NBA Week 9 in Review: Festivus Miracles

I haven’t explained one aspect of the week in review series well enough, and this is more of a personal crusade that I hope catches on: this is numbers-heavy analysis *with* videos included and appropriately referenced. There are too many damaging discussions out there about how analytics is somehow hurting the game because it’s all about stats without watching the actual game, but the reality is that these analytics wonks watch as many games as nearly everyone. This is the false dichotomy fallacy: you can use advanced stats and watch the game. In fact, doing one informs the other activity, and vice versa. Thus, I decided to make a conscious effort to use real video, or snapshots, from NBA games because these numbers aren’t pulled from the aether; they’re pulled from the actual basketball court. And with that short explanation, let’s review a few miracles we saw from the Christmas week in the NBA[1. Most of these stats were collected after the week ended.].

C.J. McCollum: Latin for McNeck

C.J. has passed a famous and arbitrary statistical benchmark: he’s averaging 20.4 points per game. Damian Lillard has missed games for the first time in his NBA career, and C.J. has capitalized on this opportunity. He recently scored 35 in a win over the Kings, just a hair under his season high of 37 back in their season opener, along with 11 rebounds and 9 assists. He’s taking nominative determinism to an extreme with the shooting guard position, pulling-up for jumpers as much as anyone outside of Stephen Curry. But he’s been a real positive force for the Blazers, and he’s assisting on a fair number of his teammates’ field goals — he’s not a black hole.

Like many undersized shooters, CJ is adept at hitting three-pointers quickly off the dribble or a catch; here’s an example of two he hit in a row. He’s a master of the in-between game though with a varied set of floaters like this bank shot, which is essential to a small scorer without explosive athleticism because of all the seven-foot trees who camp at near the rim. He makes a ridiculous number of awkward, HORSE-esque shots. Here’s a step-back midrange jumper, which is a space he heavily uses. There are a number of shots that require clunky descriptions too, like a one-legged fadeaway floater he uses to shoot over Koufos. And here’s something that keeps a defender confused: after coming to a stop after dribbling, he doesn’t shoot right away and hesitates just long enough to make the shot unpredictable. It’s not a natural basketball move, and C.J. takes a ton of shots we would definitely call ill-advised. He’s finding success as a scorer, and it’ll be interesting to see how well his efficiency holds near an acceptable level[2. He’s shooting around 44 percent on long two-pointers, which may not be sustainable.], but he has the necessary skills — and the Blazers will need it if they want to continue chasing the No. 8 seed in the West, which is a statement few people considered before the season started[3. Except NylonCalculus, of course.].

The Ten Games of Kobe Bryant

Perhaps this is just a mere coincidence, but a week after Kobe Bryant announced his retirement he started playing much better, mostly because he actually started hitting his shots. For whatever reason, Kobe has refused to significantly alter his game as he’s aged, and it’s led to some sad basketball. At his apex, Kobe was taking many difficult shots, but it was in a different basketball environment where league averages were uniformly lower and he was athletic enough to get to the rim and the free throw line to boost his efficiency; what’s left now is a tired athlete with a endless amount of impossible shots, and it’s tough for a basketball player to find success with that shot selection. For the last ten games, this has worked, but there have been stretches in the past where he pretends to be a point guard for a while or locks down and concentrates on defense — and it doesn’t last. He always reverts back. So while we can enjoy this recent league average efficiency shooting and increase in assists on his farewell tour, I wonder if it’ll last. Or maybe Kobe changed right as he saw the end of his career come into view like a deathbed conversion.

Cleveland Cavaliers Slowing the Pace

In the most anticipated game of the season, the Warriors defeated the Cavaliers, but it wasn’t a typical win. This was truly a sequel to the finals as the game slowed down, the Warriors had their worst offensive rating of the season to that point according to Basketball-Reference, and Curry had an atypical shooting night. Interestingly, the Warriors also had significantly fewer three-pointers than usual, and about half as many fast break points per possession. Ironically, the defensively challenged Cavaliers give the Warriors fits through their defense, but their offense was shut down enough for a Golden State win. The Cavs were a ghastly 5-of-30 from behind the arc including 0-of-5 from Love and 0-of-6 from Irving. It’s conceivable Cleveland can win consistently playing this way as long as they make their outside shots. They just need to be healthy and have a way to deal with the smaller units from the Warriors — will they keep Love on the court?

