Individual teams likely have kept track of these things all along.
But what is typical? Is it relatively easier to score, rebound, etc, when you are off the bench, presumably playing with and against fewer starters? Or are some things easier when your teammates are mostly better than you are?

This study doesn't include any full-time starters, nor any scrubs. These are players who started at least 14 games (and >480 min) and also came off the bench in at least 23 (>640 minutes).

Here are the averages -- totals divided by 470 -- per 36 minutes:

Code: Select all

per36  MPG    FG     FGA     FG%   3fg   3fga   3fg%    FT    FTA     FT%
sta   31.2   5.10   11.35   .449   .95   2.59   .365   2.38   3.11   .764
sub   21.5   5.01   11.41   .439   .99   2.81   .353   2.57   3.41   .754


per36  ORb    DRb    Reb    Ast    Stl   Blk    TO     PF     Pts    total
sta   1.63   4.24   5.87   3.10   1.10   .58   1.94   3.04   13.51   22.23
sub   1.67   4.24   5.92   2.82   1.14   .63   1.97   3.54   13.59   22.12

total is just Pts+Reb+Ast+Stl+Blk-TO
As subs, these players shoot more often but with less accuracy, vs starting; perhaps fueling the increase in OReb (yet not DReb).
Harder to get an Assist and easier to get a foul, when you're not starting.

Note: A similar study of just the current season had some similar results. The difference in OReb% was more exaggerated, as I recall.

[Crow]

I consider those to be small differences. They give the surface impression that with starters or with subs doesn't matter much.

I wonder how the results would vary for sub-groups. Players by 2-4 age bands or former regular starters vs players without that track record.

Hypothetically RAPM could be run for players as "starters" and "subs". In most cases the sample size will greatly favor one or the other but for a few players there might enough data and enough reason to focus on that choice with them (Ginobili, Odom, etc.)

[Mike G]

I suspect many players in this 21-31 minute per game range are likely to play better when given more minutes. Something about the continuity, vs just coming in and expecting to hit your first shot.

Note that FT% is better for starters. That isn't affected by who's on the floor for your team or the other team.

If there's a "minutes advantage" associated with this range of player-minutes, then mathematically removing that may reveal the "sub advantage".
In general, a player who's play is seen to improve may get more minutes; and some have confused this with "More minutes causes better play". But that's backwards.

In the course of being promoted to greater minutes, a player is often designated the starter. And so, if there were no advantage to playing off the bench (in accumulating stats), you'd expect these players as starters to just have better overall stats. But they don't.

Rookies may earn a starting job during their first season, and they create data points that suggest their productions as a starter have come easier. Likewise, the player on his last legs is demoted to a backup role, and there he may well be less productive.

A number of players have several years as a sometime-starter in this list. Some are chronically better off the bench, some should be starting, and some have mixed results. I'll have some examples tomorrow, hopefully.

[Mike G]

It must be tomorrow.
Of 470 player-seasons surveyed, here are some outstanding examples of players whose per36 numbers were much better as starters.

Code: Select all

start >> sub        Pos   yr     mpg    G    Pts   ORb   DRb    PF   Ast   Stl   TO   Blk
sta   Brandon Rush   F   2009   36.3   19   15.8    .5   4.9   2.7   1.1    .7   1.2   .8
sub   Brandon Rush   F   2009   19.9   56   10.0    .8   3.4   2.5   1.4    .8   1.7   .7

sta   Brian Shaw     G   2001   31.5   28   10.6   1.1   5.4   2.9   5.7   1.1   1.8   .5
sub   Brian Shaw     G   2001   18.3   52    6.1    .8   4.7   3.4   4.5    .8   2.0   .5

sta   Jamal Crawford G   2003   32.4   31   17.8    .4   3.2   2.3   6.7   1.2   2.1   .4
sub   Jamal Crawford G   2003   20.2   49   13.2    .3   2.8   2.3   5.4   1.6   2.8   .5

sta   Melvin Ely     C   2005   32.9   17   14.1   3.4   5.2   3.5   1.7    .8   2.0  2.1
sub   Melvin Ely     C   2005   17.6   62   11.8   2.9   3.4   4.8   1.7    .7   2.8  1.2

start >> sub        Pos   yr     mpg    G    Pts   ORb   DRb    PF   Ast   Stl   TO   Blk
sta   Jared Dudley   F   2011   34.1   15   17.2   2.6   3.6   1.9   2.3   2.3   1.3   .4
sub   Jared Dudley   F   2011   24.3   67   13.8   1.6   3.4   2.5   1.7   1.2   1.3   .3

