07-08 Warriors: +/- Analysis of our 0-5 Start

by jlagace on Nov 10, 2007 7:32 PM PST

from basketballvalue.com:

OnCourt OffCourt WeightedDiff(per48)
Baron 473-501 66-97 +30
Ellis 422-454 117-144 +10
Barnes 210-220 329-378 +9
Buke 400-437 139-161 +4
POB 68-72 471-526 +4
Biedrins 409-458 130-140 -1
THud 27-33 512-565 -3
Belinelli 104-126 435-472 -10
Harrington 359-409 180-189 -11
Pietrus 208-253 331-345 -14

(our other players haven't played enough significant minutes to be included.)

What this tells me:

Baron - As we've seen by his plethora of minutes played, we're still not even a winning team when he is on the floor, but we're truly a disgusting team when he's off the floor.

Monta - same goes for Monta

Barnes - despite his slow start, airballs and occasional turnover this season, our team has statistically done its best when he's been on the court. He really is the glue guy after all, he needs more PT.

PF/C - It shows that we haven't been that bad of a team with Biedrins off the court, which echoes my previous sentiments that he isn't the cornerstone to this franchise that alot of people believe him to be. POB has a pleasantly surprising positive weighted rating, however he clearly hasn't spent enough time on the floor to make this stat that reassuring. Further supporting what alot have people have said already, POB should be the first guy off the bench and get more than 5mpg. He's not a rookie anymore, and yet Nelson hasn't even given him the opportunity to fail or succeed.

Belinelli - He's gotten a surprising amount of PT thus far and has not produced. When he's on the floor it feels like he's not even part of the offense. Either he's not working hard enough to get the ball, or his teammates don't have enough faith in him yet to make a good play. Either way, I'm not worried. He'll produce more and more as the season progresses.

Pietrus - Starting Pietrus at PF for those games has played out to be a big big mistake. Despite his efficiency, the Warriors have been a real travesty when Pietrus is on the court. Granted, Nelly didn't have too many options. In hindsight, Harrington/Barnes at PF and Biedrins/POB at C seems like a more successful rotation.

I would still really like to see how Lasme's shot altering defensive presence could contribute to our lackluster defense.

I know alot of these comments are mere reproductions of the whiny diaries posted everyday. I just thought it was pretty cool to look at it from a purely unbiased statistical perspective and wanted to share.

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 33 comments | 0 recs |

Do you like this story?

Comments

interesting stuff..

...but you can't trust it entirely. According to this, we'd be better off keeping Monta in the game and scaling back Beans and (especially) Harrington.

While I (usually) like what Barnes brings to the game, he hasn't been starting either. So his good numbers here could reflect that he's coming in when the opposing side starts benching their starters.

In Beans's case, we know he's coming in against the first string, and we definitely need his presence in the lineup for significant minutes. Way too often, he's the only big guy in a lineup where the other "big" is Harrington, whose career average is maybe 6 boards in a full night's box score. Blaming Beans for his so-so +- does him a disservice, especially because I'll bet that those numbers are still better than the team's +- on the whole.

OTOH, the numbers on Baron back up what we see: he's absolutely essential to the team, and there's no way we can think about cutting his minutes below 40 a night unless the rest of the team starts picking up slack or his body just won't hold up.

Same goes for Pietrus; it's not hard to conceive that he's a lead weight for the team. Not too many options when we're running 6-deep these days.

Harrington, I'm not sure what to make of those numbers. I liked his game against Dallas, and he's the only thing resembling a game-ready PF on our roster. Hoping those numbers improve when SJax returns. Course that applies to the whole team.

Anyways, interesting stuff. Thanks for posting it!

by sarchasmic on Nov 10, 2007 6:25 AM PST reply actions

wurd

very interesting stuff

by Bill Curley 0wns j00 on Nov 10, 2007 1:48 PM PST reply actions

sample sizes

You're drawing false conclusions from the stats.

