Unadjusted plus/minus for trios

by sfwarriorcvg on Feb 5, 2008 11:20 PM PST Comment 20 comments

http://www.82games.com/trios0708.htm

via Dwyer's blog on Yahoo!, thought this was cool, taking all different combinations of three players possible from each team's roster (that's a lot of work! Hundreds of possible combos, like POB, Perovic and Wright!!) and finding the trio with the max plus/minus. You'll be surprised who the Warrior's "best" trio is, hint, it's not who you'd assume it'd be.

I like the plus minus stat because I feel it gives more information on how much impact a player has for his team. The efficiency stats are good for comparing players from different teams w/ different schedules, but it's hard to derive a team specific ranking of players. W's fans know Stephen Jackson helps this team win, but PER tells us that he's a middling small forward. Jax's plus/minus isn't overwhelming (+8.3), but his PER is slightly over average (15.2) (PER average is set to 15 for the NBA) From those stats however, we can say that Jax, on average, creates another 8 points for W's on a given night through offense and defense, but compared to other SF on individual stats, is an average player. Anyway, I thought that was an interesting link.

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!

0 recs | Comment 20 comments

Story-email Email Printer Print

Comments

the CRAZIEST stat...

...is our best trio vs. our worst trio, which is a cumulative swing of 242 pts, and is simply a substitution of one player in the trio, the other two being a part of both:

Monta, Jax, Beans +188
Monta, Buke, Beans -54

by 321 IN n OUT on Feb 2, 2008 9:30 AM PST reply actions 0 recs

Buike is terrible

according this. Pretty strange.

by Calamity on Feb 2, 2008 2:34 PM PST up reply actions 0 recs

ditto

if you check his individual +- for the season - its -92, worst on the dubs. thats from the nba.com "lenovo +-" stat page. Ya its a corporate sponsored stat so thats probably should just mean to put less stock in it anyway. Also its the away games which send his +/- down, its +3 for home games. "Hey were on the road and buike just came in lets tank for awhile" Seems odd to me too... but interesting. I think Ive seen arguments here regarding this stat only being significant for large sample sizes, well its 40 or so games in and it'll be interesting to see what happens to it after another 20 or so games. I havent watched as many games as I'd like though and wonder what other people think...

by frankus maximus on Feb 2, 2008 3:10 PM PST up reply actions 0 recs

The trio numbers

Aren't adjusted the same way the other +/- status are, they're just crude. 82games adjusts the player stats for the usual stuff like strength of opposition, minutes played etc. You're right though, this stat is probably less precise than other standardized statistics because it is at the level of entry/exit from games, not minutes. However, after 50 games we should get some idea of what's going on. Buike's PER is below average, but he's improved considerably since last year, turns the ball over less and scores a little more. I think since these are crude stats the interpretation is murky, I just was surprised the top trio wasn't jax monta and baron.

My girlfriend hates that I'm into basketball again, and that I yell at the TV during games.

by sfwarriorcvg on Feb 2, 2008 3:47 PM PST up reply actions 0 recs

Buike played a lot of minutes

during our 0-6 stretch to start the season where we were getting blown out regularly. That might have something to do with it...

It's almost like the Warriors have 6 guys out there... they always have a guy open! - Jon Barry commentating game 3 last year

by gsw4life on Feb 4, 2008 12:08 PM PST up reply actions 0 recs

Buike

I think he has such a low plus/minus stat because he plays too much time at the wrong position. I believe Nellie sticks him at SF next to Baron and Monta too much.

I read some article that was talking about something similar to this. Al Jefferson has an absolutely horrible plus/minus number playing center this year, but a really good rating when he's playing power forward.

by jlagace on Feb 2, 2008 3:45 PM PST reply actions 0 recs

Do you remember

The link to that article? Sounds interesting.

My girlfriend hates that I'm into basketball again, and that I yell at the TV during games.

by sfwarriorcvg on Feb 2, 2008 3:47 PM PST up reply actions 0 recs

nah

can't find the article again, and I wouldn't know where to start looking either.

If you guys are really into the plus/minus thing, here's a nice site that I like to check from time to time, and they update it every couple of days too:

http://basketballvalue.com/topplayers.php?year=2007-2008

Shows even more just how ridiculous the Baron Davis all-star snubbing was.

by jlagace on Feb 2, 2008 4:18 PM PST up reply actions 0 recs

I don't get it...

