As I mentioned in my piece last week, I’m lucky enough to count myself a writer for Nylon Calculus, the new analytics arm of Hardwood Paroxysm (under the Sports Illustrated/Fansided banner). Several of my colleagues are supremely gifted data scrapers and manipulators, and are doing simply remarkable things with extrapolations of publicly-available data.
One bit of scraping I’m eternally grateful for has been done by NC writer Darryl Blackport, with some assistance in compiling from Krishna Narsu. Most of you who read me regularly will remember my frequent references to SportVU player and team data available publicly on NBA.com, but they’re not the only bits of such information on the site. Perhaps slightly less well-known in last year’s public roll-out were Player Tracking Box Scores – game by game only, but with several bits of data that aren’t available within the season-long database, and several others that allow for extrapolation if laid out over the full season. And extrapolate Darryl did: he scraped every individual box score, compiling season-long statistics by both team and player that expound on the data already available. And while I obviously can’t share any full databases, I want to highlight a few bits and pieces I found relevant within a Jazz context, using both Darryl’s extrapolations and other bits of publicly-listed data. Keep in mind these are snapshots, not anything remotely comprehensive, but they still have some interesting implications.
The Jazz had a plethora of issues shooting the ball last season, a fact that’s easily attainable without any sort of advanced information. They spent time in the early parts of the year flirting with all-time levels of awfulness from the floor before smoothing things out to simply bad, finishing the year in the league’s bottom third for both effective field-goal percentage and true shooting percentage. SportVU box score data gives us some further insights: they track contested shots versus uncontested ones[ref]Contested shots are deemed any shot where a defender is within four feet of the shooter.[/ref], one of the snippets of info that doesn’t appear on their publicly-housed yearlong stats. Now, the distance-only aspect of this differentiation means they need to be taken with grains of salt, particularly contested numbers – the closer to the hoop a shot is taken, the higher the chances become that said shot was “contested” under these guidelines given defenses’ proclivity for placing themselves in that area, up to the point where nearly every non-fast-break layup attempt (even those with no true challenge, essentially 90-95 percent shots for NBA players) will fall under the “contested” label.
That said, tracking the other end of the spectrum, or “uncontested” shots, can provide us with less noisy data. These shots can’t be convoluted by the possibility of non-challenges, because challenges simply aren’t possible with no defender within four feet. Accordingly, again excepting breakaway layups and dunks, such shots will trend heavily toward open jumpers, and therefore can be of some use.
As far as the Jazz went here last year, the picture wasn’t pretty. Utah ranked 29th in the NBA for uncontested field-goal percentage at just 40.7 percent, over 7 percent below Miami’s league-leading mark and mere tenths of a point above tanktastic Philadelphia. Again, these aren’t perfectly contextualized numbers, but they seem to match up decently with team success overall: 12 of the league’s 16 playoff teams were in the top half for uncontested percentage, meaning just four fell in the bottom 15, and vice versa. The top five teams for this category, in order, were Miami, San Antonio, Dallas, Oklahoma City and Phoenix, all of whom were in the top eight league-wide for per-possession offensive efficiency.
He’s been piled on unfairly by some in Jazzland recently, but unfortunately Trey Burke comes in as the worst offender here for Utah’s rotation players. He shot just 38.4 percent on 477 uncontested attempts – this was a top-40 attempt number for the entire league, and of these 40, only Josh Smith shot a worse percentage. Gordon Hayward was nearly as inefficient, posting the ninth-most uncontested attempts league-wide and converting at just over 40 percent, only three spots ahead of Burke among this same top 40 for attempts. Easily best among Jazz regulars was Enes Kanter at just over 46 percent, but the Jazz had only four players (Kanter, Gobert, Evans, and Jefferson) over the league average of almost exactly 43 percent. It speaks to an overall lack of jump-shooting prowess on the roster last season, and Utah will hope the additions of sharpshooters Steve Novak and Rodney Hood can boost things somewhat along with rejuvenated shooting years from Burke and Hayward, among others.
One element that could be involved in some of the still-present traces of noise in the above numbers involves another we can snapshot, in this case assists and assists per opportunity. This isn’t exclusively a box score tracking stat, but SportVU tracks “assist opportunities”, or passes by a player followed by a field-goal attempt which, if made, would result in an assist for the passer. Inserting a simple formula, we can find each team’s assist-per-opportunity, which is really a measure of two things: how well a team shoots the ball after passes, and how good those passes are in the first place.
The Jazz ranked dead last for assists-per-opportunity last year, and also dead last in a similar category, assists per total passes. Though there were a few more exceptions than the uncontested shooting numbers above, the top of the lists for both these areas mostly included top-10 offenses and vice versa. Quantifying what portion of Utah’s showing here is shooting skill versus bad passing is impossible given the information available currently, but I unfortunately lean toward the former – passing accurately has much more room for error and is intuitively far less integral than shooting. Not to beat a dead horse, but the Jazz just weren’t good shooters last year from any viewpoint, and it’s even possible that their miserable early season showing was closer to reality than the slight improvement their overall offensive efficiency seemed to indicate later in the year.
One area the Jazz stacked up well in was team rebounding, per SportVU’s rebound chance stats, defined as any time a player is within 3.5 feet of an available rebound. Utah ranked eighth for defensive rebounds per chance (62.6 percent) and ninth for offensive rebounds per chance (54.5 percent). The defensive number is of particular importance seeing as they gave up the second-fewest total misses in the league, and a failure to rebound at such a good rate would have sunk their already-league-worst defense to even further depths.
Within the roster, Hayward and Burke both get credit here – Hayward was the top rotation player for Utah, snagging 68 percent of his available boards, while Burke trailed just him and Diante Garrett, grabbing 62.5 percent. These elements can help compensate some for deficiencies in other areas, and the Jazz will surely be pleased at placing 13 roster members last season, including eight rotation players, over the league average of 57.8 percent[ref]Alec Burks and Jeremy Evans were the only two regulars below it.[/ref].
Again, these are just a few small pieces in the massive jigsaw puzzle that is player tracking data and its potential extrapolations. Like absolutely everything in this league, they must be analyzed in proper context and through an unbiased lens to be of optimal use, and here’s hoping the amount of data available makes this process easier and more detail-rich in future years.