Minutes played is sorely overlooked when assigning value to the +/- split.

5 Games is also virtually worthless as a sample size.

And of course, the last part is that we've played zone, which makes it hard to pin any defensive woes on any one player.

Questions? Complaints?

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

by OptionZero @ Golden State Of Mind on Nov 10, 2007 2:19 PM PST reply actions

thats what i was thinking

also it definately depends which teams we played against. Pietrus played a lot against Utah, utah was going to kill us regaurdless, so pietrus's stuff declines.

by Proof on Nov 10, 2007 7:47 PM PST up reply actions

yup....

you gotta take +- stats with a grain of salt. There are just too many variables that are not accounted for, plus with just five games it tells you pretty much diddly-squat. I don't even trust adjusted +- stats completely (although they do have Baron at number 7 for last year!).

I believe that Barnes is an important component to this team, but anyone that's watched the games will tell you that he hasn't been playing very well. His good +- stat has more to do with random luck due to the few times he's been on the court. He just HAPPENED to be on the court when the dubz were on a roll. They didn't score more points because of him.

Also, there's no argument here about Baron being the core of this team, but when your backup PG is going to be THUD or Monta, of course you're going to have a high +-.

by kinetic on Nov 11, 2007 1:21 PM PST up reply actions

hmm

You're saying they didn't score more points because of him, but how can you really prove that? Sure, he wasn't the one scoring all the points, he wasn't the one getting all the steals, assists, and blocks; but there had to be some reason why the Dubs did better with him on the court. Maybe his hustle uplifted the play of his teammates? Maybe his oncourt leadership and encouragement? Maybe the mere winning familiarity of having him on the floor just like how it was back when the Warriors were winning? It's not like his sample size is THAT insignificant either, he's been averaging 17mpg.

Also, you're using an argument about Baron to discredit the EXACT same conclusion that the +/- statistic makes. So I don't get what you're trying to say?

by jlagace on Nov 11, 2007 6:04 PM PST up reply actions

How can you prove that?

Um, dude, the burden of proof is on you, since you're the one who presented all the numbers and leapt to a bunch of conclusions about them.

Of course some of the results of the numbers jibe with reality -- that's the way small sample sizes work. One doesn't need to disprove all your conclusions to show that your methodology is flawed.

Get back to us in another 20-30 games. I feel pretty confident saying that a lot of those #s you posted will be radically different.

by Sleepy Freud on Nov 11, 2007 6:22 PM PST up reply actions

No the numbers won't be different

The statistics I've posted for the first five games will always be the same statistics for those five games.

Again, all I wanted to do was analyze our start to the season for the sake of having a discussion. Is that really such a horrible horrible thing?

by jlagace on Nov 11, 2007 7:41 PM PST up reply actions

not following you..

I just thought it would be interesting to some to analyze our first five games, why we've lost, and what we could try to do better. I tried not to make any absurd generalizations about the team we will be when we're in midseason form or anything like that.

And these stats do mildly take minutes played into consideration, the adjusted weighted diff stat I posted weighs it per 48 minutes.

Yes it is true this stat needs to be taken with a grain of salt, but so does every single NBA stat. They're all skewed by who a player is playing with, who they're playing against, and by how many minutes they've played. The thing I like about the +/- stat is that it avoids the "money stats", and gives at least some merit to a player's intangibles: how they affect how their team plays, including leadership, hustle, deferring to hot teammates, etc.

by jlagace on Nov 11, 2007 5:32 PM PST up reply actions

context

Stats do need context.

Sometimes the context says the stats are providing an inaccurate conclusion (which is the case here, imo).

Questions? Complaints?

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

by OptionZero @ Golden State Of Mind on Nov 11, 2007 5:40 PM PST up reply actions

curious

Which specific conclusions are inaccurate?

by jlagace on Nov 11, 2007 5:49 PM PST up reply actions

specifically

Biedrins. He played major minutes; the team has played three times the available minutes with him than without him (184 on, 55 off). He's the only shotblocker on the team; watch the games and see how often the D gets busted on the perimeter and he has to scramble and make up for someone (Ellis) mistake.

