All Posts By

VP Sinha

2016-2017 in Review: A Tale of Two Teams

By: VP Sinha

 

Question: Can You Guess Who These Players Are?

 

Player 1 – 13 points, 7.4 rebounds, 2.2 assists per 36 minutes, on 53% TS% (39% from 3).

Player 2 – 6.4 points, 6.6 rebounds, 1.2 assists per 36 minutes, on 35% TS% (17% from 3).

 

As you may have guessed, this is a trick question. Player 1 and Player 2 are the same player – Luol Deng. That first stat line isn’t from his time in Miami; it’s what Deng put up playing without Jordan Clarkson in 872 minutes this season. The latter encapsulates Deng’s pitiful performance next to Clarkson in 614 minutes this season, and the trend holds steady no matter the month. Deng struggled in November (44.7% TS% without Clarkson), was excellent in December (62.9% TS%), and was decent in 2017 through his deactivation (53.0% TS%). But consistently, Deng played considerably worse next to Clarkson. He posted TS% of 33.3%, 38.2%, 34.5%, respectively, while sharing the court with JC. The stability of these numbers through Deng’s peaks and valleys suggests a fundamental incompatibility of these two players.

Jordan Clarkson Luol Deng

If you look at the breakdown, you’ll note that Deng’s usage drops to a laughably low 11.9% next to Clarkson.

Who the hell is going to stay in a rhythm when you’re hardly touching the ball?

 


 

If we extrapolate from Clarkson’s play style, we could infer Deng doesn’t work well with Clarkson’s heavy ball-pounding ways. Deng prefers to spot up and maybe attack the closeout, but the weak side might as well not exist to Clarkson. This means Deng can’t get into a rhythm, so the shots that he gets end up bricking, which reinforces Clarkson’s do-it-all mentality

Compare these numbers to Deng’s numbers playing with Lou Williams instead of Jordan Clarkson. Williams was also a ball pounder, but he was a far more dangerous scorer who commanded more attention from defenses, and he had better vision in kicking out to shooters. In relatively limited minutes (109) with Williams and without Clarkson, Deng’s usage actually rose to 15.2% from 14.2% on the season, and his TS% was a scorching 72.1%. That number would go down in a larger sample size, of course. But because Williams would at least spoon feed Deng good looks, his ball-pounding didn’t have quite the same deleterious effects on Deng’s shooting percentages.

This contrast highlights an essential truth that people too often gloss over about the NBA: fit matters.

It matters a lot. Deng has been ostensibly terrible nearly all season, and yet it turns out he’s actually been more or less the guy we wanted, even as he’s played at a suboptimal position all season (he’s a 4, not a 3). Deng is not close to a star, but he plays solid defense, he is a passable shooter (especially at the 4 and if we discount his horrible November), and more importantly, he understands how to play offense and defense. He doesn’t muck things up, and that has its own value, but he needs to be spoon fed shots in rhythm. It turns out Clarkson doesn’t do that in his freeform style of play, so Deng is flat out terrible next to Clarkson. This brings me to the main point.

 


 

The Lakers’ 2016-2017 woes boiled down to a fundamental incongruity in roster construction.

The unit of D’Angelo Russell, Nick Young, Luol Deng, Julius Randle, and Timofey Mozgov posted +6.0 net rating in 406 minutes.

Starting Lineup Data

 

You know that Trail Blazers team that looked revitalized after acquiring Jusuf Nurkic (13-3 in March)? Their starting unit with Nurkic posted a +6.2 net rating in 229 minutes, and this is later in the season when opposing teams have begun resting or blatantly tanking. There are caveats with the Lakers’ +6.0 net rating – most notably, when they weren’t playing well, Luke would pull at least one of their starters, and so you can argue the starters have an artificially bloated net rating, but that’s beside the point. A +6.0 net rating is on par with a starting lineup that helped the Blazers go 13-3 in March, and even if it’s bloated by a few points or so, that’s incredible for the starting lineup of a roster that won 26 games.

