Another week of the NBA is in the books, capped off by Martin Luther King Jr. day. (Yeah, I wonāt address the ludicrous ending to the Clippers-Rockets game just yet. We need to see the repercussions from infiltration-via-Swiss-distraction kerfuffle.)
The holiday has become one of the most visible days for the league with many games and multiple ones nationally televised. Obviously, the NBAās large share of African-Americans for most of its existence is the cause, and this year got fairly political. This is not a league where theyāre looking forward to visiting the White House after a championship. Multiple people associated with the NBA had something to say about the current environment in the USA, and Iāll close this preface with a quote from the surprisingly erudite/woke Stan Van Gundy.
Pistons coach Stan Van Gundy on the idea of "racial progress" in the 50 years since MLK's assassination (via @MarcJSpearsESPN) pic.twitter.com/UJQ0QZkCps
ā Vincent Goodwill (@VinceGoodwill) January 15, 2018
Lou Williams is on fire
The Clippers have seen a resurgence as of late. They still have a few new guys in street clothes, but Blake Griffin and Milos Teodosic have been (mostly) playing. But they havenāt been the primary catalysts ā Lou Williams has been on fire for weeks. Since his game against his last team, Houston on Dec. 22, and through last Sunday, heās been averaging 31 points per game with a 66 true shooting percentage. Heās been hoisting 20 field goals up, half of which are 3-points at 46 percent, and heās had a 93 percent free-throw rate to boot ā with how often he gets to the line, thatās huge. The All-Star talk concerning him is a bridge too far, but heās tearing up the court for another Sixth Man of the Year.
The crown jewel to this hot streak was his 50-point game in a blowout win over the Warriors ā for virtually anyone, thatās the game of the year. (Okay, so he did it cheaply with a 3-pointer with a few seconds left in a blowout win, but thatās still some tough competition.) You can see how he did it in the video below. They were not easy shots. Heās one of the best practitioners of the off-the-dribble, contested 3-pointer. Only James Harden (whoās lapping the field), Damian Lillard, and LeBron James have more unassisted 3-pointers, per pbpstats.
Whatās remarkable is that Lou Williams is accomplishingĀ this peak season as a 31-year-old, 6-foot-1 combo guard. I imagine most people would have guessed heād see his best seasons in his mid-20ās, but heās gotten better by expanding on his skills, becoming a deft, crafty scorer who can burn other defenses from every spot from the arc and in. The secret to his efficiency ā and this eludes many scorers of his ilk ā is that he actually gets to the foul line, despite his size. Heās improved even in free throw shooting in his old age, and as a result youāve got a guy hitting a 61 true shooting percentage with a 30 percent usage rate. Thatās the kind of rare scoring season you usually see from legends or at least All-Stars ā the only guy on that list who wonāt be going to the Hall of Fame is probably just Isaiah Thomas. Most scorers-trapped-in-a-point-guardās-body donāt make it this far or develop this much. He should be lauded, and itāll probably come in the form of a Sixth Man Award.
Single game plus-minus
Gregg Popovich, that lovable curmudgeon, had some comments recently about that new box-score stat, single game plus-minus, calling it useless and dangerous for the impressions it creates about players. As someone who values plus-minus, who checks its various incarnations regularly, and as someone whoās calculated RAPM before, I ā¦ agree. Single game plus-minus is bogus. I donāt even trust single season raw plus-minus. Thereās too much noise. Iāve also been seeing that single season form used frequently from many different types of people. Since adjusted plus-minus was first implemented about a decade and a half ago, Iām sorely disappointed weāve lost that progression, weāre just using the raw, unadjusted forms, which are objectively worse by a sizable difference. But hopefully we can find someone to bring the superior form, RAPM, roaring back into NBA useā¦.
Zach LaVine returns
The Bullsā main prize in their Jimmy Butler trade finally made his debut last Saturday in a game against Detroit. Since Butler is, once again, somewhat quietly having a fantastic season, LaVineās value will be important if they donāt want to totally regret the trade, even with how wellĀ Lauri Markkanen andĀ Kris Dunn are playing. Giving up a superstar hurts. So far, the LaVine experience is largely the same, although heās probably too hesitant (and recovering) for any electrifying dunks. He scored well, shot from distance, but he wasnāt much of a playmaker and his defense was .. well, an optimistic would say he was extremely rusty.