Ellis. He has a massive positive effect on paper, but he pushes Azubuike to SF. Neither is big enough to defend their respective positions in such a lineup, meaning team defense suffers as a whole. He has also played alot of minutes, so his "off-court" time is small...too small to be worth anything analytically.

Overall, +/- is a better long run analysis, and most certainly not to be used by itself. If you yourself agree with that, then why bother causing a ruckus when others disagree with the conclusions you've drawn from them?

At this point, I could probably draw better conclusions from watching the games with my naked eye. 5 games of 82 with makeshift lineups...I might as well ask a blind man for his best guesses on whats wrong.

Questions? Complaints?

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

by OptionZero @ Golden State Of Mind on Nov 11, 2007 7:35 PM PST up reply actions

thanks for specifying

Biedrins: You're right, Biedrins is important. The point I'm emphasizing is 130-140 in the 55 minutes when he's off the court. Based on that, I'm not saying for something drastic to happen and that he shouldn't start. I'm saying when either Harrington or POB have been at center, we haven't gotten totally destroyed.

The main conclusion I've drawn from our Center situation is that POB should get more minutes, not that Biedrins should get less. Which implies Harrington should get less minutes at center. And that's based on the improvements I've seen from POB in the preaseason, in the few minutes I've seen of him in the regular season, and his positive +-; not JUST his positive +-.

Ellis: I don't want to see Ellis start at SG as much as you do, and I never said anything about it. I do suggest in this moment of urgency (we need to win a game), that Belinelli gets little to no minutes and we stay exclusively with our 3-guard rotation of Baron-Monta-Buke.

The part I find weird is people are disagreeing with conclusions that I didn't make, but disagreeing with conclusions that could be suggested by the statistics without taking them into context.

by jlagace on Nov 11, 2007 8:47 PM PST up reply actions

Yes but

As others have said, five games is way too small a sample size to conclude anything remotely meaningful. You do well to add the caveat that "our other players haven't played enough significant minutes to be included" -- but this pretty much goes for everyone you've listed. Troy Hudson, for example, could easily slide to the top or bottom of your list if two or three shots (by him or others) had rimmed out rather than in (or vice-versa).

I'm not enough of a stat-nerd to know at what point the sample sizes become meaningful, after five games they just don't pass the smell test.

But OK, as long as we're talking about five games ... I couldn't find plus-minus numbers on the site you linked, but I couldn't help but notice that JRich is by far the worst player on his team in terms of "weighted difference" (the same as weighted plus-minus, I think, but adjusted per possession rather than minutes played). This actually jibes pretty well with his "money stats," which are terrible across the board...

Again, though ... small sample sizes.

by Sleepy Freud on Nov 11, 2007 6:09 PM PST up reply actions

about sample sizes...

It's like if I said "Hey that guy had a good game and is playing well, maybe he should play more next game."

And somebody else says "Yeah, but it's only one game, that doesn't mean he's a good player."

Well, I had never said he was a good player in the first place; all I said is he had a good game.

What I'm saying now is, "This is an analysis of the first five games, maybe we should play to our most recent strengths and try to minimize our weaknesses." And at this point, it's clear that what the Warriors are trying to do isn't working so well.

by jlagace on Nov 11, 2007 6:19 PM PST up reply actions

Statistically

If you treat a player's performance in each game as independent and you want to compare players I'd go with 8-12 games. That's usually enough to net you significance in a study.

There's several issues though:

I don't believe games are independent meaning if you stink it up or you piss Nellie off in one game it will effect how you play the next game.
The samples aren't equal obviously if one player plays 35 minutes and another player plays 5 minutes.
Your sample should include half home games and half road games because some people go Jekel/ Hyde.
Looking at a player's performance in isolation is really misleading. This is a team game. I actually think the +- 5 man unit measures are more useful.

http://www.82games.com/teamunits.htm
(Anyone know where to get this season's stats?)

