World Conference on Science and Soccer

[mtamada]

http://wordpress.up.edu/wcss2014usa/

It's in its final day today; I only found out about it a couple of days ago. It's the fourth annual conference.

Most of the presentations are by coaches, kinesiologists and exercise scientists, and the like but there are a few presentations each day by sports analytics types. Susan Rudd (who won a contest by StatDNA similar to the one that the Sacramento Kings are running, and got hired by them shortly thereafter) is not presenting but her partner at the OnFooty.com blog, Ravi Ramineni, is. That was the only name that I recognized (not counting soccer luminaries such as Jeff Agoos). There are presentations which appear to be using SportVU data or similar video data, and researchers are experimenting with putting accelerometers on players.

The "soccermetricians" who attend the Sloan conference seem to agree that less progress has been made in soccer analytics than in basketball analytics, although I wonder if it is catching up. Given the worldwide size and wealth of pro soccer leagues, there's plenty of incentive and resources to do so.


This is the first time that the conference has been held in the USA, Portland, OR to be precise, which seems like a curious locale until one remembers that the University of Portland is a women's soccer powerhouse (and is where the conference is taking place), and the MLS Portland Timbers have sold out every home game they've ever played, and the NWSL Portland Thorns draw crowds of 13-14K, three or four times more than other NWSL teams draw.

I would drop by to see some of the presentations, but the conference fee is $450, and even a one-day registration is $200. The complexities of analyzing spatial data, which NBA analysts are only starting to wrestle with, are even greater in soccer (22 bodies to track instead of just 10, and on a larger playing surface) so as work on soccer progresses, researchers there may come up with breakthroughs or new techniques which could be applied to basketball.

[schtevie]

Hmmm. If there is not much found use for SportsVU data (and its ilk) in soccer, that doesn't augur well for such sources to have much effect in basketball.

[mtamada]

Right, but I think we're still a few years from being able to determine how useful the data are. It's a completely new and much more complicated type of data to analyze.

[schtevie]

Ah, but isn't it fun to speculate? And here's the thing. I enjoy the "low hanging fruit" metaphor. And in basketball analytics there has been a bunch of that available to be picked (not that much of it necessarily ever was) none of which, of course, derived from SportsVU type data. But with soccer (box score or +/- stats being being dead trees) one might expect (or at least hope for) some easy pickings to come from such information in ways which wouldn't expect in basketball, e.g. in soccer, it being far more important to establish the preconditions to scoring success and determining key features of the motion of players through what are much larger physical spaces would seem conducive to real opportunities using SportsVU technology.

Hence, the speculation: if there isn't (yet) any profit realized in its application to soccer, why is it reasonable to believe that in basketball, a game played in much tighter quarters, where "target zones of probable success" have long been well-known (i.e. those just around the basket and past the 3 point line) and strategy and tactics have long been established to maximize/minimize opportunities therein, that it is plausible to believe that one can expect that SportsVU-type data will lead to any epiphanies on how to improve play (never mind if so identified, that they could be practically implemented)? I would be interested to hear any hypothetical story as to plausible, potential, realizable competitive gains in basketball from such motion tracking. Maybe a recommendation that NBA players should fast break and press more often? (That's a joke....)

[mtamada]

Ah now I see your point. It's like Sherlock Holmes' dog that didn't bark. Or, where's the beef? I do not know, although I do think it's possible that the real future (and maybe even the present) is not in SportsVU or Vantage type data but rather Synergy where humans code the data. We see analysts talking about how successful a given player is at running or defending the pick-and-roll, using Synergy reports which have broken down the pick-and-roll plays into their various permutations for how the defenders react and whether the picker chooses to roll or pop or whatever and what the ballhandler chooses to do. I don't know if coaches use that information but it would seem to be useful stuff to have.

This article from 538.com seems to describe Opta as the soccer version of Synergy (i.e. using significant human data entry, although they don't seem to do any analysis or even description even though their job title is "analyst"), and Prozone as the soccer version of SportsVu (i.e. more dependence on cameras and computer-based optical tracking systems).