Setting A Baseline with Synergy Sports

facebooktwitterreddit

Feb 11, 2015; Oklahoma City, OK, USA; Oklahoma City Thunder forward Kevin Durant (35) attempts a shot against Memphis Grizzlies forward Jeff Green (32) during the third quarter at Chesapeake Energy Arena. Mandatory Credit: Mark D. Smith-USA TODAY Sports

With today’s release (or really, re-release) of a consumer-facing Synergy Sports application breaking down play types, (and perhaps a return of the SORELY missed video indexing system) the stats-minded fan has received yet another massive datadump. Between the still not-yet-fully explored public SportVU data[1. I’ve written a ton about insights to be gleaned from that particular data set if you haven’t seen it.], shot chart information like those provided by Nylon Calculus’ own Austin Clemens as well as the excellent Peter Beshai, new metrics created by everyone from the NBA itself to ESPN, the amount of information available has increased exponentially. These newly available numbers offer far more granular detail, in some cases drilling down to the molecular or even atomic-level building blocks of the NBA game.

As with any advancement in available information, deciphering which bits are useful and interpreting their meaning is an enormous challenge. With no real[2. Sorry All-Star Game] game for almost a week, it doesn’t take much of a soothsayer to predict this new data will find it’s way into some of the endless “who’s better, who’s best” arguments. I’m not going to try to answer those questions here. Well, not all of those questions, but I did want to provide a little bit of context for the Synergy data.

First of all, in the past, I’ve been mildly skeptical of Synergy as a pure analytics tool – the raw numbers and rankings are the result of what has to be an imperfect categorization process, devoid of context and further provide a subtly biased view of a player.

As an example of the numbers possibly misleading, Alexis Ajinca and Ryan Hollins both show up as extremely efficient post up scorers. Even aside from sample size concerns, and especially once the video indexing feature gets added, what is likely to be shown is very limited players scoring with huge mismatches. What isn’t measured and won’t be seen will be all the times these players receive the ball in the post and do nothing but burn 5 seconds off the shot clock trying futilely to find an opportunity. While this is better than a turnover, it’s not good. But since it doesn’t use a possession, it won’t “count” against them. But simply going by their PPP, one might suggest feeding Big Alex as a viable strategy in New Orleans, and I don’t think anyone really believes that to be good advice.

My reticence to use the numbers whole hog isn’t really a criticism. As Synergy CEO Garrick Barr told me, the video functionality is intended to provide that missing context to a trained coaching or scouting eye. It should not be taken as rankings. Further, numerical data and video-based scouting really should not be considered different disciplines.[5. Much of the dreadfully boring “nerds vs. jocks” debate arises from this attempt to silo things which are in actuality friendly neighbors.] As long as there is important qualitative data not being captured in the numbers,[4. Meaning for the foreseeable future] scouting and video will play an absolutely vital role in player assessment.

That isn’t to say these numbers are useless. They can tell us such things as getting out of the way and letting Kevin Durant cook is a pretty damn good offense unto itself[3. Durant’s 1.24 points per isolation possession used are 12 points per/100 more than any other player with more than 50 isos on the season, with Kyrie Irving’s still stellar 1.12 points per iso the next closest.] Mostly, they can tell us a little about what is and isn’t working for a team.

A bit more context, since the numbers are only possession-ending plays, they subtly penalize ball-handlers. Assists aren’t credited but turnovers are which drag down the apparent efficiency of players who do a lot with the rock. Still, the numbers show that on a raw numbers basis, that creating with the ball is generally speaking less effective at putting points on the board than getting open for a look off the ball:

(The worst category “Misc” consists of plays so broken they can’t even be described as isolations, hence the miniscule efficiency. “Set Halfcourt” is all plays barring putbacks and transition opportunities.)

Comparing the percentage of time each category is employed to the percentage of points scored with each illuminates further:

A final takeaway from this brief look is the importance of getting easy buckets in transition. It’s cliche, but it’s also demonstrably true that scoring against a set defense is simply harder as indicated by the huge gap in expectations between half court and transition plays and supports and reinforces the notion that the shot clock can be a powerful defensive weapon.

It also means that perhaps a little bit of wildness can be excused and in fact encouraged. As an extreme example, relatively speaking, Tyreke Evans has been terrible in transition this season. His .91 points per transition chance is well below the league average of 1.1, while he turns the ball over almost 50% more often (17.6% of his transition plays are turnovers vs. an average of just over 12% leaguewide). He is in the 19th percentile of all players with sufficient transition opportunities. BUT, this .91 points per possession is still better than a league average half court possession. So even as poorly as his frequent fast break flurries have been, his ability to get into transition situations[5. Evans is 7th in the entire NBA in transition plays recorded.] arguably helps the Pellies offense.

In any event, there’s a ton more to decipher and decode here and I for one appreciate Synergy and the NBA giving us all this new data and information two days before Valentine’s Day.[8. My wife probably disagrees.]