APBRmetrics

[onlxn]

Howdy, folks. A friend and I have just started a new semi-statsy GSW blog; while we are but novices in the numbers game, the sheer absurdity of the Warriors has given us plenty of fodder already. As an example, the team's rebounding differential is now -9.8 a game. If sustained over the full year (which, Biedrins's pubis permitting, it won't be), that would be the worst rebounding margin in league history. The current record-holder? Don Nelson's 1989-90 Warriors. You get used to that sort of thing around these parts.

But whilst lamenting the idiocy of our version of smallball and grimacing at the bizarrely huge minutes given to Vladimir Radmanovic and Mikki Moore, we came across something sort of incredible. Through December 15th (so excluding yesterday's loss to the Spurs), we have been outscored by 8.0 points per 48 minutes with Monta Ellis on the floor. When Monta sits -- not often, but often enough for about four games' worth of off-court data -- we have outscored our opponents by 12.4 points per 48 minutes. His net plus-minus is -20.4, not only the worst of any NBA starter by a mile, but more than twice as bad as Kevin Durant's much-discussed showing last year.

If you've been watching us play consistently (God help you if so), it's not exactly news that Monta has been ineffective. And there's some padding in those off-court numbers, particularly a couple huge C.J. Watson/Acie Law-led garbarge-time runs that are still skewing things. Nevertheless, this is a pretty crazy number. And it begs a question that I have not been able to answer in my Web wanderings: what is the worst net plus-minus on record for an NBA starter?

Thanks for the great work you folks do, and (in advance) for the answer.

[Crow]

Looks like a good start on the blog.

The Franchise Fix series might be worth a look by fans of other teams as a review checklist.



It does look like Ellis has this distinction this season among minute qualifiers / for clear starters (though Drew Gooden has been a part-time starter and is slightly worse for +/- on/off).

http://tinyurl.com/yz7fdk8

Nobody sustained much more than half this level of bad for a full season and qualified on minutes in the last 2 seasons.

Looks like the change from previous seasons is mostly about offense and a bit more about the quality of the offense when he is off compared to on and thus this is a layered story not just about Ellis.

Digging at 82 games thru more years could turn up other notable cases. I checked very briefly but didn't find any of this level but I leave further exploration to you or others.



My super quick, lightly informed take would be

Trade Ellis. Got to or things don't change.

Maybe Maggette and Biedrins if you got a good offer. Radmanovic if you can.

Get a big who can defend / rebound.

Try Watson in a bigger role. Curry in a smaller one as the backup PG only. Maybe you get a vet PG for a slice of the time.

Play Azubuike at SG, (Bell if healthy), Morrow at SG- SF.

Randolph 2/3rds SF - 1/3 PF and see what happens.

Play Wright big minutes, someday.

[onlxn]

Thanks for the kind words, and for the hookup... I had poked around basketballvalue but somehow failed to parse the rankings properly.

[Mike G]

[onlxn]

Randolph deserving more minutes is a drum that I (and most Warriors fans) have been banging for weeks now. He's still error-prone, often failing to cut to the hoop when given an open lane and getting blocked a lot underneath... still, he's easily our best available frontcourt player, and probably one of our best players, period. Nellie supporters claim Randolph doesn't deserve to start until he stops making so many mistakes. But if frequent mistakes disqualified you from a starting job, we'd be opening play down five-on-zero every night.

[onlxn]

(Note: I am always down to talk some Warriors, but don't want to distract from the main goings-on here; if I am blabbing too much about one sorry team, let me know and I'll happily desist.)

[Crow]

I am not up on the health statuses. I didn't label it as such but you probably should think of my list as an outline for next season.

Pegging Morrow as SG-SF I'd try to give him bigger, slower or less challenging SGs. And if he needs to lose 10 pounds to gain a bit more speed then ask him to do that too. If you want to move away from small-ball and keep Morrow then he has to shift down some. If that won't work then you trade him too.

Randolph has been very lightly used as SF. Maybe it won't work, I don't have much basis to say but he isn't working as PF at this stage of his career. Do the Center experiment if you want but I'd want to know about him at SF at least as much and probably more.

Ellis may be fairly tough to move but I'd think that from within a group of Chicago, NY, New Orleans, Memphis, maybe Washington and perhaps some surprise they could find something.

I leave comments about Nellie and management to you.

