There’s been a really interesting shift among a subset of basketball analytics writers (myself included) towards much more micro-level analytics than we’ve previously seen.
For a long time, the focus of analytics was on proving which player is better than which other player. ORTG/DRTG, On/Off court +/-, PER, and RAPM were all the focuses of writing and analysis that itself tried to argue that a team is better with X player in a lineup than they are or would be with player Y.
Such analysis certainly has its uses, though there’s something decidedly juvenile about using complex analytical tools to settle an argument that amounts to “nuh uh, my player can beat up your player.” Still, when it comes to answering questions about efficient time management and minutes distributions, all of that information is useful. Similarly, all of that macro-level data has pretty obvious applications for free agent acquisitions and determining a cap-friendly value for the players that a team is interested in.
That data doesn’t, however, tell us any of the nuances of how or why a player might actually be better than another player, why that player may actually fit or not fit in any given lineup, and, most importantly, it doesn’t really teach us anything about how basketball operates.
Really granular, micro-level information, though, (particularly that provided by the SportVU cameras) can fill in all those holes that macro-level analysis does not. As an example, RAPM and on versus off court +/- indicates that Trevor Ariza was way more impactful on the Wizards than he was on any non-Wizards team, which is great. But it does, certainly, beg the question of “why?”
When we can use SportVU to tell us that John Wall was the best player in the league last season at creating corner 3’s, and Ariza got a massive increase in corner 3’s, and that his TS% was way higher on the Wizards, we can get a much better sense of what caused Ariza’s impact. That information is essential, too. Will Ariza’s shooting translate to the Rockets, now, and away from Wall’s incredible playmaking? Maybe not. That’s helpful information.
Because we learn so much more from the really deep stats, there’s been a large shift to discussions on that granular level. I do genuinely think that’s for the best, too. As fans, we’re drawn to arguments of “better or worse,” but that just doesn’t get us anywhere.
All of that said, knowing when to consider players’ skill fits in a particular context versus settling on their macro-level “impact” can be dicey, from a team-building perspective and a fan one.
Micro-analysis tends to be framed as the team-friendly analysis, that provides decision value for teams, but that’s not always the case.
Take, for example, Jose Calderon.
Jose’s a really interesting test case. On the micro, he’s hyper elite at a few things. He’s probably quietly one of the NBA’s top 25 greatest shooters ever (one of only six players to average 50% from the floor, 40% from 3, and 90% from the free throw line for a season), and he’s led the league in assist/turnover ratio time and time again. He never turns over the ball, passes well, and he shoots like a maniac.
On a play-by-play, skill-by-skill level, it would appear that Jose should be insanely valuable to any team that struggles with spacing and that needs someone who can carefully distribute the ball. There are a lot of teams who need that, so he should be hugely valuable to those teams, to the extent that it compensates his struggles.
On the macro, though, it appears as though Calderon hasn’t been valuable for the last 3 seasons. As age has really affected his lateral quickness, Calderon’s negative impact on a defense (well over -3 DRAPM for a while: astoundingly bad, per Jeremias Engelmann’s RAPM data) has just flat-out outweighed his strong positive impact. There’s not really a good reason to put him out on the floor if, no matter what his offensive impact is, he can’t make his overall impact positive.
This is the odd situation that the Mavericks found themselves at the heart of this offseason, where, despite his massively valuable contributions last season, they jettisoned him because his defense was untenable.
What I’m trying to get at here is to ask: is there a point where context just isn’t helpful anymore? Is there a point where it’s necessary to fall back on macro-level data and say, “this just isn’t going to work no matter what system we put you into?”
That’s really a tricky question to answer. It’s a matter of asking, “is there a situation where if I know exactly what his flaws are, I can put him in a perfect situation that mitigates those flaws?” Or, are his problems just so large that his issues are overwhelming independent of context? I mean, theoretically, there has to be a team for which having one of the best shooters of all time is more important than the negative value of adding a bad defender.
Looking at SportsVU stats tells us that Calderon — despite more or less being a secondary option to Monta Ellis last season — was 12th in the league in passes per game, but he also only averaged an astoundingly low 4.7 assists per game despite that passing number. His hockey assists numbers (i.e. passes that led to assists) were totally average. We can learn something from that.
Jose’s speciality, then, is getting the ball moving and starting plays. He’s a smart and careful passer, and most valuable to a team that struggles with ball-stoppage and ball-hogging (congratulations, Knicks fans) less than he is valuable to a team that struggles to get scorers in a scoring position.
Similarly, Jose is fond of pull-up jumpers. Per SportsVU about 30% of Jose’s points per game were on pull-up jumpers, a greater percentage than Kevin Durant or teammate Dirk Nowitzki, both noted for their propensity to shoot off the dribble.
That makes Jose doubly valuable to any team that’s struggling with it’s spacing. Not only is he a 45% 3-point shooter (which, holy crap), but he hits a good number of his shots coming around screens and when given too much space. That means that other teams can’t give him any room at all, even when he’s on-ball.
So, theoretically, for a team that already has an elite defense that needs ball distribution (but not precise “action passing”) and spacing in particular, might Jose become valuable? Or, might our micro-level analysis allow us to come up with a situation in which doing a micro-analysis of Calderon yields more information than simply saying “he’s not valuable?”
Here’s the thing though: based on some prior quick and dirty regression analyses I’ve done on the change in DRAPM from going from a bad defense to a good one, Caldy’s best-case-scenario even on a good team is something roughly a -2.9 (he was at -3.81 last season), and it’s unclear how much better his offensive impact would become from being on a team that can really use his skills.
If his change in offense is proportionate to the change in D, the best case scenario is about 2.8. So, it doesn’t seem that Calderon’s at a point in his career where he’ll ever be “impactful” all around, per APM metrics.
Maybe, in the absolute best-world scenario, he could have a non-negative impact, but that’s really the best-case scenario. And what team really fits that best-case anyway? The Bulls? Joakim Noah already passes the way Jose does. The Pacers? They’re bad, but not exactly desperate for spacing among guards.
Calderon, then, is an odd case where talking about why his value is what it is doesn’t really reveal anything about how he can fit on a team, even though it feels like it should. Micro-level analysis can be interesting for us, because we learn more about how Calderon plays, but it wouldn’t help a team, really. His defense is just too much of a problem for him to be valuable.
There aren’t many players in the league like Calderon. First, there aren’t many players who are elite at anything the way Calderon is elite at shooting and passing and who are also not obviously valuable to a team, and second, there aren’t many players whose offensive and defensive impacts are so extreme.
But Calderon does pose as an interesting thought-experiment of sorts. Despite the shifting focus to measuring players mostly based on the contexts in which those players do well or do poorly in order to maximize their utility, here we have a player whose value on both offense and defense is so extreme where context more or less doesn’t matter.
What I’m ultimately trying to argue is that macro and micro analyses of players and their contributions have a tricky relationship, and learning, how, when, and why we need to manage our usage of those analyses can be important as we continue to develop our online contributions.
