As the world turned the page on another year, weāre just one more week into the NBA season. Weāre nearing the halfway point, and the All-Star Game is right around the corner. All the bizarre numbers at the beginning of the season ā the Los Angeles Clippers performing like an all-time team, the Chicago Bulls shooting well from outside, the Golden State Warriors looking mortal ā have regressed heavily.Ā The state of the league is pretty stable, though thereās still the possibility of a massive second-half change; those happen occasionally.
The common thoughtĀ before the season is that we were headed toward another Cavaliers-Warriors showdown in the finals, and that does still appear likely. But the NBA is fast and chaotic, and changes can quickly take hold. We now accept Giannis Antentokoumnpo is an MVP-caliber player, and that would have been unthinkable just a few months ago. Letās prepare for the second half.
Are there too many 3-pointers?
Thereās been some discussion bubbling recently about the continuing upward trend of 3-point field goals. The league-average ratio of 3-point attempts to all field goal attempts is just shy of one-third, and the 3-point-shooting center is no longer a rare beast. This is remarkably differentĀ than what weāve seen in the past, and the changes have actually been accelerating. We can all agree, for the most part, on why teams are doing this and why itās beneficial, but is this entertaining? Are we going to reach a point where this isnāt aesthetically pleasing?
The equilibrium point for 3-point rate is shaped by a simple fact: a 3-pointer is not the best shot in basketball. Instead a shot at the rim is the best shot, and going to the free-throw line is (usually) better too. So if the league traded all its midrange shots (from ten feet and out) for ones behind the arc, youād have a ratio of 0.56 3-pointers per field goal, or roughly 0.5 per true-shot attempt. I understand those are all assumptions, and that if you stop taking mid-range shots it can change the frequency of your open looks at the basket or behind the line; but this is just a rough estimate. Houston right now, for example, has a 3-point to field goal ratio of 0.46. Theyāre approaching the theoretical limit where youāre in danger of missing out on more valuable shot types, like free throws, if you keep pushing outside shots.

