NBA Week in Review 19: The Specter of Spring

The NBA regular season isn’t far from its final stages, which means soon we’ll prepare for the playoff grind and dream of free agency and the draft. But there are still many high quality basketball games left and a few playoff spots are set in stone. Spring brings the race for the playoff seeds and the dance of the post-season. It’s only a few days away, but we’re still in winter — and let’s take a look at the past week in basketball.

David and Goliath

In chasing 73 wins, your team is a target and every game is a circus. The NBA is competitive and there’s more parity than one would think — thus you get games like the Lakers upsetting the Warriors, which is definitely a surprise but still (marginally) within the realm of possibility. For NBA stat-heads, the game was a fun exercise in numbers. The gambling spread lines were as wide as 18 points before the game, even though Los Angeles was home. The numbers from FiveThirtyEight stated the game was the 23rd biggest upset ever, but ignoring homecourt advantage it was the biggest upset ever — there was no other game listed where the difference in Elo (team strength rating) was as large. Using this win percentage formula here and some team stats before the game, the Lakers’ odds of winning were actually near 6 percent. That seems a bit high, but that’s without the knowledge of how healthy the Warriors were and how good they’ve been in clutch situations.

Digging into the numbers further, there are a handful of other teams in NBA history with an SRS of at least 10. Looking at games they played against teams as awful as the Lakers (an SRS of -9 or worse), the Warriors were the first team to lose such a lopsided game. That’s a win percentage of 2.9% for the David’s here, but the Lakers had the advantage of being home. Adjusting for that, and the Lakers would have 4.0% odds of winning — close to 2.9%. There are only 35 games like this in NBA history, so that appears to be a reasonable estimate and a loss like that eventually will happen. The Warriors were 4 for 30 from beyond the three-point line — that’s the kind of luck required for an upset of that magnitude, and it had little to do with Los Angeles’ defense.

In fact, one could argue that the Warriors have been exceptionally lucky this season. They’re currently leading the league in opponent three-point percentage, which is not a stable stat — it’s mostly luck. They have 11 wins in 12 games decided by five points or less, and they’re also perfect in their five overtime games. Their clutch stats are league-leading, of course, and they’re ridiculous. While Curry’s magic and the power of the death lineup explain some of the clutch performance, generally all those stats do not transfer over to the next segment of games, like the following season or even the playoffs. I’m not trying to denigrate what the Warriors are doing, but it’s useful to point out. And if you want to break a record like 72 wins, you need all the luck available.

Giannis: Legs and Arms Akimbo

There are few players with the League Pass and Twitter potential of a guy like Giannis Antetokounmpo. But he has legitimately improved, and he played superbly last week. For instance, he had a triple double against the Thunder last Sunday. Against the human tornado that is Russell Westbrook, he had several plays like this one where he grabbed a rebound (or a steal) and sprinted down the court for a quick two points. But he’s seen growth as a playmaker, where he has a long career ahead of him as a point forward. For instance, here’s a nice touch pass to Greg Monroe, who slipped inside; lesser inexperienced players hold onto the ball too long.

I would, however, be excited about his defensive development. Here Dion Waiters gets isolated against the Greek Freak, and makes a mistake by trying to challenge him, as Dion is wont to do. Giannis gives him a little room, Dion drives, and he gets swallowed pretty easily inside. But Giannis guards frontcourt players too — Serge Ibaka posts him up here and gets nowhere. While that’s only Ibaka, it’s pretty normal to see him guard power forwards and then switch often onto quick guards. The Greek Freak is still all legs and arms and potential, but he’s improving more and more on both ends of the court. This play, I believe, is the full Giannis experience: he grabs the rebound, pushes it hard up-court, nearly loses the ball, resets with composure, makes a nifty guard-like move to get inside, and draws a foul at the rim with his long arms. Giannis Antetokounmpo — not only does he possess all the letters but he plays every position and role too.

