NBA Week 6 in Review: December Frost

The Golden State Warriors decided they don’t want to lose a game ever again, and their conference is disintegrating with former powers slumping and the east rising. Curry breaking the three-point record last season is like Babe Ruth breaking the home run record with a mere 29 and then hitting 54 the next year; Curry’s on pace to shatter his own numbers and clear a bar that doesn’t seem possible — that’s how you never lose a game. Dominance can get boring because the results aren’t surprising, but luckily the league is healthy and interesting just about everywhere, with the exception of a small area in the northeast. With some hindsight and analysis, let’s look back at the past week and learn from what’s happened.

Stephen Curry: Off-the-Charts Scoring

One of my favorite articles and ideas is the efficiency frontier. Simply put, for high shot volume players, there seems to be a natural limit on how efficient they can be — they hit a boundary known as the efficiency frontier — that depends on their usage. In fact, this boundary is quite nearly a straight line, and the players who straddle it are pushing against the edge of performance with some of the greatest scoring seasons ever[1. The efficiency frontier is defined as 90 – 0.89*USG%.]. But right now, Stephen Curry is so far past the line he can’t really be categorized. He’s beyond the horizon; it’s a Ruthian level of dominance we rarely see in modern, well-established sports.

If you zoom out and look at every qualifying season, the effect is greater. Players rarely score as efficiently as Curry does. When they do it’s in a smaller role and they certainly aren’t a team’s lead distributor either. Fittingly, this frontier concept was explored by a Golden State Warrior fan, building a term that would soon be destroyed by his own kin. Some regression should be expected with Curry’s shooting, but even with a modest decline this could be the greatest scoring season ever[2. Let’s not delve into Wilt’s 50 PPG too much, but he got there by playing just about every minute in a high paced era in a vastly different, and inferior, environment.]. Michael Jordan is used as some unreachable measuring stick in many respects, but for this … he’s got nothing on Curry.

The Philadelphia Experiment: What is Failing to Fail?

The 76ers have been terrible at playing basketball, though at least that was the plan (sorry, Kobe) and an inability to win close games has killed their record. There’s been a lot of discussion about the so-called “Process” and that their entire philosophy is wrong. There’s one critical dimension where they’ve arguably failed: drafting. They’ve been able to pick up some valuable draft picks, but they’ve yet to hit a home run and add significant value. There’s Nerlens Noel, who looks like a defensive ace but has issues on offense — then again, the team is so poor on offense it’s hard to tell with big men. And then there’s Dario Saric, who was flipped for Elfrid Payton and could be a useful player but he’s still overseas. Beyond that, they’ve drafted the wrong building blocks for a progressive team, oddly enough.

Michael Carter-Williams was infamously traded shortly after winning Rookie of the Year, but he’s a point guard who can’t shoot well. Jahlil Okafor was their last prestige draft pick, but he’s a non-shooting big man with some of the worst defense in the league, and no one’s surprised; he was projected like that. Joel Embiid, of course, has been setback by numerous injuries and it’s unclear what his future holds. The franchise is forward-thinking in its general approach, but they haven’t targeted the kind of savvy pick-ups that other smart teams have. Theoretically, they’re targeting assets and athletes who can later become shooters, but right now it’s a mess on offense. They’ll have another shot at lottery gold next spring, and the Lakers pick is quite valuable, but they have to actually nail the draft picks for this strategy to work. Otherwise they’re just cycling through trash.

The Right Way Down the Wrong Path

There are some topics that are best avoided partly because it’s a well-trodden area that’s been mined of anything useful and partly because it’s so negative and, well, dirty. When a couple opposing parties engage in mud-slinging, it’s likely that the movement as a whole loses at least a little bit. That’s why I feel trepidation in writing about Wins Produced, which I’ve done several times before in the past long before my tenure at Nylon Calculus. But when I see something ultimately harmful for the discussion of basketball at an intelligent level, I have to speak up.