Bulls fans noted the latter point, and itās the biggest concern for him going forward. The ACL injury, of course, didnāt help, and he just hasnāt shown the kind of awareness on the court to defend even at a competent level. You can see an example of it below (thanks to Blog-a-Bull for finding it.) For the rest of the season, however, be careful with his defensive on/off numbers. Remember, those are extremely noisy, and partial season samples are not to be trusted unadjusted. Iām fearful some noise may come into play and people will argue heās finally good on defense ā this is a time where we should indeed watch the games. If they want to sign him to a long, large contract, his defensive issues need to be addressed.
https://twitter.com/iamvega1982/status/952369354217271296
The continuing Markelle Fultz saga
Update to the Fultz drama: we still have no idea whatās going on. Apparently the 76ers donāt either, as Brett Brown had a few words to say about their number one pick that boiled down to, basically, āThis is very weird and I donāt know whatās going on.ā Was it the shoulder injury? Did he just suddenly lose his jump shot, a psychological failure? Did he try to change it ā because, you know, good timing coming into a highly competitive league ā and the results failed? Are we living in a Berenstain universe where we have all just forgotten that his jump shot always looked like this? I donāt know. No one knows. But itās probably some combination of factors, like a shoulder problem that led to a tweak of his form, which led to a mental collapse when he had troubles converting a shot. Heās healthy, but heās still not on the court because, as Brett Brown explained, ā[H]eĀ needs to be is able to shoot a basketball.ā Letās hope this gets resolved neatly.
Nikola Jokicās defense: Yeah, itās not as bad as people think so please make him an All-Star
When listing candidates for the All-Star game, you usually see the same set of statements about the Denver big man. Thereās something about his otherworldly passing, his place of origin, maybe his scoring abilities, and then about how his defense hurts his case compared to others. But is that true? Over his past three seasons, he was rated well on defense by ESPNās RPM. The first year you could have argued it was a fluke, but heās been steadily above average for the third year in a row. Going by the box-score stats, compared to other centers, heās been above average on the boards (and Denver has been above average too as a whole), near the best in steal rate, but his blocks are definitely low. That, coupled with his European background and his athleticism, probably paints him as a poor defender in most peopleās eyes.
Digging deeper into the numbers, his rim protection stats look awful. The team usually allows a higher percentage at the rim when he plays, but they also give up fewer shots at the rim with him on the court too ā and thatās consistent over three seasons. And yes, he does foul too often, but they donāt give up more free throws when heās in the game either. Hereās my theory after watching him play: like many big men out of eastern Europe, heās a smart defender with sneaky value. Vlade Divac was like this, and so were lesser players like Zaza Pachulia. They know positioning, they know where to put their large frames even if more athletic guys can jump over them, theyāre good post defenders, and they usually have a few dirty tactics. Jokic is no Divac, but heās got enough defensive chops, and quick hands, to overcome his lack of quickness and verticality. Defense may not be quite the liability people think, and that should (in theory) boost his All-Star status to the no-brainer tier after the starters are picked.
The upside-down Charlotte Hornets season
The Charlotte Hornets are in a rough position for the playoff race, a few wins back of the No. 8 seed. Theyāre having problems keeping above the rebuilding cellar dwellers, like the Nets. Despite the makeover of the franchise and all their discipline in rebounding and keeping turnovers low in Steve Cliffordās system, and the development of Kemba Waker into a multiple-time All-Star, theyāve only had two playoff berths in his tenure and this is the fifth year. Weirdly, they have a point differential of a near 0.500 team, which would put them squarely in the playoff race. Whatās going on? Should we trust the point differential more or the win percentage?
Going to the histogram below, they have a strange distribution for point differential. They have a large number of losses with a margin of victory around negative ten points, while they have few close wins with more blowout wins. Thatās odd ā you donāt usually see a bimodal distribution in the NBA, and I doubt it sustains. Itās also a great recipe for losing a large percentage of games with an okay point differential. I would not bet on them being a super āunclutchā team thatāll keep losing close games ā they have none of those markers. They take extremely good care of the ball, they have Kemba Walker to control things in the half-court, and they have a solid defense (for a bad team.) But even so, we expect more progress than that.

Outside of Kemba Walker, the roster has been a disappointment. Michael Kidd-Gilchrist was a defensive wunderkind with a bright future; heās stalled because he still cannot make a shot. Heās major drag on their offense, one that unfortunately has no clear number two option. Dwight Howard is their second-leading shot creator, and, all due respect, that is not a great sign. Cody Zeller has been injured for weeks, and Nic Batum has had his worst season since he was a very young, skinny wing in Portland. Itās not all been negative, like how they made a competent player out of Jeremy Lamb, one of their leading scorers. But they need a lot more help to compete in the Eastern Conference, which is now more competitive. They should start winning more close games and break away from that bizarre point differential distribution, but it may not be enough at this point in the season.