The original bench unit of Clarkson, Williams, Ingram, Larry Nance, Jr., and Black was even better, putting up a +7.9 net rating in 196 minutes together. However, that lineup played a completely different style compared to the structured offense and defense that the starters played. Offensively, they ran far more early offense sets or just high ball screens to get Williams and Clarkson looks on offense, with Nance and Black running cleanup duty. Defensively, the frontcourt supplied plenty of speed and length to cover up mistakes from the backcourt. In response, the backcourt opportunistically jumped passing lanes or otherwise picked pockets at elite rates: Clarkson’s 3.3% STL% and Williams’ 2.5% would outpace a duo of, say, Chris Paul (3.0%) and Kawhi Leonard (2.7%) on that front.  We get the sense of a far more improvisational unit than the methodical starters, which worked well for that group’s talents.

But therein lies the problem. The minute these two units mixed they were disastrous, with few exceptions. For example, replace Ingram for Deng on the starters and you get a -32 net rating in 63 minutes.

When you are essentially running two different systems on the same team, is it a surprise the mixed units fared poorly? 

The Lakers lost quite a few games in November because these mixed lineups were horrible, especially the Nance/ Randle small ball lineup (-11.2 net rating on the season).

When injuries struck in December and these units were forced to intermingle, the Lakers plummeted in the standings. In Luke Walton’s defense, he recognized this shortcoming. People might remember his desperate attempts in December to preserve that bench 5 together even as the starters faced injuries to Russell and Young. This included starting Calderon and Huertas or starting Ingram and letting him play nearly 40 mpg to play with both units. Some were calling for his head for not starting Clarkson earlier, but the lineup data suggests it was the right move. 

 


 

The Lakers picked Walton to install a modern offense, specifically one in the vein of Golden State, San Antonio, and Utah, as opposed to the fast-paced, gunning style of Mike D’Antoni. They overpaid veterans to help implement the offensive and defensive systems – and the plan worked! Russell has the poise and feel of a veteran, and Young, Deng, and Mozgov are, for all their warts, experienced players. People bag on the Deng and Mozgov signings, and I get that their contracts are terrible and that their individual production was underwhelming. But it is also true that they, especially Deng, were crucial components to a core group that outperformed other starting units at a pretty significant rate, and that this unit only improved as the season went on. The starting 5 posted a +3.2 through November, +6.5 in December, and +9.5 in 2017.

What’s more intriguing is the four-man unit of Russell-Young-Deng-Mozgov was quite good no matter who the fifth guy was, except when it’s Clarkson. Russell’s steady play with this group of veterans particularly merits notice: in 494 minutes, he averaged, per-36 minutes, 21.8 points (on 57% TS%), 5.3 rebounds, and 7.6 assists (to 3.7 turnovers) while solely orchestrating the offense. The unit played even better if we replace Randle with anyone besides Clarkson. Those lineups were +16.1 net rating in 89 minutes, with an elite defensive rating of 95.6. The only time this unit crashed was using Clarkson as the fifth guy, where in 15 minutes they put up a -82.4 net rating. Yes, that’s a small sample size, but it’s not trivial given how hugely negative that net rating is. But again, given Clarkson’s play style, is it a surprise that the structured starters would be so ineffective with him?

As we look back on the data from the 2016-2017 season and look forward to the type of system that Luke looks to install, we can see clearly that trading Williams was undoubtedly the right move. Lou was brilliant for us, but he was almost too good. He ended up breaking Luke’s system and running a lot of high ball screen type actions instead, and he inspired some selfishness in Clarkson’s game as well. The Lakers should look to form a coherent roster, where all the players run mostly the same system and there aren’t huge stylistic shifts as one player is substituted for another. In this regard, Deng and Mozgov are necessarily the liabilities people think they are, at least on the court. Yes, they aren’t world beaters, but they can play supporting roles in absolutely functional lineups. (A digression, but this is why I’m skeptical of Brewer and Nwaba as rotation players on the team next season. They might be thriving in these energetic, defense-leads-to-offense type lineups, but their inability to shoot from the 2 or 3 will sabotage attempts to run Luke’s playbook.)