Cleveland’s strategy for Curry, the head of the dragon, is aggressive and suffocating. Here’s a play where Australian pest Dellavedova sticks close to Curry, who tries to drive inside and sees Kevin Love quickly leave Draymond Green open to help clog the lane; it leads to a missed flaoter from Curry, who was crowded. The general plan is to rush Curry with an extra defender, preferably leaving one of Golden State’s poorer shooters open like Iguodala, Shaun Livingston, or Draymond Green (before this season at least) It worked in the finals, as Draymond was notably 5-of-19 from behind the three-point line, but with his playmaking and if he’s still shooting 40 percent through the rest of the season, then it’s not a long-term solution. You also can’t trap Curry at the top when Green is the center because then the lane is free, unless the Cavs commit by sending another defender (whoever is guarding Iguodala or Livingston) straight to the rim. It’s a delicate balance, but if the Cavaliers can continue to slow down Curry and hit their shots the next finals meeting could be more competitive[4. It helps that Curry had a minor calf injury before the game, but it apparently wasn’t a concern and Kyrie Irving wasn’t 100% either. But Curry missed some following games, so perhaps this game needs an asterisk.].

Reggie Jackson: Fluke Busting

Mr. Jackson played extremely well for the Pistons after being traded last season, but many, including myself, were skeptical. Many players can have a magical stretch of games that fool us into thinking they’re something they’re not. But he’s been excellent this season so far, even winning a Player of the Week award and helping Detroit reach a winning record, like they’re victory by a mere one point over the Heat before Christmas. Andre Drummond is getting all the attention, but Reggie is scoring more than he did last season for them and he looks like the kind of legitimate penetrating guard that works well in Van Gundy’s spread, pick-and-roll offense.

Philadelphia 76ers Drink the Phoenix Suns

There’s a cause for celebration: the 76ers won, and the D’Antoni era is off to a wondrous start. By a cosmic coincidence, the win was against the Phoenix Suns, who regarded the loss as a rock-bottom point. To add insult to injury injury to insult, Eric Bledose left the game at half-time with a non-contact knee injury, and it turned out to be a torn meniscus. The 76ers were short-handed too, however, as Jahlil Okafor missed the game mostly due to precaution. This was also the team’s first game with Ish Smith, a veteran point guard described by the franchise like John Stockton; they curiously used two second round picks to pick him up in a trade after passing on re-signing him last summer.

How did Philly win? The answer isn’t easily explainable, and it wasn’t entirely due to Ish “the Houdini of the Hardwood” Smith. Their assist rate was a little below their season average, and their turnover rate was a little above. They didn’t take any more three-pointers than usual, and they actually took 23% of their shots at the rim compared with 37% before that game. Looking at single game KOBE (expected shot percentages based on a number of factors like shot distance, defender distance, and more via SportVU), the 76ers did not have a substantially different game, and the Suns under-performed. But they did get to the foul line at a higher rate, and they obviously made more of their field goals. While the addition of Ish Smith and the recent return of Kendall Marshall give the team veteran passers who have led real NBA teams, another factor is that the team couldn’t use the poison pairing of Jahlil Okafor and Nerlens Noel. Seen in the table below, the team has been indescribably awful when those two are on the court, which is even more concerning given how many times those two have been in the starting lineup. They have been merely terrible with just one of those guys on the court, and at least they’ve defended competently with Noel[5. Strangely, the team is performing quite well without either of them on the court, shooting around 60 TS%, and it’s not because of some garbage-time warping stats because I tried filtering out 4th quarters too and little changed.].

