clock menu more-arrow no yes mobile

Filed under:

The problem with ESPN's Real Plus-Minus

ESPN has a new basketball stat. The stat loves Manu Ginobili. I am not a fan of this stat. Does not compute.

Pat Lovell-USA TODAY Sports

On Monday ESPN gave us just what we all wanted: Yet another supposedly all-encompassing basketball stat. A month ago gave us Kate Upton and various other supermodels posing seductively and wearing practically nothing. The former website continues to be as popular and profitable as ever, while the latter is hemorrhaging money.

(Clearly my finger is not on the pulse of America, but I digress.)

So what is this Real Plus-Minus?. Here's the short answer from the column that introducing it:

Real plus-minus (RPM) was developed by Jeremias Engelmann, formerly of the Phoenix Suns, in consultation with Steve Ilardi, University of Kansas psychology professor and former NBA consultant.

It follows the development of adjusted plus-minus (APM) by several analysts and regularized adjusted plus-minus (RAPM) by Joe Sill.

RPM reflects enhancements to RAPM by Engelmann, among them the use of Bayesian priors, aging curves, score of the game and extensive out-of-sample testing to improve RPM's predictive accuracy.

Basically, as I understand it, RPM intends to be ESPN's version of baseball's WAR (or Wins Above Replacement), a handy end-all-be-all stat that captures the net value of a player, taking everything into account from teammates on the court with them to the opponents, the score at the time they've entered the game, the venue, the coaches, the pace of the two teams and "other factors."

For example, let's say we're talking about Matt Bonner. If he's on the court with Tim Duncan, Manu Ginobili, Tony Parker and Kawhi Leonard and against the Milwaukee Bucks, then it would be natural to assume that his plus-minus in those minutes will look pretty good, and similarly his offensive rating, defensive rating and net rating numbers will comply as well.

What RPM proposes to do is figure out what a guy like Bonner -- or anyone, really -- is worth if his teammates aren't Duncan, Ginobili, et al. but rather league average players and his opponents are league average players and the score is tied and the venue is neutral, etc. The interesting twist is that unlike conventional plus-minus, they break down the value both offensively and defensively and then come up with the net rating, per 100 possessions, based on adding the two, similar to the more popular net rating we've become accustomed to, based on offensive and defensive ratings per 100 possessions.

It was amusing and entirely predictable that one of the main ways ESPN introduced the stat was by settling the supposed debate for the Sixth Man of the Year award between Chicago's Taj Gibson and the Clippers' Jamal Crawford. *Spoiler Alert* The column concluded that Gibson is the more valuable player because while Crawford is indeed the superior offensive player (over twice as much per 100 possessions than Gibson, according to this metric) the defensive difference between the two is even more pronounced, in Gibson's favor. RPM has Crawford's defense as such a detriment that he's an overall negative player, more harmful than helpful.

The amusing part, naturally, is that the article made no mention of one Manu Ginobili, whose RPM dwarfs both Gibson's and Crawford's. One would think that Ilardi's article would trumpet Ginobili's merits for the "6MOY" instead of lauding Gibson's. Incidentally, Gino ranks ninth in the league in RPM, one spot behind a fella named Tim Duncan, but even he isn't the highest-ranked bench player on the list, but we'll get to that in a minute.

The better argument Ilardi makes in introducing the stat is in using it to show that guys like Kevin Durant, Russell Westbrook and even Serge Ibaka are way more valuable for the Oklahoma City Thunder than sixth man Reggie Jackson, when stats like net rating or conventional plus-minus per game would conclude otherwise.

You look at the overall leaders in RPM, and it looks legitimate, more or less, with a couple of notable exceptions, as to what you would expect a who's who of the league's best to look like. Where RPM starts to lose me though, is with Kawhi Leonard, and while I admit I'm looking through the prism of black and silver tinted glasses, Leonard's ratings still have me baffled.