 

Williams was the biggest culprit in terms of stylistic conflicts, and while he was our best player, he also masked some roster construction issues that need to be fixed. With Lou gone, we can build a proper lineup for Walton.

 

But there is still a glaring question to be answered.

 

What Should The Lakers Do With The 2014 Draft Class? 

 

First, there’s Clarkson. I have to confess that I’ve grown skeptical of where Clarkson fits in on the Lakers moving forward. His absolute best stretch of games was two years ago now, where he thrived running simplified high pick and rolls.

Can that translate into the sets and principles that Walton is trying to establish in LA?

His decision making has improved as he’s gotten away from Lou’s shadow, reflected by his ATR as a starter rising to 1.5, from 1.2 as a reserve, though he was abysmal scoring the ball (16.1 points per 36 on 49.5% TS%). His 3-point stroke is inconsistent and subpar for a guard, and he forces shots near the rim because he misses the reads. Stylistically, he’s a bench 2, which is complicated by the fact that defensively, he’s probably best guarding the 1.

Can he rein in his shot-happy tendencies? Can the Lakers find an appropriate backcourt partner for him on the bench?

A more nuanced question awaits Randle. Randle is an interesting spot where the starters were pretty good with him, and the original bench lineup was pretty good with him. He has shown plenty of improvement this season, and a consistent 3-ball might be a game changer. But both the starters and the bench were better with other players, which explains why he’s at #67 on ESPN’s RPM for power forwards (for comparison’s sake, Nance is at #23). Small ball lineups with Randle at the 5, or with Nance and Randle together, have mostly been terrible. His defensive awareness is perhaps the worst on the roster, and lurking in the background is his looming contract extension. Randle is not a league average starter yet, but he stuffs the box score and someone is going to pay him a lot because of it. Will Randle be able to outperform the contract he earns next summer?

As the Lakers’ Front Office looks to evaluate the state of the roster, I can only hope that they identify these positive takeaways in the midst of all the negatives. Smart management means amplifying the good and mitigating the bad. This season was as much a consequence of intra-roster incoherence as an indictment on the performance of our individual players per se. That isn’t to say that we don’t need more talent – we certainly need more talent – but you can see the foundations of an overachieving team hidden somewhere in the rubble of this past season. Despite question marks about Randle and especially Clarkson, we can see how some of the major core pieces of the Lakers’ roster fit together, even the much-maligned and overpaid 2016 free agent acquisitions. Perhaps Magic and Pelinka should consider staying the course set out this season and making incremental changes with an eye on roster coherence, instead of effecting sweeping changes.

Lies, Damned Lies, and Brandon Ingram’s Advanced Statistics

By: VP Sinha

What advanced stats tell us about Brandon Ingramand what Brandon Ingram tells us about advanced stats.

 

Despite the debacle in Dallas, Lakers’ prized rookie Brandon Ingram has had a very nice January, now showing consistent flashes expected from the #2 overall pick. His stat line this month sits at a healthy 11.1 ppg on 53.2% TS%, including 14-35 (40.0%) from 3, 3.9 rpg, and 2.3 apg (to 1.2 topg). Yet even factoring in his 2017 renaissance, Ingram’s stat line more than halfway through his rookie campaign is a mere 8.1 points, 4.1 rebounds, 2.1 assists on 45.9% TS%. All-in-one “advanced” statistics reveal a ghastlier picture: 7.7 PER, -0.01 WS/48, -4.2 BPM, and -5.44 RPM (-2.59 ORPM and -2.85 DRPM) – and his RPM is dead last in the NBA by a fair amount.