There’s a lot of background to cover about David Berri, his metric Wins Produced, and his followers, and you can sift through the internet and read about it in various places. There’s a negative article out there about Kobe Bryant, but what’s most irritating and damaging is this idea that because Berri is a professor and the metric is defended by math it’s unassailable. Statistics is not elementary math where you just make sure everything is summed and multiplied correctly. These are open systems where models are built without every possible variable of influence. Basketball is not a perfectly solvable game because it involves five human beings interacting on the court with five other humans.

Dr. Berri is an economics professor and he became prominent in the mid-00’s. As criticisms of his Wins Produced work were raised he often answered opposing arguments with his academic credentials, stating that, essentially, the critic has not published enough in the field to be taken seriously. Everyone is subject to mistakes, even experts, and economics as a field is littered with erroneous assumptions and broken models[3. One of my favorite examples is the Hubbert peak where In the 1950’s, a now famous prediction was made about peak oil in the United States. The “Hubbert peak,” named after a geologist, stated the country would hit peak production in 1970, and in fact it actually happened — it was accurate. But the prediction for the world’s peak production was off the mark, and in recent years we’ve seen a second peak hit the U.S. The model did not accurately predict the effects of technology, like our recent ability to process tar sands and shale deposits economically. Statistical models are mere estimates of reality, and they can break down in unforeseen ways. Be wary.]. Every economist should understand that point, and an appeal to authority is not a particularly rigorous way to judge methods. The Wins Produced metric has not held up to scrutiny, and Kobe Bryant deserves a better career retrospective.

Spinal Tap’s Small Forward: Luc Mbah a Moute’s Turn

In the unheralded vacancy left by Matt Barnes, the Clippers have had little stability at small forward. Doc Rivers is so desperate (I assume) that he’s been trying Luc Mbah a Moute as the starting small forward. When the season started, he was listed far down in the depth chart, and it wasn’t even at small forward. No offense to Luc — he seems like a good guy and he’s a superb defender. But he can’t shoot, and last season he started on the hapless 76ers. Right now he has a decent three-point percentage. That’s deceiving: he’s a mere 3-8 on the season with a career 30 percent average. This year, he’s a step below the 50 true shooting percentage Mendoza line, and he has a minuscule 9.4 usage rate. The Clippers starters should be blowing the doors off teams like they did with Barnes; with Mbah a Moute they’re outscoring teams by a modest five points per game and overall the team is roughly break-even when he’s on the court. He has the defensive chops Los Angeles needs, but opposing teams don’t really have to guard him — the search continues, Doc.

Rondae Hollis-Jefferson Injured

As I wrote in the Nets preview, the franchise’s future is grim and their current outlook is just … bleak. The team has so little of interest and so few assets that Rondae Hollis-Jefferson was nearly christened a miracle by the Catholic church. But, of course, he got hurt and he’ll be out for roughly ten weeks — this is the Brooklyn Nets after all. The most common refrain after his injury was how the team performed at its best with him on the court, usually cited with raw plus/minus numbers. The Nets offense has been about the same with him versus without, but it’s their defense where he’s really been an apparent difference-maker. Notably, he has one of the highest assist rates in the league, but he has contributions elsewhere too.

Looking through video, Hollis-Jefferson is definitely solid but he’s not spectacular. But for a rookie, that’s really encouraging. For example, he guarded Carmelo Anthony a few times and held his own, although he did get beat on an alley-oop and was lucky Anthony missed the attempt. But his quick hands are his best asset and this playlist of every steal he had versus Boston recently is a great showcase of that as he’s able to play the passing lanes well and control loose balls. His offense needs work, particularly his outside jump shot and his ball-handling, but for Brooklyn anything is a plus because they have nothing. Boston may have lost a recent game against the Nets, but they’re winning the war: their Brooklyn draft picks could be incredible.