Measuring spacing from way downtown
People who say the analytics ānerdsā need to get off their spreadsheets and watch more of the game donāt understand how it all works (not to mention theyāre using code from R/Python/other programs over spreadsheets.) Hereās how a lot of stats are born: you watch a game and notice something that piques your interest. Maybe itās Robin Lopez boxing out for a guard, who gets credited for the rebound, on multiple possessions.
You can hit the data and see how his teams do with rebounding with and without him, compared to perimeter players who actually out-rebound him like Justin Holiday, Kris Dunn, and Carmelo Anthony last year. His teams actually rebound better with him, contrary to what youād believe from his defensive rebound percentage. Then you go back to the video and find that, yes, he doesnāt grab the easy rebound and lets someone else take it. Then you notice other players do this, like Steven Adams, and dig into the numbers of other players, and notice a trend. Big guys are often underrated (in terms of rebounding) by box score stats, while perimeter players with surprising rebound rates often donāt move the needle on the boards. Now you have a hypothesis thatās been validated with other personnel, and it changes the way you view games and how you watch rebounding.
Basketball stats arenāt meant to be consumed in a vacuum, and theyāre usually credited by curious minds who watch a lot of games. That leads me to this weekās inquiry: Houston not only takes an ungodly amount of 3-pointers; they take many from far behind the 3-point line too. Iāve written about this before, and itās a great market inefficiency. No one else was thinking about it, but you can generate more open shots because fewer people will guard you there and youāre pulling defenders further away from the basket ā and when you have James Harden and Chris Paul, you can stress the defense so much they break.
This can be quantified too. When SportVU tracking emerged, we had some new terms enter the NBA lexicon, like spacing and gravity. Teams get coordinates for every player (and the ball) on the court, and while there are many ways to quantify the spacing of every player, Iāll focus on two: distance from the ballhandler and the convex hull. The former is obvious, so let me explain the latter. A convex hull is like drawing a bunch of straight line to connect the points to create the smallest area possible. The convex part means the perimeter canāt go inside a line created by two points ā i.e. no dents or indentations. Itās easier to see in the pictures below. Itās been used before by Stephen Shea, a math professor and sports analyst; Justin Jacobs, a statistician and current head researcher for the Orlando Magic; and I imagine many others.

Letās move to some concrete examples. Iāll start with a simple set-up. Letās imagine two players in the corners, one on the right wing, one on the left, and one player in the high post, presumably about to set up a pick. To make things easier, letās ignore where the defenders would position themselves ā that all depends on their respective gravities, and thatās not the point here. You can see the setup below.

How do you quantify that spacing? If the guy on the right wing has the ball, the average distance from every other player to the ballhandler is 28.6 feet. That seems pretty good. And if you calculate the area of the convex hull created by those five players, you get an area of 699 feet squared. Need some context for those numbers? If you move the guy in the left corner to the low block ā this really cramps the spacing on one side of the court ā you get 24.7 feet and 543 squared feet, respectively.
Now what happens if you, say, move the guy on the left wing back a mere three-and-a-half feet to 28 from the basket. Plenty of the Rockets shoot from there, and itās just one player. The average distance increases to 29.2 feet and the hull to 770 square feet. Thatās a sizable change; the distance alone from the player at 28 feet to the ballhandler changed by two-and-a-half feet, which can be enough to kill help on a drive. The convex hull didnāt change by the same amount that sending the corner guy to the post did, but the difference was nearly 50 percent of that. Thatās impressive for just moving a guy back 3.5 feet.

This kind of spacing is a truly outside-of-the-box type of thinking. You canāt just think of the 3-pointer as a shot you take on the arc; you have to think beyond, and imagine all the cascading effects. Next time you watch a Houston game ā and yes we of Nylon Calculus watch the games ā take note of how often they shoot from behind the 3-point line, and, most importantly, note how defenses guard in those sets. Opposing defenses are being stretched beyond their capabilities, and the Rockets play some of the best offense youāll see in NBA history. And one quirk in positioning, something you can notice while watching the game, helped them accomplish that feat.
Next: Nylon Calculus -- How have team offensive play types changed?
By the way, if you want to calculate the convex area too ā I know everyone on the playground is talking about this ā you can follow the code here in R and use the coordinates below as a starting point where the left corner is 0, 0 and the basket is at 0, 25. And yes, thatās some trigonometry. Itās easier to think about locations around the arc in terms of straight lines from the basket from which you can find the x-y coordinates using good old triangular functions. You just gotta pick out your angle, and remember the distance from the rim is the hypotenuse.
x1 <- c(0, 24.5*sin(pi/4), 24.5*sin(pi/4), 15, 0)
y1 <- c(2, 25-24.5*cos(pi/4), 25+24.5*cos(pi/4), 25, 48)