Leonard ranks 14th (14th!?) among small forwards, behind such luminaries as Atlanta's DeMarre Carroll, Dallas' Jae Crowder, Chicago's Mike Dunleavy and the Clippers' Matt Barnes. Carmelo Anthony ranks ninth. I know it's fashionable to rake Anthony over the coals for his lackadaisical defense and ball-stopping on offense, but I'm sorry, at some point common sense has to intervene. If Anthony is your ninth-ranked small forward, your stat is faulty.

Even more questionable than Leonard's overall ranking among small forwards though is his defensive ranking at his position: 11th. Behind ancient fossil Paul Pierce of Brooklyn, behind Dunleavy, behind Boston's Gerald Wallace.

LeBron James, by the way, is a negative defensively -- as in below average. 43rd for his position, just ahead of New York's Steve Novak. According to RPM James is a worse defender than Durant, worse than teammate James Jones, worse than the Spurs' Austin Daye and worse than the Clippers' Hedo Turkoglu.

So you'll forgive me if I'm not buying what RPM is selling.

Meanwhile, offensively Leonard is the 27th ranked small forward, behind such juggernauts as Barnes, Carroll, Crowder, Minnesota's Corey Brewer, the Clippers' Danny Granger and noted shooting wizard Josh Smith of the Pistons. Leonard is shooting 53 percent from the field, 38 percent from three and is among the leaders at his position in both offensive and defensive rebounds. I don't think all of that is due to his teammates making him look good.

Leonard aside, the other way RPM loses me is when it lionizes backup big men such as OKC's Nick Collison, Miami's Chris Andersen and, to a lesser degree, Gibson. However, in that regard, net rating does the same thing.

Collison, 33, is averaging a career-low 16.9 minutes per game and the Thunder drafted rookie Steven Adams to ostensibly replace him. More interesting to me than just those minutes is the fact that almost none of them ever come in crunch time for the Thunder. Collison didn't play down the stretch in any of their recent games against the Spurs, Rockets or Suns and really hasn't all season. Do we blame coach Scott Brooks for this, or do we understand that Collison is a guy who excels in his limited role?

The thing is, Collison couldn't play a lot even if Brooks wanted him to. He averages 5.0 fouls per 36 minutes, meaning he'd be a regular threat to foul out as a featured player. He's valuable, but he's not some diamond in the rough. He's a bench guy who happens to be on a team with superduperstars on it. That's it.

Andersen is somewhat more credible in that he's less foul-prone than Collison, plays more minutes, plays down the stretch at times for Miami and he has more legitimate scoring and rebounding numbers per 36 minutes. He's no star, but he's not just a guy either.

Ultimately we have to take any efficiency stat with a grain of salt when it has bench players up among the lofty heights of the Durants and LeBrons. If a guy plays less than 20 minutes a night, in my mind he's automatically disqualified from the discussion because in all likelihood he's compiling his numbers by playing against fellow second-teamers. We have to give their coaches the benefit of the doubt of not being TOTAL idiots, and accept that the players in question have specific limitations, whether it's endurance, being foul-prone, or being able to succeed only in certain match-ups and situations.

The reason I've always given Ginobili's numbers more credence than other reserves (besides my biases, of course) is because over the course of his career he's shown he can be completely kick-ass as a starter, playing against other starters. Plus, Gregg Popovich still plays him down the stretch, which matters. If your coach doesn't trust you to be on the floor in the final five minutes of a close game, what good are you ultimately?

Still, I think at some point we do have to recognize volume. Minutes matter. Being able to produce for longer stretches matters. It's well and good Ginobili plays like an All-Star for 23 minutes, but there is still the matter of the other 25. If Ginobili gives you A- production for 23 minutes, that's not as valuable as a player who gives you B+ production for 35. Even as biased as a I am, I'm not blind to that reality.

For that reason, I think ESPN's version of WAR, which uses RPM and incorporates the total minutes guys have played to figure out true production, is closer to the mark, but I'm still not satisfied with it in its current incarnation.