[onlxn]

A quick addendum relating to the original question... in the eight seasons for which there is readily available plus-minus data, no NBA player has fared as poorly in that measurement as Monta is currently faring. Since 2002, only ten players who've played a third or more of their teams' minutes have garned a double-digit negative net +/-:

1. Hakim Warrick, -12.0 (44% of minutes, MEM '07-'08 )
2. Andray Blatche, -11.7 (42% of minutes, WAS '07-08 )
3, Al Harrington, -11.4 (61)% of minutes, IND '03-'04)
4. Michael Curry, -11.2 (39% of minutes, DET '02-'03)
5. Sean Rooks, -11.0 (33% of minutes, LAC '02-'03)
6. Andres Nocioni, -10.6 (47% of minutes, CHI '04-'05)
(tie) DeShawn Stevenson, -10.6 (67% of minutes, ORL '05-'06)
8. Greg Buckner, -10.2 (35% of minutes, DAL '06-07)
9. Ryan Gomes, -10.1 (57% of minutes, BOS '06-'07)
(tie) Eddy Curry, -10.1 (38% of minutes, NYK '07-'08 )

Only three of these guys were on the floor for their teams more than half the time, and none rates worse than -12.0 points per forty-eight minutes. Monta's sitting on -20.4 points per 48, in 84% of our minutes.

Now, obviously the small sample size skews things, as ten other '09-'10ers are also still sporting -10+ NPM totals in decent minutes.... most of them won't be at year's end, as they'll have to either get better or get benched. Still and all, only Drew Gooden rates as feebly on a per-48-minute basis as Monta, and only Jeff Green comes close to Monta in the sheer amount of time given to an ineffective player.

Long story short: Monta has the most robustly terrible plus-minus numbers in the game, and, quite possibly, some of the worst plus-minus numbers in recorded league history. This guy's unique, and not only because of his playing style.

[QMcCall3]

[onlxn]

(82games has updated their numbers: through Saturday, Monta's net plus-minus now stands at -17.8. Still worse than any full-time starter on record, but better than it's been, and the gap between him and Drew Gooden is widening, at least.)

[QMcCall3]

[onlxn]

An update: Monta's net plus-minus now stands at -11.1. Still only 304 off-court minutes to compare, so obviously there's some wonkiness here, but, damn.

In the seven past seasons 82games has on record, only three players who played in more than half their team's minutes recorded NPMs of -10.0 or worse... three players this year are on pace to join that list. The current "leader"board:

1. Jeff Green, -13.2 (76% of minutes, OKC '09-'10)
2. Al Harrington, -11.4 (61% of minutes, IND '03-'04)
3. Monta Ellis, -11.1 (85% of minutes, GSW '09-'10)
4. Troy Murphy, -10.9 (50% of minutes, IND '09-'10)
5. DeShawn Stevenson, -10.6 (67% of minutes, ORL '05-'06)
6. Ryan Gomes, -10.1 (57% of minutes, BOS '06-'07)
_________________
Golden State Worriers: Angst & Analysis

[Crow]

One year has some utility for seeing changes but I am generally moving away from it.

Here how these guys look on hoopnumber's regularized 3 year analysis ending with last season (equally weighted and rounded)

Jeff Green -1.9
Al Harrington +0.1
Monta Ellis +0.1
Troy Murphy -1.3
DeShawn Stevenson -1.4
Ryan Gomes -3

Here how these guys look on basketballvalue's 2 year analysis (well 1 1/2 years since it includes this season)

Jeff Green -6.6
Al Harrington +0.7
Monta Ellis -2.1
Troy Murphy -3.5
DeShawn Stevenson -2.3
Ryan Gomes +0.3

Unless a really big Adjusted holds up for multiple seasons I think it is probably overfitted and too extreme to be believed literally.

By basketballvalue's 2 year analysis 28 players rate worse than -5. By 1 year there is it probably close to 70. By hoopnumber's regularized 3 year analysis there are only 5.

Only Jeff Green was -5 on one of the lists, the easier one to achieve that on.

These guys rated the worst in 1 year but it doesn't necessarily make them the worst players with actually that estimated level of terrible impact. Maybe worst performance in a specific role over that time period. But with some unknown amount of noise so it is hard to say exactly. Round it off to not "good" and probably "below average" in most cases? You could go further but not everyone would and I am not sure if you should.

[onlxn]

Oh, I'd agree... I don't regard this as some comprehensive list of the league's worst players, or anything like that. Still, as you say, you'd have a hard time arguing these guys performed well in those roles those years. And it's striking to see Monta on this list, given his production. It's not like defense explains Monta's horrible showing, a la Durant last year... all of the Warriors' improvement when Monta doesn't play comes on the offensive end. And while our Monta-centric offense is clearly suboptimal, as it precludes the use of more efficient weapons like Morrow and Curry, it's still wild that the disparity is this big.
_________________
Golden State Worriers: Angst & Analysis

[Crow]

Looking at past Adjusted Factors for clues he might be hurting team FT rate a bit and Offensive rebounding a lot (neither of which would be picked up in boxscore based metrics) but that offensive past doesn't give enough explanation of why things are so bad this season.

Previously he scored very high on team eFG% impact. But that was before the last trade and at least to some extent before the others.

Maybe the Adjusted Factor level detail for this season if/when it is provided will explain more but I'd guess a lot of his high negative will drop away thru the regularization.

And while I am certainly not an expert to judge whether that process is right or best, I lean toward thinking so right now.