Adjusted +/- Player Ratings

by San Francisco Slim on Oct 30, 2007 1:10 PM PDT

This is from 82 Games.

Most of you probably know plus/minus player ratings. That is simply the team's performance when a player is on the floor vs. the team's performance when that player is off the floor. Usually a team is 10 points worse (per 40 minutes) when a superstar is off the floor, as opposed to 10 points better when a terrible player does not play. Everybody else is somewhere in between.

But there are flaws to that system of evaluating players. Namely, no two players are ever playing with and against the same players. So, it's a different context or playing field for every player.

But, what if that context was determined? I.E. each player that any one player plays with and against is rated. So, if Baron Davis is playing with a bunch of crummy players against a really good team, his impact on the team will be judged in that context, as opposed to when he is playing with good players against a really bad team. That seems to be a nightmare of a task to figure out, but that is exactly what the masochists at 82 Games have done.

The result theoretically shows the impact of every player on his team regardless of his opponents and his teammates. The beauty of this rating is that it imcompasses all aspects of play - offense and defense.

Here are the top 20 players according to this system. While the exact rankings may be somewhat surprising, the group of players seems to be intuitively correct. The only surprises are Rondo, Anthony Parker and McDyess. Rondo and Parker have top defensive reputations and McDyess used to be a dominating player who has apparently recovered from injuries more than most people have given him credit.

As you see, Baron is 7th, between Kobe and Dwayne Wade.

Top 20 Players for 2006-2007 Season
Adjusted +/-
(per 40 min) 2006-07

1. Garnett, Kevin     12.35
2. James, LeBron     12.01
3. Duncan, Tim     10.89
4. Arenas, Gilbert     8.80
5. Kidd, Jason     8.74
6. Bryant, Kobe     8.70
7. Davis, Baron     7.72
8. Wade, Dwayne     7.56
9. Pierce, Paul     7.24
10. Ginobili, Manu     6.77
11. Nowitzki, Dirk     6.45
12. Brand, Elton     6.30
13. Artest, Ron     6.25
14. Parker, Anthony     6.13
15. Nash, Steve     6.04
16. Rondo, Rajon     5.59
17. Deng, Luol     5.58
18. Billups, Chauncey     5.42
19. Redd, Michael     5.42
20. McDyess, Antonio     5.07

Here's the link, which also rates all of the other players.

http://www.82games.com/ilardi1.htm

Notice, the Warriors regular rotation featured four above average players (Davis, Biedrins, Jackson and Richardson) two average players (Barnes and Ellis) and two below average players (Harrington and Pietrus, whose detractors will delight in his horrible ranking). Not only does the Warrior's player ratings seem to intuitively jive with the opinions of those of us who followed them closely, but it substaniates their outstanding season ending play when they were all healthy.

Also, notice a lot of surprises with player ratings as you go down the list, Tony Parker, Jefferson, Stoudemire, Carmello Anthony, etc..

This FanPost is a submission from a member of the mighty Golden State of Mind community. While we're all here to throw up that W, these words do not necessarily reflect the views of the GSoM Crew. Still, chances are the preceding post is Unstoppable Baby!

Tweet Comment 10 comments | 0 recs |

Do you like this story?

Comments

context

You have to also look at minutes played on/off and also the backup for that particular player.

Useful stat that like all other stats, needs context.

Questions? Complaints?

(AIM: JetForze; email: Jon.d.ma@gmail.com)

by OptionZero @ Golden State Of Mind on Oct 30, 2007 2:04 PM PDT reply actions

The adjustd +/-

goes a lot farther than the non-adjusted +/-, in my mind. It's still a work in progress, but the idea is that this may end up being a system which ultimately can account for both offense AND defense.

by San Francisco Slim on Oct 30, 2007 2:36 PM PDT up reply actions

adjustment

The adjustment is actually a reasonably good attempt at exactly those factors mentioned and does try to compensate for the time and backup quality issues.

Where it tends to have problems has less to do with the quality of the backups and more to do with teams where starters play most of their minutes together. Teams with shallow rotations and teams that did not suffer much in the way of injuries (e.g. Detroit, where the starters played together for the most part and sat together for the most part) tends to make all their players look more similar.

by jae on Oct 30, 2007 3:13 PM PDT up reply actions

Right

And when players who usually play with the same group, like the Pistons, DO occasionally vary with the players they play with, that variation apparently has a disproportionate effect on their rating.

That's one reason the Warriors rating intrigues me, because Nelson seemed to play them in all sorts of combinations. So, their ratings would theoretically be more meaningful, than let's say the Pistons. And the Warriors' ratings make intuitive sense, particularly in context with their elevated play at the end of the season. Baron having a superstar rating, Biedrins, Richardson and Jackson having pretty good ratings, Barnes and Monta being average, etc..

by San Francisco Slim on Oct 30, 2007 3:42 PM PDT up reply actions

seriously

i was watchin an exhibition game and the announcer was so high on anthony parker that i thought there was some rookie sleeper that i didnt know about. he def is the biggest surprise on this list.

by ktweezee on Oct 30, 2007 2:56 PM PDT reply actions

Harrington and Ellis

Both play out of position and also replace our top +- players.

I know that adjusted +- is supposed to account for who players play with and against, but the fact that those two are often in the game when the warriors have no true point guard (in Ellis' case) or center (TMNT's case) is not something that stats can measure and likely lowers their +/-'s.

by BingBluNT on Oct 30, 2007 3:35 PM PDT reply actions

These are ratings for last season

I think you have to look at them in the context of last season.

Monta actually led the Warriors in minutes played. He played in all sorts of different combinations, often playing big minutes when Baron was on the DL and doing the same thing for J-Rich. Harrington started the season alongside O'Neal in Indiana, playing both forward positions, then morphed into a forward/center with the Warriors, playing mostly as a starter.

by San Francisco Slim on Oct 30, 2007 3:49 PM PDT up reply actions

Warriors:

Davis, Baron     7.72
Biedrins, Andris     2.6
Jackson, Stephen     1.42
Barnes, Matt     0.11
Ellis, Monta     -0.22
Harrington, Al     -2.46
Croshere, Austin     -2.51
Hudson, Troy     -5.59
Pietrus, Mickael     -7.45
Azubuike, Kelenna     -9.62

I'm a little bit surprised by Buike's low low rating.

by jlagace on Oct 31, 2007 7:12 PM PDT reply actions

Not really

when you consider his running mates when he was in games. He spent more time playing along side POB than he did Al Harrington. Nuff said.

THIS IS OOOAKLAND!!!

by Tim N Chris Burger on Nov 1, 2007 11:11 AM PDT up reply actions

Kaz

I'll give him the benefit of the doubt that this has a lot to do with small sample size and the circumstance of his minutes last year. Either way just using the "eye test" I expect to see a much better adj +/- this year. His is really the only one which is counter intuitive, which in my mind lends credibility to the stat being at least somewhat reliable.