So Antwan Jamison is the best player in the NBA and KG, Kobe, and CP3 are around 20th? Seems like not the best statistical reference...

by Dubs fan in Boston on Feb 4, 2008 6:49 AM PST up reply actions 0 recs

Some

of this has to do with the fact that Buike logged a lot of minutes when Jackson was gone, when quite frankly everyone around him was playing like crap.

by greala on Feb 2, 2008 6:45 PM PST reply actions 0 recs

Nah

That can't be it because Baron, Harrington, and Biedrins and the crew played alot of minutes in that same 6 game losing streak too, yet their plus/minus ratings have recovered quite nicely.

Also, in the grand scheme of things, that 6 game stretch is relatively insignificant when compared to the 42 total games the Warriors have played up to this point.

A funny nuance that I've noticed about the stat is that alot of rookies and sophomores have really bad ratings, ratings that are far worse than their team's total net plus/minus, and that makes sense.
Kevin Durant posts a -12 plus/minus on a team with a -7 rating. Sure he knows how to put up big gaudy scoring statistics, but he doesn't know how to win in the NBA yet. Al Jefferson has a -18 on a -9 team. Aldridge has a -6 on a +1 team. Azubuike posts a -12 on a +2 team, etc.

Getting great stats on a bad team is easy, just look at JRich's stats during our dark ages. It's no surprise to me that he's posting a -2 rating on a talented yet underachieving Bobcats squad, with Gerald Wallace posting a +8, Felton a +4.

by jlagace on Feb 3, 2008 3:31 AM PST up reply actions 0 recs

Azubuike's +/-

Also, in the grand scheme of things, that 6 game stretch is relatively insignificant when compared to the 42 total games the Warriors have played up to this point.

Actually, Azuibuike was logging some heavy minutes in those games, and not so heavy since, so those 6 games account for 20.5% of his minutes this season. Considering that the Warriors were -66 during those 6 games, the 6 games are very significant for his overall +/-.

It's almost like the Warriors have 6 guys out there... they always have a guy open! - Jon Barry commentating game 3 last year

by gsw4life on Feb 4, 2008 12:17 PM PST up reply actions 0 recs

Overrated Stat

The +- is one of the most misleading stats out there. It is abusrd to think that during the course of the game with 10 players on the court at a time that one player is directly affecting the entire outcome/result of the game. There are so many variables that are outside the control of a single given player. Player A could be shutting down who he is guarding, grabbing rebounds, shooting well, playing hard and having a solid game all around, but his 4 teammates are not playing any defense, are all cold from the field offensively resulting in Player A's team getting blown out while he is on the court. As a result his +- would be horrible thanks to the outside factors of how his 4 other teammates are playing and how the opposing 5 players are playing at a given time, and yet player A would be playing a solid game individually. For example just this past week when GSW played NO, Chris Paul's stats were unbelievalbe and he played an amazingly efficient game: 28 points, 12 assists 12-24 from the field while carrying his team and yet his +- for the game was -16. To compare in that same game Melvin Ely had a +4 while anyone who watched the game saw that he was just terrible: 0-2 from the field with 2 TO's. Did Ely have a better game than Paul? No. To correlate +- to a player's performance during a game, or even a season, is simply wrong and incorrect.

by innelliewetrust1989 on Feb 2, 2008 7:23 PM PST reply actions 0 recs

good post

RIP mAC dRE aka aNDrE mACAsSi aka MuHaMMeD I-E aka pILL cLinToN

by jmefrisco on Feb 2, 2008 10:06 PM PST up reply actions 0 recs

Sample sizes

Sample sizes are extremely important regarding the plus/minus statistic. In your example, you're only looking at a sample size of one game. And Melvin Ely only played 4 minutes in that game, there's no way you can attach any significance to his +4 plus/minus.

With taking a sample size of the whole season, looking a certain player pairs should raise a few red flags. Take New York for example. Jamal Crawford is a starter and he plays alot of his minutes next to starter Eddy Curry. Crawford has a plus/minus of +6, and Curry has a rating of -14. Are you meaning to tell me that Eddy Curry's horrible rating is more attributed to his teammates bad player instead of his own? That's not possible, because Crawford has a good rating.

Once again, you can't just look at one game and make assumptions about a statistic's invalidity like that. That's like saying I can look at the Hornets/Warriors game and say that the Hornets are a sucky team because they got blown out.

Extrapolating from such a small sample size would tell us absurd things like New Orleans regularly gives up 116 points per game, Stephen Jackson is a high percentage shooter, David West usually gets 5 blocks a game, and Chris Paul is a -16 plus/minus type player. Now do you see what kind of category your example falls into?

by jlagace on Feb 3, 2008 3:14 AM PST up reply actions 0 recs

We can take that a step further too...