Ultimately PER, +-, and other "objective" measures don't tell the whole story, but I doubt anyone believed they did. They can be useful, but if you don't watch the games or know hoops by themselves they can be incredibly misleading as we all know.

I'd go with a sample of 20 games given the problems with minutes and home/ road, but that doesn't mean you can't make some good deductions with a small sample. I don't need 20 games to tell me that this team simply can't rebound, hit free throws, or defend the interior. I don't need 20 games to tell me that this team needs some talent upgrades if they're not going to waste BD and Jackson's prime as well as what could be Nellie's last season coaching.

by Atma Brother ONE on Nov 11, 2007 6:47 PM PST up reply actions

agreed

Blog-a-Bull makes heavy use of "+- pairs" to evaluate how their backcourt is doing.
One example:

Hinrich is purportedly a very good defender, but his +- had been disappointing. A closer look at the +/- for Hinrich + Duhon revealed what the problem was:

When you put Duhon next to Hinrich, Hinrich has to defend bigger SGs, who often abuse him in the post. Nothing really against Hinrich or Duhon individually, but it was stupid to put them together because the two do not fit together.

Questions? Complaints?

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

by OptionZero @ Golden State Of Mind on Nov 11, 2007 7:28 PM PST up reply actions

Right but

Drawing conclusions about the team based on over 1000 total player minutes, is quite different from using +/- analyses based on samples as low as 20-30 total minutes (Baron's off-court minutes, for example).

I think plenty of meaningful things can be said after five games, but stuff like "we're better with Monta than with Azubuike" or "we're much better with Barnes than we are with Harrington" aren't among them.

by Sleepy Freud on Nov 11, 2007 7:44 PM PST up reply actions

Definitely

I agree with that although I'm slightly tempted to wait 5 or so games till after Captain Jackson comes back this Sunday morning against the Raptors. My conclusion so far is that this team minus Jack is just awful. With him I'm projecting the 7th-10th best team in the West. I really hope Mullin is working the phones right now.

by Atma Brother ONE on Nov 11, 2007 7:51 PM PST up reply actions

hm?

I never said Buke was better than Monta, and I never even tried to imply it. All I commented on was that in those 5 games, the team has been way worse with Monta off the floor than with Monta on the floor.

I never said to start Barnes over Harrington or anything like that either. But I would like to see Barnes take more of Pietrus' minutes in the future.

Those are the kind of conclusions one could take from those stats without any context. I simply saw those stats, applied them to context, and made some conclusions.

by jlagace on Nov 11, 2007 7:59 PM PST up reply actions

Which is perfectly fine

I thought it made for an interesting discussion. It's always cool to play around with numbers.

Smart people can see things differently and that's perfectly fine. I don't think anyone can be right or wrong about something like this.

by Atma Brother ONE on Nov 11, 2007 8:05 PM PST up reply actions

No, but

Going by the numbers presented alone, the two conclusions I drew ("Barnes > Harrington" and "Monta > Azubuike") are no more or less true than any of the ones came to. Which is to say, the numbers paint a very, very incomplete picture, if not a misleading one.

Again, I'm not arguing about the conclusions per se but about the methods you used to arrive at them.

by Sleepy Freud on Nov 11, 2007 8:38 PM PST up reply actions

Ok I gotcha

Just understand I didn't come to those conclusions JUST BY the +-. Rather, I made conclusions based on what I've seen and used the +- stats (taken in context) as support.

by jlagace on Nov 11, 2007 8:53 PM PST up reply actions

I have an idea...

Re watch some games, armed with stats, and see if you come up with the same conclusions, backed up with anecdotal evidence. Faetati posted some links to get some of the games in another diary, so that part is covered. Its dead time until tuesday anyway.