In general, it’s not wise to project a rookie’s success based on general numbers that don’t capture the nuances of a player’s situation or account for development. Players improve and work on their flaws, which limits the predictive value of a rookie season. Within the last three years, we’ve seen the architect of RPM compare Julius Randle’s skill set to Anthony Bennett’s as well as 538’s Neil Paine project Andrew Wiggins’ career impact to be similar to James Posey’s.  Sometimes, common sense should simply supersede what the numbers say.

But surely these numbers mean something, right? Jahlil Okafor averaged a fairly efficient 17.5 ppg as a rookie last year, passing many “eye tests,” but many advanced stats painted him as likely the worst player in the NBA in terms of on-court impact. One year later, it would seem that the advanced stats mostly had it right. So what gives?

Well, the short answer is that the NBA’s advanced stats are useful but flawed tools that always require context.

In the case of Ingram, we can find that his poor numbers reflect both his poor performance as well as limitations of advanced stats. Let’s take a closer look on some of the key numbers.


 

Scoring Struggles

When we look for explanations for why advanced box score stats (BPM, WS/48, PER) mark Ingram as one of the worst players in the league, we should begin with Occam’s razor: are there particular statistics (such as RBD% or TO%) in which Ingram places at the bottom of the league? If so, those are probably dragging his overall box score stats down. For Ingram, it’s his efficiency at an abysmal 45.9% TS%. The rest of his box score stats are unimpressive, but they are not particularly bad. So, odds are his abysmal advanced box score stats are mostly explained by TS%, and I don’t find it particularly illustrative to look at these all-in-one stats.

As we take a closer look at Ingram’s TS%, there’s no doubt Ingram has been very poor scoring this season as a whole, and what makes his efficiency more troubling is that his usage sits at a modest 15.7% – this is not necessarily the story of somebody being asked to do too much to the detriment of his efficiency. So if the goal is simply to evaluate his performance thus far, we would state that Ingram, on account of having difficulty throwing a pea into an ocean, has been a mostly terrible player and leave it at that. All-in-one box score stats just reflect this reality.

But for Lakers fans, what matters less is how he’s performed this season and how he projects to play down the line. Can we still expect him to become a star, or even very good, offensive player? Or was Tayshaun Prince a more accurate comparison than we’d like to admit? To better answer that question, just looking at TS% is not enough. We need to know how he is struggling and whether or not we expect that to carry forward. Is he having issues scoring at the basket, is his jump shot just absent, or both? Can we expect his shortcomings to improve?

As it turns out, the scrawny teenager’s issues do not include getting to and finishing at the basket. Ingram moves like a gazelle, and he glides through the smallest seams of the defense with ease to finish at the rim. Ingram finishes at a decent 55.2% of his layups, with just 31% of them assisted. The assisted rate is crucial: though Giannis Antetokounmpo made 62.1% of his layups as a rookie, 41% of those were assisted. Ingram finishes at a pretty decent rate despite creating more layups for himself than most other players. For comparison’s sake, rookie Kawhi Leonard shot just 49.6% on layups with 37%  of those makes assisted, and this season’s LeBron James makes 70% of his layups with about 35% assisted. Ingram also has healthy free throw rate of 0.347, even if his free throw percentage leaves much to be desired (67.2%).

Though Ingram lacks the burst and explosiveness of a Giannis or LeBron to attack the rim for dunks at the same rate, he’s already a crafty driver with some serious shake off the dribble. Despite this finesse, he seeks contact in a way that belies his frame. As he gains strength and improves his athleticism, we can expect to see more plays like this assault on the hoop. Based on his year so far, Ingram projects to be a dangerous slasher and finisher at the basket in his own right.

No, it’s not finishing at the rim that is driving Ingram’s TS% down. Rather, it’s his surprising inability to shoot. The thought was that out of college, Ingram will at the very least be a legitimate 3&D guy. Yet his 3-ball, and jump shot in general, is MIA. Per NBA.com, he is shooting merely 29.7% FG% on both 2 point and 3 point jumpers, and most of his makes are assisted. His catch & shoot percentage rests at a poor 45.1% (all shooting percentages will be eFG% unless noted otherwise), and he’s worse pulling up at just 33.1%. Perhaps the most damning stat of them all — he shoots 42.3% on “open” and “wide open” shots. Prepare your eyes.