sta   Rodney Rogers  F   2004   33.2   15   16.2   2.5   5.7   3.6   3.9   1.7   2.0   .9
sub   Rodney Rogers  F   2004   16.9   54   12.4   2.6   5.1   5.3   3.3   1.4   2.8   .6

sta   John Salmons   G   2008   38.1   41   16.6   1.2   3.8   2.3   3.3   1.5   2.6   .3
sub   John Salmons   G   2008   23.9   40   11.2    .8   4.0   2.6   2.4   1.1   2.2   .6

sta   Bob Sura       G   2004   38.8   18   15.3   2.6   6.0   2.6   5.5   1.1   2.7   .1
sub   Bob Sura       G   2004   15.6   62   11.2   1.8   4.1   4.0   4.7   1.5   1.9   .4

start >> sub        Pos   yr     mpg    G    Pts   ORb   DRb    PF   Ast   Stl   TO   Blk
sta   Lucious Harris G   2003   32.3   25   16.7   1.5   3.2   1.8   3.1   1.1   1.0   .3
sub   Lucious Harris G   2003   22.4   52   13.0    .9   3.0   1.7   2.7    .9   1.5   .1

sta   Chris Wilcox   F   2006   31.5   24   16.6   3.1   7.3   3.2   1.4    .9   1.6   .6
sub   Chris Wilcox   F   2006   14.6   53   12.9   2.5   6.4   5.4   1.0    .7   1.8  1.0

sta   Andray Blatche F   2010   36.7   36   20.5   2.2   5.8   2.5   3.4   1.4   3.3   .9
sub   Andray Blatche F   2010   20.8   45   15.1   2.9   5.3   4.8   1.7   1.3   2.4  1.5

sta   Stromile Swift F   2003   28.7   26   17.2   3.2   7.2   4.2   1.1   1.4   2.5  2.9
sub   Stromile Swift F   2003   17.9   41   14.3   2.4   6.0   3.2   1.1   1.3   2.3  2.2

start >> sub        Pos   yr     mpg    G    Pts   ORb   DRb    PF   Ast   Stl   TO   Blk
sta   Greg Monroe   F-C  2011   32.7   48   13.2   4.1   5.9   3.1   1.7   1.5   1.2   .8
sub   Greg Monroe   F-C  2011   20.4   32    9.6   3.9   5.3   3.4   1.6   1.5   1.7   .7

sta   Bobby Jackson  G   2003   34.5   26   21.0   1.2   3.5   2.4   4.2   1.6   2.1   .0
sub   Bobby Jackson  G   2003   23.6   33   17.1   1.2   3.5   3.0   3.6   1.4   2.4   .1

sta   Moochie Norris G   2002   37.2   26   12.8   1.1   2.9   1.9   7.2   1.1   2.4   .1
sub   Moochie Norris G   2002   22.9   56    9.0   1.2   2.7   2.2   5.9   1.5   2.6   .1

sta   Ben Gordon     G   2010   37.3   17   20.4    .5   1.8   2.3   3.6   1.0   1.9   .2
sub   Ben Gordon     G   2010   24.4   45   16.3    .6   2.0   3.5   3.4   1.0   2.8   .1

The difference between their (Pts+Reb+Ast+Stl+Blk-TO) sta-minus-sub are between 5.0 and 7.2
Before checking each player's history and circumstance that year, I'm guessing that most are either:
a) youngsters getting promoted to starter by in-year improvement, or
b) being demoted as their careers wind down.

2009 Brandon Rush: Rookie for Ind (36-46), started a few games during the season, and his last 12. Given the green light to shoot, and the Pacers close 8-4.
2001 Brian Shaw: Lakers, age 34, his last full season. With Fisher out for 62 games, he and 37 y.o. Ron Harper trade off.
2003 Jamal Crawford: Chi, age 22, replaces Jay Williams as starter about 2/3 of the season, and holds the job for several years.
2005 Melvin Ely: Cha, age 26. After team falls to 9-32, he replaces Jason Hart in the lineup. This year and next would be his zenith.
etc.

[EvanZ]

It's really interesting that you had Rush there. I've been pushing (psychically, anyway) for Rush to replace Dorell Wright as the starter. Rush is one of those guys that seems to have fallen into the position where even if he is better than the starter, the reason given for him coming off the bench is that he provides "bench scoring" and "energy". That makes sense in a few cases (Terry comes to mind) where defense may be a big liability. But I'd prefer Rush starting for Wright (or Ellis) and Wright (or Ellis) coming off the bench. In my dreams, I guess.