NBA Week 8 in Review: Against the Night

As the northern hemisphere passes the darkest day of the year, the NBA season is maturing and the vaunted Christmas games are almost upon us. Most storylines are established and most contenders are beginning to dig in — now it’s a seeding race, though there are always a handful of surprises. For those opening this time capsule in the future, at this point in the season the Warriors and Spurs are dominating the West while people just assume the East is Cleveland’s with a remarkable lack of consensus for the next best team. People don’t understand what to make of Boston, Chicago’s offense is a mess, and Indiana has shed its old identity entirely. So, as always, let’s look back at the past week and make sense of a few random topics.

The Great Staggering of Durant and Westbrook

With the highly criticized Scotty Brooks gone, people had hopes that the Oklahoma City Thunder would maximize their skills and improve their player management. But Billy Donovan hasn’t been an improvement, and the team has been doing one curious thing: they’ve gone with bench-heavy units for long stretches, often without Kevin Durant, Russell Westbrook, and Serge Ibaka. Through last week, Durant and Westbrook have played 636 minutes together — yet Durant has only played 742 for the entire season.

The idea here is that teams should stagger the minutes of their stars so that one is always on the court when the game is still relevant. Presumably, this helps the team because shot creation duties won’t be forced onto role players or leave teams vulnerable to overmatched bench units playing against starters. Research suggests there can be diminishing returns to high usage players playing together as well. Thus, OKC would probably get more value if they play as many minutes as possible (while the game is still competitive of course) with at least one of Russ or KD on the court. But looking at the stats so far, OKC is doing exceedingly well when both players are on the court and it’s larger than the sum of the parts (i.e. you can look at Durant and Westbrook when they’re on their own and estimate what they “should” add up to together.) If you want a name for that phenomenon, I believe synergy qualifies. I even used Ibaka as a constant with years of data and found similar results.

And so begins the eternal struggle: theoretical versus empirical data. This probably deserves a column of its own because staggering is a practical problem, and there are many adjustments to make to figure out the ideal substitution patterns of these two superstars. As a final note, the Thunder staggered their two stars a lot more than usual on Monday — and they beat the Clippers.

Four Overtime Fun

Most normal NBA fans don’t know how uncommon these four (or more) overtime games are, but they actually are quite rare. Since the lockout season in 1999, there’s been only one other game with four, and since 1964 there are only nine such games. The record is six overtimes back in 1951 between the Indianapolis Olympians and the Rochester Royals, which was pre-shot clock, explaining how the final score was 75-73 even though the game lasted 78 minutes. Thankfully, this last super-long game ended with a score of 147 to 144 and featured a handful of great plays. Naturally, Jimmy Butler was involved and played 56 minutes, and he’s one of only two players with a 60 minute game since 2013. Somewhere the ghost of Tom Thibodeau is laughing.

Old Bears

The Grizzlies have been hanging around the 0.500 mark in the past week, and I assume that’s far from the team’s goal for this season. They re-signed Marc Gasol and they’ve been adding small pieces to their grit-n-grind core, attempting to reclaim their defensive glory while bolstering their offense. But they’re at a crossroads now: do they hang onto their main group and improve at the edges, hopefully adding another high quality player, or do they cash out now and rebuild for the future?

Not too long ago, the Celtics decided they had gone as far as possible with their core, and they traded their veterans, still producing, for young players, draft picks, and cap space — even jettisoning  the beloved Paul Pierce, who had been a Celtic his whole career. The Nets were a team desperate for relevance quickly, and they paid a premium — this is what Memphis should hope for. Perhaps they can trade Tony Allen, Zach Randolph, and a couple other role playing veterans for a contender needing a boost. Allen is borderline unusable against smart teams, but he’s a useful situational player against star wings, while maybe another team will think Randolph can be used as a counter against the Warriors, however right or wrong that may be. That’s the fast rebuild option because they can keep Mike Conley and Marc Gasol, but maybe it would be best to trade one or both of those two while their values are still high. The nostalgic desire to hold onto your veterans past their due date is not conducive to building a strong team for the future, however painful that process is.

The Materialization of Bobby Portis

Chicago already has a stuffed frontcourt, and they don’t need another headache in finding time for a player. But there are few complaints allowed when the 22nd pick in the draft has a good game. Bobby Portis was dubbed a jack of all trades during the draft, but a skillset of good finishing, a jumper, and solid rebounding has some utility. Against the Knicks, he recently had a good game with 20 points, 11 rebounds, and 2 steals. He has the athleticism and quickness to grab a steal and run the break to tip in a miss. He also has enough of a floor game to go from stand-still in no-man’s land at around 20 feet into a drive to the basket for a layup, although he looked a bit awkward.

What’s really alluring is his range, as he nails a three-pointer here popping off from a screen. With his size and mobility, he has a chance to be a stretch 4 with plus defense and rebounding, which would solve some of Chicago’s issues with spacing while keeping their defense intact. It’s still really early in his career, and nothing definitive can be written about him yet, but he’s someone to track as the season progresses.

Unappreciated Love

Kevin Love is a lightning rod for criticism due to a bit of ambient LeBron James hatred and perhaps an inability to see past his lack of size and athleticism. But if nothing else, we should treasure him for his outlet passes, which are consistently amazing. Some players will unleash a monster touchdown assist, but only rarely. It’s impressive how often these are converted and not turned over. For example, in a game last week against the Magic the pass gets tipped but it still leads to a bucket. And this guy is paired with LeBron James running the break. How blessed are we?

