The Nets' Midterm Grades Are In!

With James Harden gone and Kyrie Irving semi-AWOL, the Nets are, more than ever, Kevin Durant's team. But injuries and absences have forced Steve Nash to employ dozens of different lineups, relying on journeymen and rookies to play big minutes in unaccustomed roles. As the stretch run gears up, who has come through and who has disappointed?

ESPN has just posted its first real plus-minus ratings for the 2021-22 season. Unlike most basketball statistics, these ratings attempt to capture the totality of each player's contributions to his team's success or failure. And unlike most other statistics, they do so by focusing on what really matters--how many points were scored or allowed on each possession. The resulting midterm grades help explain why the Nets are where they are, and perhaps how they might pull things together over the next few months.

Real Plus-Minus

Casual fans see players' most obvious contributions, like points per game and rebounds per game. But basketball is a complex, fast-moving team sport, and a lot hinges on such hard-to-measure factors as help defense, hockey assists, screens, and box-outs. Raw plus-minus numbers shed some light on those contributions by measuring how well the team does, overall, when a given player is on the court. That's a start; but there are nine other guys on the court at any given time--what about their contributions?

Real plus-minus ratings attempt to take them into account, disentangling the distinct contributions of each player on each possession. Obviously, doing that requires some assumptions--most importantly, that the offensive and defensive contributions of each player are fixed and independent of who else is on the court. That is a simplification, but not a wildly unreasonable one. In effect, the ratings average over all the circumstances in which each player has played, boiling his performance on each side of the ball down to a single number.

If a player's team runs an offense that doesn't suit him, or plays him out of position, the rating will not reflect his true ability--but it will reflect his contribution given those circumstances. If two players are especially good playing together, both their ratings will be higher, but only to the extent that they actually play together. On the other hand, if a player "makes everyone around him better" than they are playing with others, the rating algorithm will rightly credit him for that success.

Obviously, if two players are on or off the court together on every possession, it is impossible to distinguish their contributions without bringing additional data or assumptions to bear. Even in less extreme cases, the algorithm will struggle to distinguish the contributions of players who rarely play apart, or simply rarely play. In those cases, especially, it is worth bearing in mind that the ratings are statistical estimates with a good deal of imprecision. (In order to reduce the imprecision, most of the various adjusted plus-minus algorithms in circulation, including ESPN's, incorporate additional information from box score statistics or previous seasons, with the relative weight of that "prior" information declining as the season goes on.)

The ratings are denominated in net points per 100 possessions on each side of the ball, with zero representing a league-average performance. The overall distribution is skewed by the productivity of stars, so most players end up with negative ratings. (The 225th-best player in the league, 7th or 8th best on an average team, has a -1.5 rating.)

Some fans have violent negative reactions to statistical evidence that contradicts their own "eye tests." Since the ratings are inevitably imprecise, especially this early in the season, and especially for players who haven't played much, you should feel free to discount or dismiss as "statistical flukes" those you disbelieve. (I do some of that myself in what follows.) Still, if you approach the data with a relatively open mind, you may garner some insights that help you see more of what is actually going on on the court.

Nets, Old and New

Remember, these ratings capture average performance per 100 possessions. With as many injuries and absences as the Nets have had this season, just showing up is commendable; nonetheless, guys who have played heavier-than-expected minutes will be penalized in the ratings if their performance has suffered as a result.

Kevin Durant (1313 minutes), +4.8. KD has dropped out of the MVP conversation, and not only due to injuries. He ranks 17th in the league in RPM, with about half the per-possession contribution of the top players, Jokic and Curry. Still, he has carried the offense (+3.7) and contributed on defense (+1.2) for a team without a lot of bright spots.

Ben Simmons (0 minutes), +3.7. Well, that was his rating last season in Philadelphia (+1.3 offense, +2.4 defense). Nets fans can only hope that he gets back to that level, and quickly.

Seth Curry (1697), +1.7. The Nets' second-best player, pending the arrival of Simmons, though most of that play was obviously with the Sixers. For all the talk of his defensive limitations, his defensive rating is excellent, +2.3. (That may be a statistical fluke; but his defensive rating last season was +1.7, so maybe it's a trend.) How well will it translate to Brooklyn?

Patty Mills (1831), +0.4. It's very hard to complain about a guy hired to come off the bench who has been forced to play big minutes and done so productively. He rates as one of the Nets' better defenders (+1.1); but on a per-possession basis, his shots, points, assists, rebounds, and steals are all below his career averages.

Andre Drummond (987), +0.3. Another trade throw-in who looks more productive than most of the Nets' holdovers. His excellent defense (+2.1) features a remarkable 24 rebounds per 100 possessions, but also 2.4 blocks and 3.1 steals. The main limitation on his playing time with the Nets may be his fit with Simmons.