But going forward, this actually bodes well for Ingram. His shooting may never live up to his pre-draft comparison of Kevin Durant, but he has a smooth jump shot. Yellow flags about free throw percentage aside, Ingram was a great shooter in college (41% on 3s), and Coach Luke Walton consistently raves about Ingram’s lights out shooting in practice. We should be careful to draw conclusions from a small sample size, but January has Ingram shooting about as well as we’d all hoped. His catch and shoot percentage has skyrocketed to 58.1%, the sort of number you expect from a very good 3&D guy. That’ll drop some, and Ingram still should make tweaks to his jump shot to get his release point higher, but it gives us glimpses of the shooter that Ingram will grow into.

There was another scrawny, sweet-shooting forward who also struggled with his jump shot as a rookie, shooting 36.3% eFG% overall on jump shots, including 20.6% from 3. That player is now considered one of the greatest shooters in NBA history. I don’t mean to fall into the same trap as with the Durant comparisons by bringing up Dirk Nowitzki. Rather, it’s reason to remain optimistic about Ingram’s shooting ability as his jumper is tweaked and refined by NBA shooting coaches. The Mavs legend also serves as a template for the Laker rookie to learn to shoot more effectively over smaller defenders.

Meanwhile, Ingram consistently shows precocious ball handling and the ability finishing at the rim, as well as playmaking ability: it wasn’t until Dirk’s 5th season that he surpassed Ingram’s assist percentage (10.4%) by a significant amount. The degree to which Ingram’s jump shot translates will determine how just how effective he’ll be offensively, but it is comforting to know the rest of his game won’t be a huge limiting factor. The sky is still the limit for the 19 year old.


Understanding Ingram’s League Worst RPM Numbers

ESPN’s RPM is quite different than other frequently referenced stats, and it is worth investigating it specifically. The details of its construction are worthy of its own post and I won’t get into them now. But what makes RPM interesting is, after using a BPM-type stat as an initial guess of how good a player is, it doesn’t use box score numbers to make its evaluations. Rather, it considers how well the team performed with a player playing versus without him playing, adjusted for teammate quality and opposition quality. Of course, a player’s raw production plays a part in that, but if someone provides great screens, has gravity rolling to the basket, or can stretch the floor, then RPM may rate them favorably despite poor individual numbers. Last season, while some stats such as PER saw Okafor as an above average contributor, RPM rated him as the worst player in the league. For the most part, I agreed with this evaluation. Ingram has now replaced Okafor in RPM’s evaluation, and so it is perfectly valid to wonder whether there is cause for concern. To answer that question, we need to look more carefully at lineup data.

Consider the Lakers’ bench. For the first two dozen games or so, the Lakers’ strongest lineup was the full, healthy bench corps, which posted an ORTG of 108.9 and DRTG of 100.7, per NBA.com. In that unit, Ingram played alongside two attack dogs in Jordan Clarkson and Lou Williams while Larry Nance, Jr. and Tarik Black were happy to do all the dirty work. Ingram would sit behind the 3 point line, often not touching the ball for entire possessions. Some may argue he played a glue guy in this lineup, but that’s not quite true, and it is reflected in lineup data. Per NBAwowy.com, the 3-man unit including Clarkson/ Williams/ Nance has a net rating of +9.6 versus +5.4 when playing with Luol Deng instead of Ingram. If you replace Nance with Black, the 4-man lineup has a net rating of +10.4 with Deng (in a much smaller sample size) versus +5.0 with Ingram. The fact that these lineups play better with Deng undercuts the argument for Ingram being a necessary glue guy in bench lineups. Even though Ingram’s raw +/- looked good playing in this bench corps, RPM correctly ascertained it’s mostly due to his teammates rather than his own contributions. This is a good example of the considerations RPM uses to arrive at its numbers.