Thereās no correct way to answer a question about aesthetics. I understand the concern about variety. If all we see are drives to the basket or 3-point shots, events are easier to predict and, thus, more boring. But if weāre simply replacing other types of jump shots, I donāt think think this is a major issue. Thereās still a lot of action and shot types around the rim, including that awkward area from three to ten feet.
But if weāre sincerely concerned about the prevalence of the 3-pointers, then we need real solutions: move the line back about a foot and make it uniform all the way around. And yes, this would mean widening the court, which is my real dream for the NBA. We could allow more angles to attack and make shots that are worth 50 percent more a little tougher to convert. I doubt thatāll happen, so for now prepare yourself for a league where 3-point ratios approach half of all shots.
Westbrook: Triple-double update
I feel as though weāre not freaking out about this as much as we should: Russell Westbrook is still averaging a triple-double, heās got a bit of wiggle room, and itās January. A decade ago, John Hollinger, preeminent NBA statistical writer whoās currently using his voodoo magic on the Grizzlies to extract wins from stone, noted that the modern-triple double season was virtually impossible. This was in the era of young LeBron James and rebounding phenom Jason Kidd, who was actually the closest to accomplishing the feat. I
would think the increase in the pace of play explains much of Westbrookās numbers, or perhaps more missed shots from all the 3-pointers, but itās more than that: heās grabbing a significantly higher percentage of rebounds than Kidd. This is an anomalous individual performance, and while his teammates are conceding many easy rebounds to him, itās still unprecedented.
What happened to the Pistons?
The Stan Van Gundy era has not been a renaissance for Detroit. Theyāve got about the same adjusted point differential as last season with a lower win percentage, and theyāre only slightly better than two years ago. Theyāre still a lot better than they were pre-Gundy, but any other progress has been virtually nonexistent. Much can be tied to the stalled development of their prized big man, who was supposed to be Stanās next Dwight Howard.
Andre Drummond, unfortunately, hasnāt changed much at all in the past couple years, and heās been arguably worse than he was before Van Gundy. This version of Drummond is much more active on offense, but has traded that usage for efficiency ā heās gone from a true shooting percent of 59.3 and a usage rate of 16.9 percent in his first two seasons to a combination of 50.5 and 20.3. I understand the desire to expand his game, but crafting Drummond into a post player, a guy who canāt shoot free throws or pass well, downplays his strengths and has been largely a mistake.
Outside of their center, Detroit does have a handful of decent players, andĀ their best player, according to most advanced stats, might surprise most fans:Ā Kentavious Caldwell-Pope. He doesnāt stand out in any particular area, but heās a well-rounded, good outside shooter who plays solid defense. But heās no oneās idea of a franchise player, unless he takes a Jimmy Butler-esque leap. Reggie Jackson was supposed to be their best player, or ideally their second-best behind Drummond, but heās had a disappointing season. Tobias Harris and Jon Leuer have been good, but, again, theyāre not franchise cornerstones.
Before this season, the Pistons enjoyed a little bit of success with the combination of Reggie Jackson and Andre Drummond, torturing teams with pick-and-rolls. Jackson was injured, and he hasnāt been too effective this season. And the Drummond post-up experiment, as Iāve repeatedly stated, should be ended. He should be used like DeAndre Jordan, focusing on shots inside, screens, rebounding, and defense ā thatās okay. You can have one starter like that, and it plays to his strengths. The Pistons should perform better over the course of the season, but Iām concerned about their long-term success: who will they become, and which players will remain in their core? This isnāt enough to compete with the top classes of the league.
The three centers
Weāve had no change with the three-center conundrum in Philadelphia, butĀ I think one fact reveals the position: Joel Embiid and Nerlens Noel have only played a little under two minutes together, per Basketball-Reference[1.]. Thus, since they have not tested this big man combination in real games, I doubt this is the combination they want going forward. Thatās no surprise, but you donāt want to show your intentions and itāll be easier for other teams to buy Noel. Theyāll likely be giving up a young asset for a lot less than heās worth. But at the very least, can we see how the team performs with Embiid and Nerlens? Experimentation is fine; itās not like theyāre fighting for playoff seeding.
The Beard and the revengeĀ of the 50-point game
James Harden just had a monster 50-point game, where he tied Wilt Chamberlain for the most points with a triple-double. These ridiculous games have been oddly common this season. In fact, we may be in a bit of a 50-point golden age for the modern era. Weāve have seven so far this season, and weāre not even at the halfway point[2.]. Back in 2011, there were only two such games, and only three in both 2010 and 2012, respectively. This surge hasnāt been from just one player either, like Wiltās effect in the 60ās. All seven games this season have been from unique players.
James Harden (@rockets) finishes with 53 points, tying Wilt Chamberlain for the most points in a triple-double
ā ESPN Stats & Info (@ESPNStatsInfo) January 1, 2017
James Hardenās 50-point game was pretty special though. HeĀ created a huge chunk of his points thanks to some high difficulty 3-pointers, often well behind the line and/or right in front of a defender. Look at the outside shots he made at the end of the second period in this playlistĀ ā those were not easy. He had a huge number of assists, and those were contingent on some excellent shooting from his teammates, many of Hardenās assists were high-quality, drive-and-create passes. And his rebounds were important for the team: as the lead shot creator and passer, itās a more efficient and faster way to start the offense. It was hard to imagine a 50/15/15 game happening in the modern era, and now Iām wondering what else is possible.
538 and win projections
One of the more popular destinations for win projections for the rest of the season is FiveThirtyEight. I understand this, and theyāve been pretty decent with their coverage. Their projections are more involved than simple ones, like Basketball-Referenceās. But their pre-season projections are not performing well. Some of their wilder picks, like Minnesota for 46 wins and Chicago for 45, which completely diverged from most of the analytical world, are starting to look worse and worse. They may have had overfitting issues with their closest player/age curve algorithm, where they tried to predict future individual performance by looking at statistically similar players. So I would advise some caution when using their win projections. Iām not quite sure how their pre-season ratings affect their current ones, but it warrants mentioning.
Avery Bradleyās defense
Defense is particularly difficult to analyze and tease out the useful information. Even with more advanced stats,Ā weāre still largely in the dark, and for some players that means weāre missing the bulk of their contributions. We understand defensive rebounds, at least to some extent, and we have a wealthĀ of information about shot-blocking and rim protection; and then thereās the area of turnover creation, headlined by the basic steal stat. But what about the class of players who donāt excel in those obvious areas: the plus defensive guards with low steal rates?
The representative here is Avery Bradley, a known defensive menace with a good reputation but disappointing defensive stats.Ā In fact, the issue is illustrated by the range of stats in the table below. Starting from the left, you have a box-score metric, then a mix of +/- and box-score stats, then pure +/- models. He performs better when weāre more detached from the traditional box-score world. Of course, you can also see another pattern: his defensive stats have been plummeting, and opponents are scoring at a blistering 112.8 points per 100 possession rate when heās on the court this season (and just 98.3 when heās off.)
Table: Avery Bradleyās advanced defensive stats
Season | DBPM | DRPM | DRAPM (single) | DRAPM (multi) |
2011 | 0 | NA | -0.16 | -0.04 |
2012 | 0.3 | NA | 2.72 | 2.4 |
2013 | 0.2 | NA | 1.01 | 1.13 |
2014 | -1.4 | 1.22 | 0.65 | 2.07 |
2015 | -0.9 | 0.73 | 1 | 1.7 |
2016 | -0.3 | -1.24 | -0.5 | 0.2 |
2017 | -0.3 | -2.31 | NA | NA |
*Single season RAPM and multi-season included.
Bradley played his usual pressure defense against the Cavaliers in a game last week, but to no avail. The worst moment was probably when he was knocked over byĀ a screen, though in his defense it should have been called a foul. He played Kyrie Irving close again far from the basket, and after having problems fighting over the rare double screen, he recovers and takes out his frustration by nailing Kyrie with a hard foul in the paint. You can see the issue with his pressure defense, especially against a spread team like the Cavaliers: if the opposing guard can drive past him, he has an open lane to the rim. You had similar issues versus Russell Westbrook, and while Westbrook is a nightmare defensive assignment, you can see how he tracks too closely and gets blown by after a crossover. Russell missed that finger-roll, but he didnāt miss another one later in the game: watch how Westbrook can easily round the corner against a defender whoās guarding him tight on the perimeter.
Thereās some evidence Bradley was a good defender before this year, but his +/- numbers have never been spectacular. Heās limited by the easiest way elite defensive guards can shape that end of the court: he doesnāt force turnovers. Ironically, this is the one season where he does stand out in the box-score: heās rebounding on defense at an elite rate for his position, but his overall defensive stats are trending well below-average. Defenders of his type are hard to measure ā other no-stats defensive guards include Klay Thompson and Joe Dumars ā but with enough seasons you can get a sense of their impact.
Assist value by position
All assists are not the same. We all accept this fact, butĀ rarely do you see any statistical adjustments. We can get near universal agreement that an assist leading to a shot at the rim is better than, say, an assist leading to a mid-range jumper. And you may have heard a handful of remarks like, āRondo stat-pads, so his assists are worth a bit less.ā Or, āLeBron James has high-value assists because heās great at nailing 3-point shooters.ā Or maybe, āBig men assists are more valuable because theyāre usually to guys at the rim or past the 3-point line.ā But is that last one really true?
This question originated in a Nylon Calculus email chain, and itās one Iāve definitely heard before; it seems plausible too. Answering the basic form of the question is pretty easy. I have dunk and layup assists going back to 1997, as well as a decent index of positions for every player in that time frame too[3.]. I have ātweenerā positions too: youāll see some graphs with positions of 1.5 for combo guard, 2.5 for shooting guard/small forward 3.5 for tweener forward, and 4.5 for power forward/center. You can see the results in the graph below.

