The Fundamentals: With advance stats stuck in neutral, focus shifts to process
By Rob Mahoney
Around this time every year, a Boston convention center becomes a hub for sport-centered innovation and thinking, as team personnel, quantitatively inclined analysts and bright-eyed undergrads come together to create the unique dynamic of the Sloan Sports Analytics Conference. It's an event that technically encompasses all manner of sports and leagues -- from rugby to hockey to pro football -- but makes no qualms about being an NBA-centric event. Rockets general manager Daryl Morey has co-chaired the conference since its inception in 2006, and this year 29 of the league's 30 teams sent representatives to explore all that the conference has to offer.
The weekend features a series of panels with the likes of Morey, Mavericks owner Mark Cuban, NBA commissioner-in-waiting Adam Silver and many others, and a research paper track that showcases sports analytics in more specific detail. Both yield quality information based on all kinds of perspectives, but the statistical momentum had noticeably slipped from this year's proceedings, due largely to the massive -- and as yet incomplete -- undertaking that is SportVU.
SportVU is STATS' system of optical-tracking cameras installed in half of the NBA's arenas. It's designed to chart the placement and movement of all 10 players on the court, the ball itself and even the three referees. Those cameras capture 25 frames per second, resulting in a massive, unprocessed data load that forms a sea of X-Y-Z coordinates. Any meaningful stretch of games would require millions and millions of coordinates, holding in aggregate the potential for all kinds of statistical analysis. At this year's conference, we saw the SportVU data put to work to quantify interior defense, measure the acceleration of players, evaluate which defensive players can effectively deter shots from ever happening, analyze specific trends in offensive rebounding and chart the specific execution of a set play. The possibilities are quite literally endless, but the data itself is so unruly -- and yet so incomplete -- that it takes time for even dedicated team personnel to really break through. SportVU is very much the future of basketball analytics, but even the most brilliant statistical minds in the game are so deep in the future that they're drowning in it.
It's only a matter of time before the subscribing NBA teams figure out the best ways to wade through those massive data dumps, and word is that many have already made incredible progress. Half the teams in the NBA may already be onboard, but the gaping holes in the data set leave the league's optical tracking hopes half-empty. It's great that the teams on the cutting edge can all keep a close eye on one another, but all of those franchises are in part biding their time until data across the entire league becomes available. SportVU has been all the buzz over the last few years for the possibilities it offers in terms of data collection, but the harvesting of that technology has turned out to be a multi-year undertaking rather than some overnight epiphany.
I can understand why that would lead some to be impatient -- we all want our flying cars, and the few glimpses we get of SportVU's applications are intoxicating. Yet the world of analytics has always operated at something of a slow churn, and is predicated on the value of process above all else. As valuable as SportVU may come to be, its specific analytic products are dwarfed by the value of the methods that brought about its creation in the first place.
The Sloan Conference, amid technical presentations and the revisiting of familiar stat-related quandaries (the importance of communication/packaging of data-driven findings, etc.), is powered by an invaluable intellectual curiosity. Some show up for the networking scene or due to the fact that it's become an industry event. Yet at its core, Sloan is still a gathering for the inquisitive, and that's a nice framing for the world of analytics as a whole. R.C. Buford, unimpeachable general manager of the Spurs, and panelist at the conference, put it well in addressing a question about a hypothetical coach who dismissed analytics out of hand:
"I don't think 'analytics' is limited to number-crunching. An approach of wanting to be thoughtful, of wanting to analyze all the situations … and then determining if the numbers fit what the system is -- I think if somebody doesn't have that type of approach, it would cause me pause to want to bring that person into our organization because I think the decisions that happen throughout the course of a season can't be done without an approach of heavy analysis. But I don't think it's necessary to limit it to number-crunching. So for that reason, if somebody just says 'I'm going to do this and do it just because,' and it's not going to be interdependent on who we're playing, who's with us and the circumstances, then you'd have to step back and say 'Is that the person that I want leading the team?'"
You could use the same logic in regard to who you'd want building a team, owning a team, or even writing about a team. What's most important of all isn't the breakthrough tech or some revolutionary study, but the fact that heady stat-types and sharp basketball minds can come together to challenge themselves and their way of thinking. So many people in the basketball sphere are constantly looking for a better way to do things, be it to run a team or merely to understand their favorite one. In such a framework, we all win. Teams make better decisions. Players take to more informed strategies. Quants allow coaching feedback to shape their analysis. And the fan experience is augmented by looking at the game in entirely new ways, be they stats-driven or otherwise.
That's because the power (of the Sloan Conference and of basketball analytics in general) is still in the process. SportVU holds the potential to forever change the world of player evaluation, but ultimately analytics are most beneficial to us as a vehicle for our open-mindedness and an extension of our scholarship. It's by those means that we can always find value -- even as we wait on SportVU to bear fruit.
For this week's GO FIGURE, we'll focus on some of the findings of the research paper finalists from the 2013 Sloan Sports Analytics Conference.