The Somber Decline of Omer Asik

Over in New Orleans, Anthony Davis is being thwarted by his teammates once again, and it’s often because of starting center Omer “stone hands” Asik. The team is on track to hit the lottery, and Asik has been an anchor for their offense: he’s pulling them down, down into the abyss. He makes it tougher for Anthony Davis and others to operate in the half-court because there’s no reason to guard him unless he’s standing under the rim, and even then there’s a chance he’ll drop the pass, like in this play where he’s wide open right under the basket and still nearly loses the ball. He’s been inept on offense, witnessed in the vine below and this clip, and his defense isn’t good enough to cover up the gap. It’s no surprise there are rumors he’s now on the trading block.

Unfortunately, this is how people think of Omer Asik now. But he used to be a really useful, underrated player. He had two great seasons as a part of Chicago’s bench mob playing suffocating defense, and the bench with Asik usually played better defense than the starters. It wasn’t a fluke either because he had one great season on defense with Houston. They were a bit above average when he was on the court, but they were awful when he was off it, and more advanced numbers verified his defensive value. But after Dwight Howard arrived and he suffered through a knee injury, he hasn’t ever been the same and he’s been aging like fellow stiff Kendrick Perkins. This season has seen a precipitous decline where his rebounding and shooting stats have fallen while his turnovers balloon. His usage rate is scary low — in the single digits — so much so that he’s truly a liability on offense. He shouldn’t be starting for an NBA team as long as he’s playing like this, and he needs a greater mobility for defense to justify his lack of other skills.

The Role of 3PT Luck in Plus/minus Metrics

Nylon Calculus has discussed 3PT luck extensively, but at a certain point theory needs to be put into practice. Three-point percentage defense is nearly an oxymoron; the best way to defend the three is to prevent it. Breaking this down mechanically, what’s happening is that players have little influence on most three-pointers because they’re usually more open than other shots, and this is backed up by my own independent research through a couple different techniques. As a consequence, individual defenders should not be graded on how well outside shooters hit their shots. This is especially pertinent with plus/minus metrics because they rely on how teams outscore each other. But how can this be implemented?

As a first step, can 3PT luck even be detected within these metrics? With some mathematical techniques and a wealth of data, perhaps the effect is so small that it’s negligible. For the best possible numbers for testing, I’m looking at the 2014/15 season because that’s the season for which there’s a full season of single-year DRPM available. It’s single-year only because other seasons would muddy the results, and I’m only looking at single season opponent 3PT%. For another baseline, I’m using DBPM via basketball-reference. It’s not perfect because it relies on the box score, but it’s a different measure and it aims to replicate DRPM without using play-by-play data.

For a simple yet effective metric in 3PT% luck, I calculated the disparity between the league average 3PT% of 35 and the opponent’s 3PT% while a particular player was on the court using data from stats.NBA.com. Then I multiplied it by the number of attempts and converted it to a per 48 minute scale[1. Excuse the use of a per minute stat mixed with a per possession RPM stat, but this is just a quick feasibility study.]. Positive results mean that a player is getting lucky; negative means unlucky. The results are in the plot below. What’s fascinating is the huge domain. From direct calculations, some players are getting hit by 4 points per 48 minutes just from 3PT% volatility. There’s also a slight positive relationship with the data, which means that players with a high DRPM are getting slightly more lucky than other players. The correlation coefficient between the two stats is 0.317.

Of course, it’s not entirely all luck, and it’s still a question of what exactly can be controlled, if anything. Some players may have aided in forcing the opponent into tougher shots — one can’t tell with these results. But for an interesting test, I’ve made the same plot but with BPM instead of RPM; this is purely box score stats and not plus/minus. The relationship has virtually vanished. Thus, it suggests, there exists some 3PT% defense volatility wrapped up in play-by-play plus/minus data.

For a better illustration of the futility of 3PT% defense at an individual level, I’ve provided the best and worst performers with at least 300 opponent attempts in 2014 and 2015. If this is indeed a skill, we should see some of this carryover into an adjacent season. But this isn’t the case: the players who look great in 2015 were, as a group, around average the year before with results all over the place. The same is true of the other table.

Table: best performers 3PT% defense in 2015

Table: worst performers 3PT% defense in 2015

While it’s apparent now 3PT% defense can’t be trusted, the fix is complicated. One can’t just delete the “excess” three-pointers because it changes a game that has already been played. Teams play differently after made shots than missed ones, and one can’t ignore offensive rebounding, which happens about a quarter of the time after a missed three-pointer. There’s also the same amount of this defense that does stay constant, due to, say, teams forcing tougher three-pointers, thereby lowering expected value of shots and over a long enough time frame the actual percentages, but this is likely only a small effect[1. The gap between best (so far this season Miami, contesting 67.5%) and worst (OKC at 53.6%) teams in terms of contesting three pointers is about 14 percentage points, which translates to about 2 or 3 extra uncontested shots per game depending on the number of attempts allowed. That difference in turn translates into the expectation of one extra made three about every five games.]

Based on the correlation between DRPM and 3PT% luck and combined with a conservative estimate, one could say about 50% of that difference can be applied to a plus/minus metric or possibly more. But this needs further testing with years of data. Finally, here’s an example of the results below with the top players last season with an adjusted measure. The changes are modest, but notable. Khris Middleton and Jusuf Nurkic were probably being overrated, while Rudy Gobert and Tim Duncan were apparently being underrated by a fair amount. Off the table, Hassan Whiteside saw his rating increase by 1.2 points, which makes more sense given his prodigious shot-blocking. The changes for a few Memphis players are sensible: Marc Gasol leaps ahead of Kosta Koufos, while Nick Calathes sees his strangely elite DRPM rating fall by 1.5 points.

All hail Lord Duncan.