Rise of the Raptors

Toronto spent much of last week chasing Cleveland in the standings, and they’re still dangerously close. The East was in dire need of a challenger for the reign of the Cavaliers, and they’re at least good enough for an interesting series — the Kyrie Irving-Kyle Lowry dogfights could be classic. They’ve accomplished this with one of the best offenses in the league and a capable defense, which is much better with some of their bench units. As I’ve outlined before, they’re primarily driven by Lowry, not their leading scorer DeMar DeRozan. This season is no exception. Raw plus/minus or on/off stats can be wonky, but I’d still like to point out their offense has actually been better with DeRozan off the court and their defense has been vastly better — it’s an extreme number but they’re over 12 points per 100 possessions better when he’s off the court.

The Raptors, meanwhile, rely on Lowry’s offense and his occasionally great defense. He should receive strong consideration for second-team all-NBA, and I’m pointing out DeRozan’s short-comings as more evidence of Lowry’s prowess. Hopefully he can stay healthy down the stretch, which has been an issue in the past; but he came into the season in remarkably better shape. The NBA is certainly unpredictable — a short point guard with a full injury history is having a career season at age 29. But it led to an intriguing Toronto team. One note, however, is that the gap between Cleveland’s wins and Toronto’s is smaller than their gap in SRS or any other decent team rating system — Cleveland’s still a bit better. Nevertheless, the Raptors have the talent to stretch a playoff series out, and while I’ve attacked DeRozan he has actually improved too.

DeRozan to Dantley

DeMar DeRozan has played a good stretch of basketball lately and he’s had an improved season, scoring more efficiently than he has since his rookie year. In fact, he had a bizarre little stat-line the other week: 38 points with 0 three-pointers and 24 for 25 from the line. Going back to the 1964 season, it was only the tenth ever game with at least 24 free throws made. It’s an Adrian Dantley-esque box score line, racking up the free throws without outside shooting. One would think, however, that Wilt Chamberlain would grace the list, but modern players and guys from the 80’s take up the bulk of the leaderboard. Pete Maravich and Tiny Archibald actually have the most free throws for a single game before the merger. And in the playoffs, the record belongs to Dirk Nowitzki, oddly enough, with 24 made. DeMar DeRozan is cementing his legacy as one of the best free throw creators ever, and there’s nothing anti-analytic about it. Free throws, especially from a guy who can shoot, are some of the very best shots in the game. Hitting 24 free throws is no sin.

Lillard: 50 Again

In the same game as DeMar DeRozan’s 24 free throws, Damian Lillard had 50 points (again.) His all-star snub rampage has continued, and his scoring outbursts have echoes of Stephen Curry. He hit multiple extra-long three-pointers — and those were usually off the dribble. In fact, he hit a (first quarter) buzzer beater eerily reminiscent of Curry’s now famous game-winner. There’s another parallel. When Curry was snubbed and — this feels so long ago — David Lee went to the All-Star Game instead of him, he had a sensational last third of the season, scoring 26 points per game with 7.4 assists on excellent efficiency. He ended the year with a good argument for an all-NBA team[2. Saying David Lee was both the All-Star and All-NBA representative for the Warriors feels broadly ludicrous now.], and so will Lillard this season[3. Goran Dragic actually accomplished this in 2014, missing the All-Star Game in a loaded conference but playing so well down the stretch he was granted a spot on the all-NBA squad.]. Lillard is actually averaging 26 points per game so far with nearly 7 assists and he’s nearly the same age Curry was that season — I’m not saying the similarities mean Lillard is going to blow-up like Curry and become an MVP, but it’s time we appreciate his talents.

Injuries, Injuries, Injuries

The number of significant players who are out for the entire season is frustrating and a little scary. There’s Marc Gasol, Joakim Noah, Eric Bledsoe, Tiago Splitter, Tyreke Evans, Eric Gordon, Wilson Chandler, Michael Carter-Williams, and Michael Kidd-Gilchrist (et al.) We haven’t had a complete disaster of an injury yet, like Kevin Durant last year, which led to Oklahoma City missing the playoffs. But it’s a real problem and I’m sure players are concerned too.