• The highlight of the research paper competition came from Sloan veteran and Grantland contributor Kirk Goldsberry, who harnessed Sport VU's optical tracking data to measure some elements of interior defense. In particular, Goldsberry focuses on two different measures: the efficiency of opponents who shoot near the basket with a certain defender nearby and the number of shots a team allows when that same defender is within proximal range of the hoop.
By now, the most extreme case studies of Golsberry's presentation have become rather well-known. Larry Sanders, the pogo-stick center for the Bucks, allows opponents to complete just 34.9 percent of their shots close to the basket. The presence of Golden State's David Lee, on the other hand, correlates to opponents finishing 53 percent of those attempts. Neither of these marks is entirely representative of the player's defensive abilities, but they give us an effective starting point from which to evaluate how effectively those two defenders protect the rim within their teams' respective schemes.
Here's the top 10 in terms of lowest FG percentage allowed when the defender in question is close the basket: Roy Hibbert, Sanders, Elton Brand, Serge Ibaka, LaMarcus Aldridge, Jermaine O'Neal, Kosta Koufos, Kendrick Perkins, Joakim Noah, Dwight Howard.
You can read Goldsberry's paper here.
• Matthew Goldman and Justin Rao's paper entitled "Live by the Three, Die by the Three? The Price of Risk in the NBA" is full of interesting tidbits on a number of different subjects. One such morsel: As a result of their study, Goldman and Rao find that the NBA has its own "rubber-band effect," by which the offensive efficiency (or the points scored by a team per 100 possessions) increases by 10 percent when that team is trailing by five points or more. One possible explanation: trailing teams make a greater effort to get the ball to their best players, which generally results in more effective offense. One can certainly see this effect in the shooting frequency of Kobe Bryant, but I'd suspect the same holds true for many other superstars who benefit from a more altruistic iteration of "hero ball."
You can read Goldman and Rao's paper here.
• The ideas presented in many of these papers were hardly finished thoughts, as the researchers looked to tap into specific areas of study without necessarily running a complete analysis of the target variables. Still, there's valuable information to be gleaned or confirmed as a result of this process, especially due to the fact that several of the researchers had access to data sets that aren't publicly available.
One product of that access was the following chart constructed by Philip Maymin, who studied bursts of on-court player movement in his paper, "Acceleration in the NBA: Towards an Algorithmic Taxonomy of Basketball Plays":
The image on the left represents the logged locations of players over the course of the project's sample, which included some 30,000 NBA plays over the 2011-12 season's 233 regular and post-season games. The image on the right represents the spaces on the court in which players are accelerating most, with the highest frequency of bursts coming closer to the basket.
You can read Maymin's paper here.
NOTES FROM AROUND THE ASSOCIATION
1. Derek Fisher already stealing minutes
Thunder head coach Scott Brooks has an established weakness for Derek Fisher; he's so wooed by the point guard's veteran savvy that he's willing to overlook most every limitation of Fisher's game. That fact alone made Fisher's recent return to Oklahoma City a bit disheartening. OKC could certainly live with playing incumbent backup point guard Reggie Jackson a bit less, even if Fisher is far from a preferable option at that spot. Yet when Fisher's presence on the roster is costing Kevin Martin minutes …well, that's a bridge too far. It's not at all uncommon for a coach to favor a certain player despite most tangible evidence, but there comes a point where the difference between two options is so vast as to be impossible to explain away. The context in which it makes sense to play Fisher off the ball as opposed to OKC's efficient (and highly complementary) super-sub is slim, if it exists at all, and one can only hope that Brooks' interest in giving Fisher such frequent minutes fades as the weeks wear on.
2. Walking back the offensive basket interference rule
I understand the basic intent behind the NBA's basket interference rules, but I'm not quite sure why the "cylinder" concept applies to offensive players looking to finish lobs. As it stands, the rim creates an imaginary threshold that extends upward, and no player on either team may touch the ball when any part of it crosses that boundary. That's sensible enough when both teams are competing for a rebound on a live ball, but why use the same specification for players completing alley oops that happen to be caught over the rim?
Sports are arbitrary by nature, and we needn't get tripped up by challenging rules purely on the grounds of being unnecessary. Yet I can't help but notice -- and mildly object -- to this little facet of the offensive interference rule, as it seems to deviate from the game's own internal logic.
3. Is Evan Turner more trouble than he's worth?
Turner still has every component of the skill set that warranted his selection as the No. 2 overall pick in the 2010 draft, but the sum of those skills makes him difficult to use in optimal fashion. Turner is a good creator off the dribble, albeit one that generally creates inefficient shots and converts them at underwhelming rates. He's a fine defender, though ultimately more of a cog in a good system (like Philly's 11th-ranked D) than a true difference-maker on that end. He's a superb positional rebounder, but won't drastically change a team's fortunes on that end. He's shooting a decent percentage (36.9) from long range for the first time in his career, but without the kind of volume that would make that percentage register in a meaningful way. He's a terrific passer and a sound ball-handler, and yet asking Turner to manage too much of a team's offense tends to be problematic.
