The Soccer Analytics panel at the 2014 MIT Sloan Sports Conference is a colossal waste of attendees’ time and should never return in its present form.
It’s taken a while for me to reach this point. I’m not pleased to reach this point. I know half of the panelists on this year’s panel. I knew all of the panelists on last year’s panel. I recognize that a number of people, some of whom I consider friends, may not like this post.
I no longer care.
Let’s recap the panel: appearing for the first time since 2011 was Steven Houston (ex-Chelsea and Hamburg), joined by Paul Nielson (Prozone), Robbie Mustoe (NBC), Jim Pallotta (Roma), and Taylor Twellman (ESPN) who served as moderator. You can read my paraphrased summary of the session at this post, but to be honest, I lost interest after 35 minutes and the quality of the liveblog went down considerably. (Any errors of interpretation with my paraphrasing are entirely my own.)
I came into the this year’s panel with very low expectations. I expected that the panelists would talk into vague generalities and rehash gripes of previous years: restrictive data practices, inability to make sense of complex in-match data, cultural resistance throughout the game. The panel managed to clear that bar, but it struggled to move beyond that threshold. So even though I tried to set myself up not be disappointed, I left the room feeling very frustrated.
I found it very telling that the high point of the session incorporated an data-backed decision making example from another sport (McLaren’s instructions to Lewis Hamilton during the last lap of the last race of the 2008 Formula One season). There were some attempts to take a deep dive into some issues, such as Gareth Bale’s transfer to Real Madrid or Gervinho’s move from Arsenal to Roma, but the discussions were too brief and superficial. Taylor Twellman asked if it would be possible to have “Oakland A’s-type” teams in European soccer, which led to a conversation about small-payroll teams that end up finishing high in the league table (Freiburg was given by Steven Houston as one example), but it didn’t go further than that. To stick with an analogy from another sport, the panel managed to get into second gear but struggled to continue moving up the gears.
It’s frustrating to come listen to a Soccer Analytics panel such as this when there has been very interesting work over the past 12-18 months. The paper by the researchers at Disney Research on automated formation detection was fascinating (more details in another post). I’ve written in previous posts about the presentations at the OptaPro Analytics Forum. It is very unfortunate that that knowledge, that those perspectives, are not given an opportunity to be presented at a SSAC panel discussion.
In my opinion, the wrong people are on the stage of the Soccer Analytics panel, from the moderator to the panelists. I recognize that the SSAC almost always has a representative from ESPN who moderates the panel, but as much as I like Marc Stein and Taylor Twellman, they do not have the knowledge of the soccer analytics space to ask the type of questions that can facilitate a deep discussion of the areas where football and analytics intersect. I nominate Chris Anderson.
I would have one, but no more than two, representatives of clubs at either research level or personnel operations level, and at least one football analytics blogger who can discuss models and metrics credibly and challenge the assertions of club officials. I guess in that respect I’m coming around to what Mitch Lasky and Zach Slaton called for last year. If the SSAC organizers need help figuring out which bloggers would be good panelists, contact me. I have plenty of suggestions.
The bottom line is that the Soccer Analytics panel is not worth attending in its current form, and if next year’s panel is more of the same, I won’t bother. There are much better uses of sixty minutes of my time.
Like I said, I may have things very wrong, and I don’t expect everyone to agree with me. Either way, let’s discuss in the comments or on Twitter.