Everybody has a type, and ESPN's new real plus-minus is no exception.
The beauty of RPM is that its assessment of players isn't made on assumptions about value, as with most popular box score metrics such as PER, WARP and Win Shares. Instead, RPM goes by what happened when the player is on the court or on the bench, adjusted for teammates and opponents. So it starts as an unbiased measure of player value.
At the same time, RPM results do reveal certain trends, including types of players that rate well, or poorly, due to common characteristics.
Taking a look at those groups can help us understand what really matters in terms of winning and losing in the NBA.
In the spirit of the popular dating apps, in which users anonymously rate potential mates by either swiping right (to indicate interest) or swiping left (to pass), let's take a look at what RPM would say when various types of players pop up on the screen.
Swiping right
• Nonscoring big men
Examples: Nick Collison (+5.8), Amir Johnson (+4.8), Taj Gibson (+4.1)
Collison and Johnson were both members of my RPM All-Star team, but they're not alone among big men with pedestrian scoring averages -- or worse -- who come out well by RPM's measurement. Gibson also is in the top 20, as is Anderson Varejao (+4.1).
The obvious common factor here is defense. Box score stats have a tough time crediting these players for their contributions since they're not elite shot-blockers. But another factor is the ability to set good screens. Gibson and Johnson both appeared on Amin Elhassan's list of the NBA's best screeners. Freeing teammates with picks is a skill that's completely invisible in the box score, but it's one that RPM is capable of picking up.
• Floor-spacing big men
Examples: Ryan Anderson (+1.9), Matt Bonner (+2.8), Channing Frye (+4.1)
Defense isn't the reason stretch 4s make RPM numbers jump. Instead, these players generally rate well offensively. Anderson and Frye both rank in the NBA's top 20 in offensive RPM, with Bonner in the top 30. (Steve Novak of the Toronto Raptors, a more limited member of this group, also rates solidly above average on offense.) That's partially due to their efficient scoring, especially in the case of the more prolific Anderson.
However, RPM also credits stretch big men for the way their shooting opens up the floor for teammates. RPM co-creator Jeremias Engelmann recently posted adjusted shooting ratings that measure the impact players have had on their teammates' true shooting percentage. Frye is the lone big man near the top of those rankings who isn't an All-Star, while Bonner also rates as having a notably positive influence.
• Players with high steal rates
Examples: Monta Ellis (+1.9), Ricky Rubio (+3.9), Josh Smith (+2.0)
Larceny isn't typically a desirable characteristic in a mate, but RPM seeks it out. Rubio is an RPM All-Star, while Ellis and Smith both rate better than their reputations would suggest. But this trend is bigger than just types. Of basic skill statistics (not including stats that rate combinations of skills such as true shooting percentage), only 2-point percentage correlates more closely with RPM than steal rate.
That backs up recent research by FiveThirtyEight.com's Benjamin Morris, who found that steals are more valuable to teams than any other counting stat in the box score. It's unclear whether this is because steals are so inherently valuable or because they serve to indicate other skills -- steals actually correlate better with offensive RPM than defense -- but either way, one of RPM's first chat questions will always be, "Do you like to steal?"
Swiping left
• Score-first point guards
Examples: D.J. Augustin (-4.7), Aaron Brooks (-4.0), Brandon Jennings (-4.4)
There's one other skill statistic that correlates as well with RPM as steals, and that's assists, which is the strongest single factor in offensive RPM. That's bad news for point guards who tend to rack up relatively few of them.
The other common factor among Augustin, Brooks and Jennings is that they're poor, undersized defenders. While some point guards, such as Ramon Sessions (-1.4), can compensate at the other end, this group's offensive ratings are much worse than the average point guard. RPM simply isn't impressed by their volume scoring, which generally doesn't translate to wins.
• Rookies
If you thought I was harsh on this season's rookie class last week, stay away from RPM's ratings. Of RPM's bottom 11 players who have seen at least 500 minutes of action, four are rookies: Anthony Bennett (-5.3), Ben McLemore (-5.4), Tony Snell (-5.5) and Archie Goodwin (-5.9). That's the same number of rookies that RPM rates as above-average contributors: Ryan Kelly (+2.3), Trey Burke (+1.4), Matthew Dellavedova (+1.3) and Giannis Antetokounmpo (+0.4).
Though those ratings are partially a comment on this season's rookie class, in general young players tend to rate worse by RPM than their box score stats because of their difficulty in getting up to speed on defense. As a result, RPM tends to prefer older players.
RPM probably isn't going to find romance in the City of Brotherly Love. At its heart, RPM is divvying up credit for team performance among the players on the court. Alas, there hasn't been much credit to go around in Philadelphia this season. Therefore, it's only natural that the two worst players in RPM wins above replacement (factoring in minutes played and replacement level) are Sixers guards Tony Wroten (-4.9 WAR) and Elliot Williams (-4.2).
Byron Mullens (-1.4) and Hollis Thompson (-1.3) are also near the bottom of the league, while Michael Carter-Williams (-1.0) rates much worse than he does by box score stats. Ultimately, RPM rates just one Sixers player as better than average: lone veteran holdover Thaddeus Young (+6.8 WAR).