clock menu more-arrow no yes mobile

Filed under:

ESPN's new advanced stat seems as flawed as a Josh Smith jump shot

The NBA is still searching for a convenient, all-in-one stat that can adequately capture a players value. Unfortunately, ESPN's new RPM stat is what what people have been looking for.

Tim Fuller-USA TODAY Sports

As a decidedly non-numbers guy, I always love discovering new stats that people have slaved over that try and impart us with a certain amount of knowledge that was lacking. People have been generating advanced stats for decades (I'm sure that at one time field goal percentage was considered an advanced stat), but in basketball the adequate all-in-one stat has remained elusive.

Unlike baseball, basketball is not a series of one-on-one matchups but a team sport where success on both ends of the floor rely not only on individual talent but also the talent, communications and execution of your teammates. For a long time PER has been considered the gold standard of advanced basketball stats, but ever since it's creator John Hollinger left ESPN for a front office post with the Memphis Grizzlies, I've been waiting for the site to move on from PER and explore something else.

Well, that something else is here and it is called Real Plus-Minus. Most fans are familiar with the basic plus-minus stat. Credit when you're on the floor and your team scores more points than the opponent. Blame when your on the floor and your team scores less than your opponent.

It's of limited usefulness because it is heavily dependent on who you share the floor with.

I'll let ESPN take it from here.

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.


The RPM model sifts through more than 230,000 possessions each NBA season to tease apart the "real" plus-minus effects attributable to each player, employing techniques similar to those used by scientific researchers when they need to model the effects of numerous variables at the same time.

RPM estimates how many points each player adds or subtracts, on average, to his team's net scoring margin for each 100 possessions played. The RPM model also yields separate ratings for the player's impact on both ends of the court: offensive RPM (ORPM) and defensive RPM (DRPM).

Problem No. 1. The stat, like PER, seems proprietary, so there is no way to look into it's guts and see if it passes the smell test much less the rigors of scientific analysis. All we have are the results, and as a Pistons fan, the results mean either this stat is instantly discredited or I know much less about the team I watch every single day than I had previously believed (very possible!).

First, the stat is not adjusted for minutes played so much like PER, you have to see if, say, Matt Bonner is really a top 50 player this year or if 11 minutes over 58 games is enough information to really know anything. Second, even accounting for Andre Drummond's defensive struggles this year, I can't fathom a stat that tells me that Andre Drummond is a net negative on the basketball court and the 172nd ranked player in the NBA. While at the same time Josh Smith is ranked 76th overall and the second highest-rated Pistons player behind Greg Monroe.

Unfortunately, ESPN's other advanced stat, their version of WAR, fares even worse. In WAR Smith is ranked No. 38 in the NBA at 7.41 WAR while Drummond ranks No. 107 with 3.01. The mind boggles.

For your convenience I've listed the Pistons RPM numbers and ranks along with WAR and Wins Produced, which is not an ESPN stat, but is the preferred all-in-one stat of many around Detroit Bad Boys, including writers Ben Gulker and Kevin Sawyer. They're smarter than I am so I'll let them speak to its efficacy.

Player O RPM D RPM Overall RPM RPM Rank (sorted) WAR WAR Rank Wins Produced WP Rank
Median NBA Player -0.91 0.14 -1.26 217 0.57 217 1.2 250
Greg Monroe 0.95 1.18 2.13 71 6.93 42 5 87
Josh Smith 0.22 1.82 2.05 76 7.41 38 0.1 337
Josh Harrellson -2.34 3.07 0.73 125 0.61 213 0.7 287
Andre Drummond -0.09 -0.24 -0.33 172 3.01 107 16.6 5
Kyle Singler -0.91 0.27 -0.64 186 2.24 137 4.4 103
Luigi Datome -0.96 -0.08 -1.04 204 0.17 265 -0.5 467
Kentavious Caldwell-Pope -1.99 0.83 -1.16 211 1.10 182 1.3 244
Charlie Villanueva -1.11 -0.83 -1.94 279 0.45 286 -0.5 458
Rodney Stuckey 1.05 -3.12 -2.07 291 0.31 241 -0.9 479
Jonas Jerebko -2.35 0.17 -2.18 299 0.06 281 1.2 253
Tony Mitchell .-2.15 -0.07 -2.22 302 0.00 312 0.5 287
Chauncey Billups -0.58 -1.84 -2.42 315 -0.01 313 -0.5 461
Peyton Siva -2.18 -0.96 -3.14 355 -0.04 326 -0.4 453
Brandon Jennings 0.12 -4.52 -4.40 398 -3.26 430 3.6 132
Will Bynum -1.25 -3.44 -4.69 406 -1.56 415 -0.5 464

A few more reactions. I'm fascinated how these catch-all stats vary so wildly among the same players, especially regarding Smith and Brandon Jennings. RPM says the team's biggest problem might lie at the hands of it's no-defense point guard, while saying the relentless long-range chucker actually makes up for that flaw and is the second most-valuable player on the team. WAR is much the same. Meanwhile, Wins Produced says that Jennings is actually and above average player while Smith is among the worst in the league.

Also, I think if we consider today's NBA game a point guard's league (which I do), then the Pistons record makes a whole lot more sense. Thanks Jennings, Bynum, Billups, and to a lesser extent Stuckey.

What say you, Detroit Bad Boys community? What is your go-to stat these days?