The differences are virtually nil across the board. The range of the proportions here are 0.364 to 0.387Ā ā basically, it means that over a hundred assists, the difference in the number of layup/dunk assists would be two. It doesnāt change if you look at more recent seasons either; itās a pretty consistent pattern. Thereās something resembling a legitimate correlation when you look at 3-point assists, as you can see below. But itās nothing world-bending, especially when you consider big men shoot significantly fewer of those shots anyway.

But shouldnāt we make some adjustments? Big men shoot at the basket more frequently, so there are fewer opportunities for them to assist on those shots. We can likewise do the same with 3-point assists to see what the pattern really is.
The adjustments were a pain, and some assumptions were made, but I wasnāt going for absolute precision[4.] ā I wanted to see if the myth was plausible. You can see the results below for dunk and layup assists. Since centers more frequently score at the rim, their passing targets are players who, on average, convert fewer dunks and layups. Thus, weād expect a lower rate of assisted dunks and layups. Centers, and big men in general, may actually offer a higher value of assists from a certain perspective. That straight line we saw in the first graph was from a confluence of factors.

Looking at the next graph, we have the opposite pattern. Centers rarely shoot outside shots, at least over this 20-year period, so their targets are more likely to be 3-point shooters. Yet they have only a marginally higher proportion of those assists over, say, small forwards. Perhaps these two adjusted graphs can be explained by geometry and basic basketball mechanisms. Big men are usually closer to the basket, so itās more likely they assist there too; and since theyāre further away from the 3-point line, they assist there more infrequently. You can factor in play types too. Outside shots are often assisted by driving-and-dishing or passes around the arc.

I would caution that these are all inexact numbers. Even the official stat, the assist, is mired with subjective calls and inconsistent scorekeeping. But weāre probably reaching when we say big men assists are more valuable. Theyāre not more likely to serve up assists at the rim, and even after some adjustments you have to concede that they should have more 3-point assists. Whatās important here instead is that assists should be evaluated player-by-player. Some centers are wondrous passers who use the high post like a pitcherās mound, finding targets with great accuracy; some have hands of stone who pass the ball out of the paint to a contested jump shooter with 3 seconds on the shot clock.
We canāt exactly say big men serve up high value assists more often. The truth, as usual, is more complicated.
[1. Looks like Philadelphia wants to experiment with the pairing after-all. Perhaps they found that the market for Jahlil Okafor was cold. Or perhaps they read my drafts.]
[2. Weāre at 8 now, thanks to Jimmy Butler on January 2nd. It tied the ārecordā for most unique 50 point scorers in a season.]
[3. The dunk leaders are obvious; itās all dominated by Dwight Howard, DeAndre Jordan, Shaquille OāNeal, et al. But the layup leaderboard is more fascinating. Random trivia answer: the player with the most attempted layups in a single season is Antoine Walker in 1997 (his rookie season. In retrospect, perhaps we should perceive this volume 3PT-shooting, layup creator differently; he was born before his time.]
[4. I had to assume that once a player of a specific position was out of the game, then no other player of that position was sharing the court too. I also simplified the positions into just five types: 1 through 5. Then I estimated an expected rate of assist types. I know positions are tricky and thereās lots of lineup mixing, not to mention usage and shot distribution concerns, but this is just a rough estimate.]