Shuttering of Synergy a Blow to Scouting, Not Analytics

facebooktwitterreddit

Sep 29, 2014; New York, NY, USA; New York Knick forward Carmelo Anthony (7) is photographed by NBA photographer Nat Butler during Media Day at the NY Knicks practice facility. Mandatory Credit: Robert Deutsch-USA TODAY Sports

As I alluded to in my Industry Watch piece, the consumer application formerly offered by Synegy Sports, MySynergySports.com, has been shuttered as of this past Wednesday:

Synergy itself has not folded. As the above message indicates, “Synergy Sports Techinology will continue…” etc. In other words, the professional scouting package, the vaguely mythical “good/super Synergy” (the Glengarry Leads, for closers only) used by professional scouts and the largest media outlets, remains operational. Secondly, NBA.com has integrated many new video features over the last year and we’ve been promised more in the near future, which should replace some of what has been lost with the service no longer being offered.

However, clips have not as yet been collated in quite the same way. Unless the functionality is replaced, video-based play breakdowns just became a lot harder. For many a basketball blogger, analyst or simply extremely invested fan, this loss of functionality represents something of a hardship.[1. this hardship definitely qualifies as a First World type of problem as it seems extremely unlikely this application was being sold for anything close to enough to cover the actual cost of development and maintenance.] The service offered the ability to look back through the last five years of game action with plays broken down by numerous types and outcomes. In some ways it operated as an automated video cut-up service. Writing a piece about Blake Griffin’s improved post plays? Simply look up all of his post plays and spend the next 20 minutes or so watching hundreds of those possessions in a row.

What has not been lost is a useful analytics platform.

If I were made the czar of the basketblog universe, removing citations to Synergy rankings might not be the first thing on my list, but it would be in the top ten. The popularization of the tool led to wide though unintentional dissemination of bad information. It’s not that the delineation of plays into categories was without value. Certainly, identifying that Dwight Howard was extremely turnover prone while posting up or that Chris Paul rarely drew fouls in the pick-and-roll was useful information. But the totals themselves for a category were both prone to mislabeling and unrepresentative of a player’s overall ability in a given area.

(Noel is a software engineer at Second Spectrum, a company making heavy use of SportVU-based data for analytic support for teams. I discussed some of Second Spectrum’s work earlier this summer in this piece.)

Such coding inconsistencies were reasonably common in predictable areas. It’s difficult for even the most trained basketball watcher to categorize a play correctly, given how often different play types blend together. Differentiating between isolations, spot-ups or post-ups was often murky. This was especially true for players like LaMarcus Aldridge and Griffin who prefer the ball in odd areas such as the short corner or mid-post.

When using the service strictly for video scouting purposes, this was no big deal. In fact, these inconsistencies may have made the service more useful, not less, as the versatility of those players between face up jumpers, drives and back-to-basket play is an important point of strategy. However, these variations made any sort of comparative rankings next to useless. Already dealing with fairly small sample sizes of a few hundred plays at most in any given category, mislabeling of plays can badly skew results.  What can be a useful check on a player’s general ability in a certain situation — a post-up, a catch and shoot — is overwhelmed by the possibility of errors.

These small sample sizes also make even broad rankings messy. If a player had only 150 spot up possessions, making or missing a single additional 3 pointer would be worth 2 points per 100 either way — a massive difference in apparent skill.

Mislabelling and small samples weren’t even the most damaging biases introduced. The application only displayed possession-ending plays, which surely gives a skewed view of a player’s offensive accomplishments; what player doesn’t look like an expert shot creator if all you’re watching is plays where he shot? Carmelo Anthony is not credited for often drawing double teams in isolations, while some notable other players avoid censure for failing to create for themselves before throwing hapless teammates late shot-clock “grenades,” forcing rushed shots under duress.

And that same frame of reference exacerbated the inherent difficulties when looking at players’ defensive outputs, as the system would only display plays where the player in question was considered the primary defender. In addition to the inconsistency of the assignment as “primary defender”, this excluded situations where a player’s defense was so bad, a teammate was forced to attempt to defend for him. Nor does it account for situations where a player’s defensive breakdowns put the whole defense in help-and-recover situations, leading to open shots elsewhere on the floor.

Of course, I can’t give demonstrations of any of these examples because my favorite game footage archive service is no more.

So, while the loss of the video packaging elements is unfortunate, the removal of a very flawed tool from our collective analytic belt should not be mourned. It represents an opportunity to replace it with better defined, more accurate and simply more informative metrics in the near future.

R.I.P. MySynergy. We’ll badly miss some parts of you.