Pull all of that together and you have a player capable of the 22-point, 10-rebound, nine-assist outing that Turner had against Golden State on Saturday, but he's also a fill-the-gaps talent that doesn't make sense as a ball-dominant scorer, first-option playmaker, spot-up shooter, or off-ball cutter. In a best-case scenario, Turner would be able to access all the best parts of his game without stretching too far in any one direction, though if doing so requires a drastic restructuring, you have to wonder: Is Turner really worth that much effort?
4. Metta World Peace, the upper-body rebounder
Something quirky to keep an eye out for the next time you watch the Lakers: Metta World Peace may have the worst technical box-out form in the NBA, with a habit of getting position for rebounds using his arms almost exclusively. It's technically about as far as one could get from proper, but MWP makes it work by getting incredibly physical, pushing and pulling on opponents and forcing referees to pay close attention to a play off the ball.
5. Surprising post games
I'm a sucker for guards and wing players who can work in the post, and this season I've got my eye on two players in particular as they feel out the benefits that come from playing with their backs to the basket.
The first is Sacramento's Tyreke Evans, who, in the midst of a solid bounce-back season, has quietly begun to work from the low block a bit. He's not doing so with enough regularity to act as a true weapon from that space for the Kings, but Evans has been remarkably effective on the occasions that he establishes himself down low and actively looks to score. According to Synergy Sports Technology, Evans draws a shooting foul on 35.1 percent of his possessions in "post-up" situations -- a shockingly high number inflated by his size advantage and his opponents' increasing desperation. As Evans begins to bump his defender closer to the rim, you can see the guard on his back begin to panic. At that point, it's too easy; Evans turns quickly, exposes the ball by bringing his shot directly in front of the defender, and draws contact on the post equivalent of the rip-through move. Observe:
Evans gets a ton of mileage out of that exact sequence, which helps to make up for the lack of diversity in his finishes otherwise. It's not as if he's been attending secret seminars on the sweeping hook shot; Evans basically uses the post as a launch point for a turnaround jumper or a quick drive, making this kind of ploy all the more valuable.
The other player to pique my interest was Indiana's Lance Stephenson, who has more of a physical advantage over opposing guards than he's often given credit for. Stephenson stands at just 6-foot-5 -- a fairly typical height among shooting guards -- but it's a sturdy 6-5. With that, Stephenson can bully smaller guards that line up opposite him via switch or mere mismatch, though he does so with irregularity.
I'm keeping tabs on Stephenson's post-up work more as an item for potential development than some area for immediate exploitation. The Pacers aren't likely to get more than a handful of possessions out of Stephenson down low over the remainder of the season, but he's especially slippery when he begins backing his opponent down from the mid-post. His style, in that sense, seems informed and aided by a lack of familiarity. Stephenson moves quickly and in such a way that makes it incredibly difficult to anticipate a potential spin before it happens.
6. Unexpectedly earning playing time: Austin Daye
I didn't expect Austin Daye to factor much into the Grizzlies' season, as the 24-year-old forward had shown little aptitude for putting his passing and shooting skills to use during his time with the Pistons. Chances were botched in Detroit and opportunities exhausted, yet his inclusion in the three-team deal for Rudy Gay made sense for salary reasons and gave the Grizz an extra shooter to potentially use down the line.
Yet Lionel Hollins seems to have taken to Daye rather quickly, and the recent import has responded with perhaps the best stretch of his career. Daye has only played a handful of minutes (11.6 a night) in 12 games for the Grizzlies, but over that small sample Daye has shot the ball very effectively, rebounded well, reined in his turnovers and made some standout defensive plays. It's been a nice run for Daye, though such a brief one that we can't yet draw any real conclusions.
7. A burgeoning connection in L.A.
The injuries across the Lakers' roster have done pretty clear damage to L.A.'s record, but they've also delayed -- and complicated -- the process of establishing a natural on-court chemistry. Steve Nash and Kobe Bryant have been forced to feel out one another at mid-season, rather than doing so in November. Dwight Howard and Bryant are still learning the best ways to work together, and how they should be spaced to orient the offense without sacrificing transition defense. The entire venture is a work in progress, and the on-court chemistry between certain groups of players will continue to trend upward as the Lakers grow more familiar with one another.
One such pairing worth keeping tabs on is the high-low connection of Howard and Earl Clark, two bigs still feeling out the best ways to work together on the court. Clark is largely used to space the floor and cut, but when he has the ball in his hands at the top of the floor he can make some interesting things happen for Howard:
Clark is both quick enough to actually get somewhere off the dribble and aware enough to find Howard for dump passes and lobs, which gives this particularly underused wrinkle of the Lakers' offense some interesting potential.
8. The accessibility of LeBron James
The uniqueness of LeBron James and his role with the Heat should go without saying at this point, but I'm struck by how little he has to scramble within Miami's offense to reset busted offensive actions. When a typical team runs a play action that doesn't quite work, often you'll see that team's best perimeter player run toward the ball. It's an effort to expedite the process of salvaging a possession, but one that often results in that player having to force a difficult shot over the top of the entire opposing defense.Dwyane Wade