This reasoning explains his horrendous DRPM numbers (-2.85, ranked #432 out of 438), which may surprise people who find him a fairly advanced defender by age. Most lineups with Ingram instead of Deng are worse defensively. The rookie has the length to bother shooters and he has good defensive instincts, but he gets bullied in the post (0.90 ppp allowed, 47th percentile) and in isolation (1.08 ppp allowed, 18th percentile). Some of this can be attributed to poor help from teammates, but regardless, the experienced Deng is for now the better defender. Furthermore, the Lakers’ small ball lineups struggle defensively, so lineups where Ingram plays alongside Deng do worse defensively than those in which Deng plays next to a traditional 4. Finally, the Lakers are a terrible defensive team in general. That the team does even worse with him in most lineups speaks harshly to his defense. DRPM consolidates all of this information and reasonably concludes that Ingram is a terrible defender, even if we can see that much of that reasoning is circumstantial.

In contrast, it’s more plausible that RPM is right about Ingram being terrible offensively (-2.59 ORPM, ranked #405 out of 438). He isn’t a great offensive rebounder, he doesn’t create that much offense for teammates, and his individual scoring numbers have been terrible. Up until recently, teams were eager to sag off of him, so he wasn’t providing spacing either. Without doubt, there is truth to this fiercely negative evaluation. But another factor to be considered is Ingram’s role. Walton has accelerated the rookie’s development by giving him ballhandling duties and keeping him involved in the offense. We’ve seen Ingram improve as a playmaker, and he’s at a level where he can make sophisticated reads like this one, where Mozgov blew Ingram’s chance at recording 10 assists against the defending world champions.

But this playmaking came only after growing pains. With D’Angelo Russell injured for much of December, Walton leaned on Brandon Ingram to carry more ballhandling duties. He responded with his on court plus/minus dropping to -14 in December, despite most of his counting stats remaining steady. Take the starters, for example. With Russell, the starters post a net rating of +5.9, with an ORTG of 110.9 and DRTG of 105.0. With Ingram, that number falls to -39.9, with an ORTG of 71.4 and DRTG of 113.3. To be sure, that is partly due to a small sample size, and it probably says more about Russell than anything else. However, per 82games.com, the top three lineups in terms of minutes with Ingram as a 1 or a 2 have a net rating of -23, -43, and -34. Notably, it’s the offense that plummeted in all three cases.

The point here is that Ingram has played terribly, and the stats reflect that. Part of why he was playing terribly is because he’s a poor offensive player as a rookie. But the other fact is he was out of position and forced to do things he isn’t good at yet. These decisions will ultimately benefit the #2 overall pick down the road; we are already seeing the fruits of it this January. Ingram is boosting his assist count while lowering his turnovers, all while creating more offense for himself off the dribble. But this sort of context is not covered by RPM, which is why I don’t worry too much about its gloomy prognosis yet.


Context is King

Charles Barkley might read this article and conclude, “Aha! I told you all along. Don’t bother with analytics,” but that is not the takeaway here. Some of the best statistical rookie seasons in the last 15 years were by such luminaries as LeBron James and Chris Paul. Blanket statements are never useful. Tyreke Evans and DeJuan Blair also had strong rookie seasons. Jahlil Okafor and Anthony Bennett had poor rookie seasons, but so too did Tristan Thompson and DeMar DeRozan. It’s far more important to understand what the numbers are telling us and how much we can expect those numbers to project forward. While Brandon Ingram has certainly did not have the first two months to his career that I’d hoped for, he hasn’t left me any less optimistic about his future. His January turnaround portends what looks to be a successful career.

VP Sinha, Writer (@shaqpropogation)– VP is a crazy talented stats guy who either understands the math behind BPM, RPM, etc, or at least has us convinced that he does. We don’t know for sure, our eyes kind of glazed over halfway through. His job at Laker Film Room is to help intertwine data with the film work in a cohesive manner.