Blake Griffin (880), +0.2. Despite his shooting woes (.505 true shooting, down from .610 last season), Blake rates as a surprisingly productive offensive player (+0.8). Misleading? Perhaps (though he has not played more than 56% of his minutes with any one player, or more than 28% with any other threesome). Or perhaps veteran savvy, setting picks, and making the right passes matter.

Kyrie Irving (507), +0.1. Kyrie is such a gifted offensive player that he can just step on the court and contribute superbly (+2.5). Defensively, not so much (-2.4). By comparison, in his prior two seasons with the Nets, his offensive rating was +3.4 and his defensive rating was -0.1. Will he be near that level by playoff time?

Nic Claxton (541), -1.4. Many fans have come to recognize that Claxton is a liability on the offensive end (-0.9); he is a career .286 shooter from beyond three feet. Fewer fans have recognized how little a switchable perimeter defender adds to a team with so many other defensive holes (-0.5).

LaMarcus Aldridge (911), -1.4. Aldridge has played a bigger role than many fans expected, and is shooting well above his career rate, but his defensive limitations (-1.3) have become more glaring as the season has worn on.

Bruce Brown (1099), -2.5. A good defender (+0.4) but a black hole on offense (-2.9). He's looked energized since the trade deadline, so perhaps as the offense evolves he'll get back to last season's level (-1.0).

James Johnson (858), -3.6. Another vet pressed into service due to injuries. He has some unusual skills, but is an overall liability on both defense (-2.2) and offense (-1.4).

Joe Harris (423), -4.4. Harris's rating is a reminder of the fragility of the metric when minutes are limited and unrepresentative (all from the first month of the season, mostly with the same lineups). If and when he comes back, who knows?

The Rookie Ladder

"Why, oh why, doesn't Nash play the rookies? They couldn't possibly be worse!" Well, yes, they could be. Minutes are very limited for Sharpe and Duke, especially; but based on what they've shown so far, none of these guys is really a productive NBA player--yet. RPM has no idea that they are rookies, yet it zeroes in on their "rookie mistakes," especially on defense.

Day'Ron Sharpe (379), -1.8. How can a guy who looks like a stud and rebounds like crazy grade so poorly (-1.9) on the defensive end? 7.7 fouls per 100 possessions is a start. And garbage time features a lot of helter-skelter defense, which is not his strength.

Cam Thomas (969), -2.9. Thomas has had some great moments, but we're measuring production, not swagger or potential. Even with his recent hot streak, he's shooting 27% from three. And even more than Sharpe, he's given up much more defensively (-3.3) than he's contributed offensively (+0.4).

David Duke (320), -3,8, Like the other rookies, Duke has been bad defensively (-1.9); but he's been equally bad on the offensive end (-1.9).

Kessler Edwards (709), -4.3. Edwards' defensive rating is not bad as rookies go, -0.8. Alas, the offense has been a whopping 3.5 points worse with him on the court. Perhaps that's just a fluke; but he is 13th on the team in true-shooting (worse than Thomas and Brown). The numbers suggest he has a long way to go to become even a Brown- or Bembry-type bench contributor.

Some Ex-Nets

Gone, but not forgotten.

James Harden (1627), +3.9. Nets fans soured on Harden in his last weeks with the team, harping on his defense and turnovers. But even diminished Harden was a top-40 NBA player (and, given KD's absences, the most productive Net in terms of contributing wins).

DeAndre' Bembry (949), -2.0. Not surprisingly, Bembry looks a lot like Brown, another undersized but scrappy, high-energy, defender. If anything, he was probably a bit less bad offensively (-2.5 versus -2.9), due to his uncharacteristically hot shooting. Wish him well, though not against the Nets.

Jevon Carter (552), -4.7. You thought he was bad. RPM agrees.

Around the League

The top ten players based on these early ratings are Jokic (+11.4), Curry (+10.0), Embiid (+9.4), Giannis (+7.9), Tatum (+7.0), Paul (+7.0), Doncic (+6.8), LeBron (+6.5), Mitchell (+6.2), and Gobert (+6.0). A good deal of face validity there.

The top-rated offensive player is Giannis (+6.6), followed by Luka, Steph, Trae, and LeBron (all +5.4), with Mitchell, Jokic, Lillard, Embiid, and Morant rounding out the top ten (+5.3 to +4.2).

The 15 top-rated defenders are Jokic, Capela, Crowder, Mobley, Looney, Gobert, Paul, Robert Williams, Jaren Jackson, PJ Tucker, Horford, Jarrett Allen, Embiid, Will Barton, and Nurkic (+6.1 to +4.7).

In most seasons, the top offensive performers are more valuable to their teams than the best defenders. That's not really the case so far this season. Could be the effect of COVID disruptions or rule changes--or perhaps the defensive ratings will regress toward the mean as the season wears on.