Does plus/minus make you an All-Star?

The Warriors are just as bad with Monta Ellis on the court as they are with him on the bench. (Rocky Widner/NBAE via Getty Images)

Tim Kawakami of the San Jose Mercury News does not think Golden State guard Monta Ellis is an All-Star, citing Ellis’ plus/minus stats as part of his evidence. The upshot: The Warriors are pretty bad, and they are just as bad with Ellis on the court as they are when he’s on the bench. Kawakami argues that truly great players make their teams better when they’re on the court. The numbers — both from this season and previous ones — suggest Ellis does not do that.

Please note: Kawakami is a very close and astute watcher of the Warriors, and his arguments against Ellis as an All-Star go far, far beyond the plus/minus stats. I’m just using those numbers as a jumping-off point for a larger discussion here.

Here’s Kawakami:

It has been suggested that Ellis’ -3.0 plus/minus per game is not a problem, since it’s a little better than the Warriors’ -3.2 point-differential, which is not a bad point.

But, as was the case last year when I examined this issue with Ellis, once again, we go to other players with heavy minutes on bad teams … and all of the good ones have BETTER plus/minus rates than their teams.

That’s because they’re good.

The team is better when they’re out there than it is when they’re not. That’s the way it is supposed to work and all excuses for that are just excuses and now mainly pertaining to Ellis and Ellis only among players with big reputations.

You ask, well, what good players on bad teams have better plus/minus rates than their teams?

– Kevin Love does: He’s Minnesota’s No. 1 minutes guy, and he’s at -3.4 per game, while the team is at -5.4. Better.

– Danny Granger: -0.4 per game, Indiana is -1.2 per game.

– Blake Griffin: -0.7 per game, the Clippers are -1.9.

– Eric Gordon: +0.1 per game, again, the LACs are -1.9.

– Tyreke Evans: -3.7 per game, the Kings are -4.7.

Kawakami is right about all this stuff. And he’s right that Ellis probably shouldn’t make the All-Star team — not in a loaded Western Conference, and not when he ranks “just” 40th in Player Efficiency Rating.

But plus/minus and on-court/off-court differentials are funky things, particularly over half a season. You can find great players who don’t appear to be helping their teams by this measure. Do you think the Lakers are better with Kobe Bryant on the bench? Well, they have been this season. The Thunder have been loads better with Russell Westbrook on the pine. Denver’s offense has scored more efficiently with Carmelo Anthony on the bench, and Miami’s defense has been stingier with Dwyane Wade off the court. And more bizarre than any of those — at least on the surface — is that the numbers tell us Orlando’s defense has been stingier with Dwight Howard on the bench.

How do you account for these things? It’s hard — sometimes impossible — but you have to start by looking deeply at the context in which those numbers accumulate. And doing that erases some of the blemishes on Ellis’ record — for this season, at least.

Here’s why: Ellis leads the league in minutes, having been on the floor for about 1,930 of the 2,271 minutes Golden State has played this season. He has played a full 600 more minutes than backcourt mate Stephen Curry. This fact alone means that Ellis’ plus/minus numbers are more likely to lean very closely to his team’s overall mark.

This becomes even more clear when you take the next step and look at Golden State’s lineup data. The Warriors’ 18 most commonly used lineups include Ellis, according to Basketball Value. That is essentially unique in today’s NBA. Look at the same data for the Blazers and Bulls, and you’ll find those teams have some pretty often-used lineups that don’t include Luol Deng and LaMarcus Aldridge — two of the other leaders in minutes.

The first lineup among Golden State’s unit that doesn’t include Ellis has played a total of 16 minutes all season — just over one quarter of an NBA game. And that lineup has indeed performed much better defensively than Golden State has fared overall. So have the next four lineups that don’t include Ellis. Those four lineups have logged the following minutes totals: 12.6; 10.5; 9.45; 8.97.

So the five non-Ellis lineups the Warriors have used most often have logged about 57.5 minutes between them — just short of five quarters. And they’ve done this mostly in short spurts. The third of these five groups — the one that has logged 10.5 minutes — consists of Jeremy Lin, Charlie Bell, Reggie Williams, Jeff Adrien and Vladimir Radmanovic. That group has yielded only about 105.5 points per 100 possessions — much better than Golden State’s overall defensive average.

But are we really ready to say that that group is a solid defensive lineup? Of course not.

This is not to say Ellis is an ace defender. He’s not, although he’s usually called on to defend the opponent’s best backcourt scorer because of Curry’s limitations — or (if we’re being kind to Curry) Keith Smart’s perception that Curry is an inferior defender. There’s no question that Ellis, undersized as a 2, is going to struggle against elite shooting guards. That he’s occasionally indifferent, particularly in transition defense, doesn’t help.

Kawakami is absolutely right that Golden State’s defense has been consistently better with Ellis on the bench over the last three or four seasons, and that trend is disturbing and merits deeper investigation. It’s just that a half-season’s worth of plus/minus numbers requires a close look in context. How could the Magic, for instance, be better defensively without the best defensive player in the world on the floor? I’d wager some of my cash that it has a lot to do with the fact that Howard is often on the bench while Orlando’s opponent on a given night has its own backups on the floor. That’s one reason why Orlando units that include the frontcourt of Brandon Bass and Ryan Anderson — at least the first few of them — appear to be so stingy defensively.

You can come to a similar conclusion about the numbers showing Miami’s defense performs better without Wade (though it still does quite well with him). If you dig into Miami’s lineup data, you’ll see that units with LeBron James and four bench players have played defense as if they were the 2007-08 Celtics. Those units get a lot of run at the end of the first quarter, when Wade and Chris Bosh sit — and when LeBron and his bench pals go against opposing bench players.

I do believe plus/minus stats and on-court/off-court numbers can tell us a lot about NBA players, units and teams — I cited those numbers in backing Aldridge as an All-Star, after all. But you can’t rely on them alone — and, to be clear again, Kawakami is not doing this — and you must investigate the context in which they arise.