South Florida Shot-blocking

With Stephen Curry hogging all the glory, I don’t think that many people have noticed Hassan Whiteside’s historic shot-blocking. Per Basketball-Reference, his block percentage is higher than any qualified season in NBA history. But there’s a caveat, and not everyone knows this about the stat. Block percentage is measured as an estimate of opponent two-point field goals blocked, not every field goal. Consequently, in an era with a ton of three-point shooting, the denominator is smaller and block rates are magnified. This method may not be wise because it’s not adjusting for a fall in long two-point jumpers and it’s possible, of course, to block three-pointers. One can simply look at the league average and rank by blocks over average per possession. Manute Bol dominates that method, and he has several of the best seasons ever in blocks per possession. Whiteside is basically playing for second-place, like how a notable rebounder is only shooting for “Best Rebounder other than Dennis Rodman.” But Whiteside’s lead, it should be noted, over the rest of the league is enormous and he’s nearly lapping the field, almost doubling-up the player, Rudy Gobert, ranked second.

The Additive Effects of Assists

While it’s generally considered there are issues with stacking high scorers on a team because there’s only one ball, and the topic of the diminishing returns of rebounding has been explored, there are other facets of basketball where human intuition is misleading. Assists, for instance, aren’t exactly additive in a pure sense, even though we analyze basketball and its best players like it is. Ten assists will “make” a great player, but cram that player onto a team full of other passers and those ten assists will go away in most cases. Advanced public metrics have this issue too, as players from Westbrook to John Wall (from last season) will be rated high, but if they ever have to share ball distribution duties, theoretically, their value will nosedive — which is exactly what happened to Ty Lawson, although part of his issues are due to his personal problems.

There are a few ways to test this issue, but here’s a straight-forward one: look at how teams fare relative to their expected assist rate based on the assist rates from last season[4. Rookies and other players with no previous season have assumed assist rates of 4 per 100 possessions, which is roughly the median I saw in my sample.]. I’ll be using assists per possession because the popular assist percentage metric on Basketball-Reference (and similar ones elsewhere) scales differently based on how many field goals that player is taking. For a quick examination, there’s a graph below showing that expected rate versus the actual. The key here is proper graphics: there’s a reason professors critique axes, and there’s a rampant problem of ill-chosen axis sizes throwing off human interpretation for a large number of graphs. Since my units and axes are uniform, there’s an important and significant observation: expected rates have a wider range, but reality bunches the results together. Thus, if you throw together a bunch of high assist players, as a whole those players will have smaller assist rates.

But looking at the teams with the highest projected rates, you see a pattern: there are a lot of teams like the 90’s Jazz in there who, to no one’s surprise, will generate a high amount of assists. Instead the teams that should be studied are ones with new personnel. If you look at a bunch of teams intact year-to-year there’s little reason to see assist rates change dramatically because those numbers have already been produced in a similar environment. Thus, looking at teams where at least 50 percent of the minutes have come from new players, we have the graph below, and the curve is even flatter. This suggests that firstly, assists trend toward some constant rate and individuals have a smaller effect than it appears, and secondly, how a team plays will drive the assist potential of that particular team because year-to-year correlation is much stronger for teams with little turnover.

Most importantly, how does this translate to well team-level success? While assists don’t fully translate, there’s a possibility of ancillary effects on the team because there’s more raw passing talent. For a quick look, I’ve projected every team’s strength based on Basketball-Reference’s BPM, and then stored the residuals, which shows you if teams out- or under-perform expectations. Looking at the data, including filtering out teams with low roster turnover, there’s actually a slight correlation between assists and teams out-performing what BPM projects. The coefficient here is roughly around 0.11, meaning for every ten assists per 100 possessions a team gains roughly over a point for its team rating — this is modest but measurable.

What could the culprit be here? There are a number of factors here, including many general measures of performance because assists correlate with court awareness, but a simple one is shown below. On teams with a lot of new players, high expected rates of assists generally lead to significantly fewer turnovers. This makes sense because although a team’s assist rate doesn’t change much, there’s still a talent infusion, at least with passing. Plus, this was with historic data going back to 1979, and it’s reasonable to think passing is even more important now.

In a season where the Warriors reign over the kingdom, where the Spurs are probably the closest competitor, assists are being put in the spotlight. But they’re trickier to analyze via the interaction between individuals and the team than people realize, and the diminishing returns don’t work quite as expected. Stephen Curry has his magic shooting abilities, but with poor passing that skill couldn’t be applied so effectively.