THIS IS OOOAKLAND!!!

by Tim N Chris Burger on Nov 11, 2007 9:07 PM PST up reply actions

Now that I buy ;-)

1 Dub.

by Sleepy Freud on Nov 12, 2007 12:48 AM PST up reply actions

sort of

off topic here, but I didn't think it warranted its own diary. What's up with the 5 days off??!! I can't remember a team having 5 days off between games except over the all star break or something. this is crazy.

Let's Go Oakland! Gas, Brake, Dip.

by OaktownFunk on Nov 11, 2007 9:46 AM PST reply actions

It is kinda weird,

but welcome, imo. THis team might just need that extra practice time.

THIS IS OOOAKLAND!!!

by Tim N Chris Burger on Nov 11, 2007 12:22 PM PST up reply actions

handcuffed by injuries, flu

remember some of these decisions are being made in the training room:

according the Merc GS can't even scrimmage 5x5 -
Barnes and Baron are being rested,
Croshere's back is still bad (spinal disc inflamed),
Hudson has a bad wheel (left hip inflammation),
MP is sick (flu),
and the one guy really emerging as a player, Azubuike, has a bone bruise (ankle)

Nelson said Perovic is going to the D League, but BWright will stick - for now.

if ever a 5 day layoff were an opportunity to get healthy and get right, it's now ...

by hardcore on Nov 11, 2007 1:12 PM PST reply actions

Nelson vs. Stats

When asked about individual players, Nellie's singled out Monta, Barnes, and Biedrins as players who need to step up their game.

Monta's defense has left much to be desired, and neither he nor Barnes have been able to spell Baron's ball-handling duties in the absence of Jackson.

I'm surprised to see Harrington so far down on that list, although even with the lost weight it's obvious he won't ever be a great rebounder, defender, or post-player: he's decent-good at all those things and a very good shooter, but that's it.

Until Brandon Wright, or POB, reaches the point where he/they can provide 20-30 miinutes a game, the lack of frontcourt depth will severely limit this team's options: Jackson, TMNT, Croshere, Barnes, and Biedrins are all we got. None of those guys can operate the post, only 1 is a shotblocker, and only 1 can create offense. At least our defense shouldn't be as horrid as it has been, but I don't think you can even say that this team is 1 trade away from seriously contending in the west.

by BingBluNT on Nov 11, 2007 2:48 PM PST reply actions

almost

TMNT and Jack can both create their own offense. Last year Jack was much better at it (he was probably in better shape, and he did take a rest whenever there was a rebound to be had.

TMNT's shown his entire offensive arsenal to great effect this year.

Let's hope Jack can give us some...without turning it over.

POB should affect shots, but he's not gotten any significant time.

Questions? Complaints?

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

by OptionZero @ Golden State Of Mind on Nov 11, 2007 4:22 PM PST up reply actions

I don't need

any fancy stats to tell me why the W's have a big fat goose egg in the Win column...FREE THROW SHOOTING and defending the 3pt. shot. Simple as that folks.

Let's Go Oakland! Gas, Brake, Dip.

by OaktownFunk on Nov 11, 2007 8:54 PM PST reply actions

Here's what I believe-

PG Baron 36 / Ellis 12
SG Buke 30 / Ellis 18
SF Barnes 24 / Pietrus 24
PF Harrington 36 / Barnes 12
C Biedrins 36 / POB 12

Of course, if the game stays close Baron needs to take more of Ellis' minutes at PG.

Although I like to see Harrington start, I believe Harrington must be on the floor when Baron rests. Right now, he's the second best player on the team at creating his own offense.

Buke and Barnes should be on the court with Baron as much as possible, because I think that's when they're most effective.

by jlagace on Nov 11, 2007 9:25 PM PST reply actions

I like that breakdown,

But I think buke should be getting 40 min a night. He's young, and fills up a stat sheet, doesn't hurt the team on either end when he's out there. Plus more of buke means less monta at the 2.

THIS IS OOOAKLAND!!!

by Tim N Chris Burger on Nov 11, 2007 9:36 PM PST up reply actions