There is a decent solution to this conundrum, however, that doesn’t involve fewer games, and I’m going to keep pushing this. The Spurs have long been resting their players to absurd degrees, and they use their bench, even their deep bench, more than virtually anyone in NBA history. This limits the exposure to their stars and other important players — that alone reduces the odds of a devastating injury. But injuries are often have a non-linear relationship with stress and playing time, so the advantages might be a lot larger than they appear from the reduced minutes. The Spurs approach can also lead to a better bench simply because they get more playing time and they can mesh with the other rotation players in real-game situations. Plus, more NBA players would have relevant jobs — that should be important to people as well[3. I understand people would have objections that there aren’t enough talented players in the world to supply the NBA with bigger rotations, but the NCAA is at times more popular and they are vastly inferior. If the NBA used three-man lineups everyone would freak out over an expansion to five-man lineups for the same reasons: not enough good players. This wouldn’t be an issue.]. There are other ways to reduce playing loads, and as long as we don’t obsess over the sanctity of per game stats we shouldn’t fear this kind of change.

The Plasticity of Assists

I reserve this last section of the Week in Review series for random, assorted little studies that don’t really need their own post or may be a work in progress. Sometimes I get too ambitious, but there’s a utility in doing simpler studies because they may not get done unless they’re needed for something else and important findings come from everywhere. Thus, let’s dive into what happens to assists when guys switch teams during the season.

The simplest procedure is grabbing every player who played with two teams with at least 750 minutes for each one in a single season and looking at assists per possession. I’m only looking at guys shipped out midseason because it eschews a couple of confounding variables like aging or player improvement — the latter is still possible but the effect is diminished. Assists per possession should control for a minutes per game change, and, importantly, I’m looking at how it changes with respect to the team’s offensive efficiency. The results are in the graph below. For those without a discerning eye, there is a slight slope to the data and a quick regression will show a small negative correlation between the team offensive ratings and assists per possession. Roughly, for every ten points of change in offensive rating, there’s a change in assists per 100 of roughly 12 percent.

My hypothesis here is that when players go from poor teams to great ones, their assists decrease because there are generally more playmakers on the great team, and vice versa. But the effect isn’t substantial. A 12 percent change means a player with 5 assists per 100 possessions loses about 0.5 assists, and that’s with an extreme case where the two teams have a 10-point difference in offensive rating. However, what isn’t being accounted for? When you go from a poor offense to a great one, you also have better teammates and ones who can hit shots from more of your passes. Thus, you have two competing forces that nearly cancel each other out.

A better stat for this application is probably assist percentage, which is just the percentage of teammate field goals a player assists. This is essentially an adjustment for having bad teammates on offense. Redoing the graph, the relationship is clearer now. The effect is larger — a 25% change in AST% with a 10 point offensive rating change — but it appears to be significant[4. The player whose AST% changed the most, by the way, was Reggie Jackson, who saw his AST% go from 24.5% on the Thunder to 51.2% with the Pistons.].

Of course, these are only general trends, so do the results change under different groups? After experimenting with the data, I found that assist percentage only faintly changed for point guards and centers, and I’m not sure why. I’m surprised point guards experienced virtually no change when switching teams. In a new environment, a player may have a new role and for a point guard that could mean becoming more of a secondary ball-handler. Think about, for example, Isaiah Thomas in Phoenix. In both Sacramento and Boston, his assist percentage was near 35, but it dipped to nearly 23 with the Suns. Since they had other ball-handlers in Eric Bledsoe and Goran Dragic, Thomas had the ball less often and, thus, had fewer assists — it’s tough to maintain your assist rate when you don’t have the ball as much.

However, the results don’t align with the train of thought. I’d be concerned about the sample size here, of course, but perhaps this is something that needs further investigation. I’d understand if centers don’t see much of a change because they’re rarely leading offenses, but point guards should have a more elastic response because better offenses generally have better ball-handlers. But maybe off-ball players, like swingmen and power forwards, respond more to the environment while point guards keep their roles — I’m not certain. Yet it’s an area that might deserve more investigation because it says a lot about how players perform on different teams.

Assists have a long history in the NBA and they’re attached to many ideas and models of performance. If we understand how assists change and react to different environments, we can better understand the players and the league itself. And for that we need a bit of help from data analysis, not just human intuition.