I assume the average starter value is probably plus 1 to 1.5 per 100 possessions (or a bit more?) and the average bench player value is -2 to -3 but who has the exact values and for which metrics? What do the average performance / pay ratios look like for each (or for 3 groups if you want early vs bench benches)?
How many player exceed 150% of league average on performance and on pay?
Avg. starter / bench metric values & perf. / pay ratios
Re: Avg. starter / bench metric values & perf. / pay ratios
Maybe it's time to do another starter-sub comparison, regarding the relative value of boxscore stats in these roles.
Here's a list of players who last season started at least 25 games, but no more than 2/3 of their games were starts. That is to say, they all got significant games and minutes as a starter and off the bench.
http://bkref.com/tiny/LPhS4
There were 47 such players; but I tossed out the rookies and those who had changed teams mid-season. Then I stopped at 1200 minutes minimum. So of the remaining 32 players, I find these standout examples -- two players who were hugely better either as a starter, or coming off the bench:I took the liberty of creating a T value, where T = Pts+Reb+Ast+Stl+Blk-TO, in per36 rates.
About 2/3 of the 32 qualifying players had worse overall productivity when starting. Ranked by T(sta) - T(sub):When coming off the bench, players shoot more and only a little better, and get a few more turnovers. Here are the per36 averages, weighted equally for each player.A given player gets 8% more OReb and 7% more blocks when coming off the bench. Also 5% more steals and 12% more FTA.
This is in spite of the assumption that a coach may start a guy when he's playing better.
It seems the difference is about .20 to .25 in plus-minus. That is, a player's BPM (for example) might change by about .22 depending whether he's a starter or a sub.
I'd also estimate a player's WS/48 to be about .008 too high if he's a bench guy, relative to a starter with the same numbers.
Here's a list of players who last season started at least 25 games, but no more than 2/3 of their games were starts. That is to say, they all got significant games and minutes as a starter and off the bench.
http://bkref.com/tiny/LPhS4
There were 47 such players; but I tossed out the rookies and those who had changed teams mid-season. Then I stopped at 1200 minutes minimum. So of the remaining 32 players, I find these standout examples -- two players who were hugely better either as a starter, or coming off the bench:
Code: Select all
Tobias Harris Det per 36 minutes
. G mpg TS% FGA FTA ORb DRb Ast Stl Blk TO PF Pts T
sta 48 32.9 .550 14.4 3.0 .8 4.4 2.0 .8 .7 2.4 1.9 17.3 23.6
sub 34 29.0 .595 15.7 3.5 .9 5.8 2.0 .9 .4 1.2 1.8 20.4 29.3
TJ McConnell Phl per 36 minutes
. G mpg TS% FGA FTA ORb DRb Ast Stl Blk TO PF Pts T
sta 51 30.5 .520 9.1 1.6 .6 3.8 9.2 2.2 .2 2.6 2.1 10.1 23.5
sub 30 19.3 .454 7.5 1.4 .8 3.0 8.5 2.4 .2 3.0 3.0 7.4 19.3
About 2/3 of the 32 qualifying players had worse overall productivity when starting. Ranked by T(sta) - T(sub):
Code: Select all
Worse as starter Better as starter
-5.7 Tobias Harris Det 4.1 TJ McConnell Phl
-4.2 Jon Leuer Mil 2.7 Jameer Nelson Den
-4.0 Nik Stauskas Phl 2.7 Austin Rivers LAC
-3.5 Pau Gasol SAS 2.1 Alex Len Phx
-3.4 CJ Miles Ind 1.8 Josh Richardson Mia
-3.1 Glenn Robinson Ind 1.5 James Ennis Mem
-3.0 Randy Foye Brk 1.3 Noah Vonleh Por
-2.8 Monta Ellis Ind 1.2 Sergio Rodriguez Phl
-2.6 Bismack Biyombo Orl 1.0 Ish Smith Det
-2.4 Dewayne Dedmon SAS .9 Iman Shumpert Cle
-2.1 Trevor Booker Brk .0 Rajon Rondo Chi
-1.8 Wilson Chandler Den
-1.5 Dante Cunningham NOP
-1.5 Gerald Henderson Phl
-1.3 Boris Diaw Uta
-1.3 Al-Farouq Aminu Por
-1.1 Rondae H-Jefferson Brk
-0.8 Kenneth Faried Den
-0.7 Joe Ingles Uta
-0.5 Tim Hardaway Atl
-0.0 Tim Frazier NOP
Code: Select all
x mpg TS% FGA FTA ORb DRb Ast Stl Blk TO PF Pts
Sta 27.9 .537 10.55 2.40 1.41 5.04 3.53 1.07 .61 1.90 2.84 12.46
Sub 21.1 .541 10.94 2.72 1.54 5.21 3.36 1.12 .66 1.94 3.12 13.15
ratio 1.32 .99 .96 .88 .92 .97 1.05 .95 .93 .98 .91 .95
This is in spite of the assumption that a coach may start a guy when he's playing better.
It seems the difference is about .20 to .25 in plus-minus. That is, a player's BPM (for example) might change by about .22 depending whether he's a starter or a sub.
I'd also estimate a player's WS/48 to be about .008 too high if he's a bench guy, relative to a starter with the same numbers.
Re: Avg. starter / bench metric values & perf. / pay ratios
Thanks for the data.