When the team is without Okafor and they play smaller, there’s more room on the court for the ball-handlers to operate. For instance, Kendall Marshall slips in front of his defender in this video, and since Carl Landry is the only guy in the middle when help comes Landry is open and gets an easy shot at the rim. And for another example, when a pick is set high and Nik Stauskas drives inside, no other teammates are hanging down there clogging up the lane, leading to a layup. The team is healthier too and they have real shooters on the team, which is useful for a team that insists on chucking three-pointers wildly — when the Suns commit too many players inside during a break, one defender is left to cover both Hollis Thompson and Nik Stauskas, leading to an easy corner three from Stauskas. Nerlens Noel actually didn’t appear healthy to start this season, but he looks better now, as he finished well inside, including this nice alley-oop. The 76ers are a punch-line, but they were better last season and with some reasonable adjustments, like playing smaller and fewer minutes for Okafor, they could challenge the Lakers for the 29th best team in the league.

Table: Philadelphia 76ers stats via NBAWOWY

Unselds

Inspired by a few recent Kevin Love outlet passes, I thought I’d undertake a small study of these super-assists to see, systematically, who’s doing them more often and if there are any surprising patterns. Love is the reigning master at this, having learned the trade secrets from family friend Wes Unseld, but nothing is definitive unless there’s a systematic, objective evaluation. If nothing else, this is an example of how and why stats are far from perfection.

Using data pulled from play-by-play logs, which provide unreliable but necessary time-stamps, and after some experimentation, I have to define these outlet assist passes as the following: grabbing a rebound and within three seconds getting credit for an assisted layup or dunk. I attempted to include instances where the play started after a different action, like a made shot, but it’s impossible to determine from that data who the inbounder is and you get situations like this play where there are two passes. It’s unfortunate I can’t include this ridiculous Kevin Love assist, but basketball stats are a world of restraints.

For a quick look at the data, there’s a summary table below from the 2015 season, and naturally Kevin Love is dominating. But there’s something off with the results: there’s a lot of point guards, and I’m not sure if that’s because they’re good at throwing outlet passes or this includes conventional fast breaks. Much to my chagrin, the majority of the time these quick assisted buckets are made happen when a point guard aggressively pushes the pace after grabbing a board. But this is not what the Unseld is in spirit or else I’d call it the Magic Johnson.

Rebound-assists (2015)

Luckily, there’s the gift from the basketball gods: SportVU data, whose movement data can be extracted from the past couple years. From this, presumably one has the location of every player and the ball for every split second of every game. In practice, however, this does not lead to a pretty solution.

Let’s consider one reasonable filter I wanted to use. Every line of information in a play-by-play file is numbered, termed the event ID or event number, and correspondingly SportVU movement data is broken up by these events. With the game ID and the event ID, which, for example, is something like “player X 10″ jump shot (player Y assists)”, it’s seemingly easy to pinpoint an exact play and analyze it. You’re also given a time-stamp for each play. Armed with that data structure, I could find the exact second that each event takes place. If the rebounder isn’t within 47 feet[6. That was my initial solution, and it can obviously be improved. I chose 47 feet because it’s half of the length of the court and the reasoning is that if you assisted on a layup or dunk and you’re still 47 feet away, it was probably a great outlet pass. I could also offset the time by selecting something] of the ball when the field goal is made, then it counts as an “Unseld.”

In practice, a large number of games did not have any available data, and then around half of the remaining candidates didn’t have any matching seconds. There are some other solutions like widening the movement data window so I’m looking at a larger swath of time and can then filter the appropriate time segment, but some plays were just plain miscoded and it’s tough to systematically find a fix that can apply for every play.

For an example of what these assists look like in SportVU form, I’ve included two graphs below. The first is one where Kobe Bryant grabs a rebound and travels three-quarters of the court before throwing a nifty pass. The second one shows Duncan grabbing a rebound near the rim and firing an outlet pass quickly to Manu Ginobili at the other side of the court. I included a graph showing the height of the ball as it screams across the court too, as the ball, but no players, have that extra dimension in the data. It’s apparent from those views how one is a true “Unseld” because Duncan is separated from the ball so early on, but it’s not always so clear because events bleed into one another and time-stamps are unreliable. If you know exactly when the team has possession it’s a lot easier to work with, however.

I’m a few years late in trying to track the Unseld with public data as Bill Simmons once requested, but this is progress and with better data, or better techniques, it’s entirely possible[7. I may also just go crazy and track these manually, which is quite a lot of work because the 2015 season had nearly 200 candidates.].