Statistical analysis can only be done with a LARGE sample size. There are plenty of anomalies and outliers throughout any statistical study, but overall the data will average out to a reasonable approximation of a player's "effect" on a game, or the price of a particular stock, or the average height of a population.

Referencing a small sample size, like single particular game in which there was an anomaly, is not a sound statistical argument against a particular study. Statistics in the NBA and sports in general are by no means perfect, far from it. But don't get in a tizzy because it doesn't work every time, just accept that it works for large sample sizes most of the time. That's really as good as you're going to get with any significant statistical analysis.

by Dubs fan in Boston on Feb 4, 2008 7:08 AM PST up reply actions 0 recs

that is a very good point

However, all I am saying is that it should not be looked upon as a reliable stat to define a player's direct effect on the outcome of the game or a player's ability/how efficient of a season he is having. Even when looking over the course of an entire season guys like rajon rando(+329, 6th highest in the league), kendrick perkins(+236), raja bell(+252), antonio mcdyess(+272) +- stats' are skewed since their teams are doing so well while they are on the court and yet these select players are not having all-star seasons or directly contributing to this many points. Consider also that all of these average to good players have better +- than players like Baron Davis, Tim Duncan, Deron Williams, Carlos Boozer and Chris Bosh, and noone can argue that antonio mcdyess is impacting his team in a more poistive way than carlos boozer, nobody can tell me that Kendrick Perkins is having a better year than Tim Duncan, or that rajon rondo and his +329 are the reason the celtics are leading the eastern conference. If you put Jason Kidd(currently at -150) on the celtics for rajon rondo, many would agree the Celtics would be a dominant team and a much improved team, but according to the +- stats, jason kidd is negatively affecting his team by 150 points this season while rajon rondo is positively affecting his team and is worth 329 points. According to the +- after this trade the Celtics would be losing 479 points which is about 10 points per game. So the celtics wouldn't do this trade? That is absurd! All I am saying is that this particular stat is not a very accurate determinate of a player's value to a team, or ability. I do think that it is a very interesting stat and over the course of a season can highlight some very unique trends, but I think people overrate the stat.

by innelliewetrust1989 on Feb 3, 2008 12:28 PM PST reply actions 0 recs

hm

Having a hard time reading your post, but I think I get the gist of what you're trying to say. I don't know which plus/minus stat you're referring to, it might be the NBA.com Lenovo one. Here's the ones I go by, they're adjusted and pro-rated:

http://basketballvalue.com/teamplayers.php?year=2007-2008&mode=summary&sortnumber=21&sor torder=DESC&team=GSW

And here's how I like to interpret the stat. I like to compare an individual's plus/minus to their team's plus/minus. So for example, Antonio McDyess has a +6 rating on a +8 team, but Chauncey has a +13 on that +8 team. So I'd take that as saying that Chauncey is the MVP of that squad, and McDyess isn't hurting the team but he isn't exactly carrying the team either.

And in the rating in the link I posted above, Jason Kidd posts a +5 rating on a -6 team. I'd say the plus/minus stat is one way to "point the finger" at certain players. In this case, I'd say the Nets suck due to guys like Sean Williams (-7) and Antoine Wright (-3) and Malik Allen (-2) getting heavy rotation.

by jlagace on Feb 3, 2008 11:58 PM PST up reply actions 0 recs

Excellent response...

This is probably why some GMs make stupid decisions: They base them on good statistical measures, but use them improperly.

And again, players who are on the floor a lot WILL match their teams performance as a whole. Look, statistics are by no means a be all, end all. For the +/-, you'll get guys who come in at the end of a blow out when nobody's trying. You'll get guys who come in/out of the game between free throws. It's not perfect, nothing's perfect, but I like this statistic far better than any of the PER ratings if only because it's plain and simple and actually accounts for offense and defense pretty evenly and looks at the overall impact as opposed to individual plays (points, rebounds, steals, blocks, FG%, etc.).

by Dubs fan in Boston on Feb 4, 2008 7:25 AM PST up reply actions 0 recs

look at 5 player combinations

I believe it is a useful stat when looking at player combinations - especially the 5 player combinations. Looking at an individual player's +- can be misleading for many of the reasons pointed out in other posts here.

But looking at the 5 player combinations it's interesting to see which ones have been most effective. In the Warriors case, their top 4 combinations of 5 players include Ellis and Biedrins. Their most effective at this point is Jackson, Harrington, Davis, Biedrins and Ellis.

BTW, I'm talking about the +- on nba.com

- Flazzle

by flazzle on Feb 6, 2008 10:29 PM PST up reply actions 0 recs