SSAC 2014: Soccer Analytics Panel as it happens

Welcome to the 2014 Soccer Analytics Panel at the MIT Sloan Sports Analytics Conference.  The panel is about to start, so let’s go!

Sorry, my liveblog tools have decided not to work…paraphrasing panelist comments.

Steven Houston (Hamburg) talking about the growth in analytics over last 12 months.  Especially from outside clubs (analysts, bloggers, media) and moving inside

Jim Pallotta (AS Roma): starting to benchmark players on physiology, what do you do with social analytics and translation to merchandising, sales, new stadium construction.  Experience with Celtics helped with use of analytics.  Still a very long way to go on uses of analytics in soccer, want to build internal platforms.

Paul Neilson (Prozone): How are teams using it to procure talent?  There is a culture of data in US sports that drives use of analytics in soccer in Europe.  Very much an evolutionary process.  Use of analytics very opaque (clubs not divulging work in analytics, reluctant to share data)

Robbie Mustoe (NBC): Discussing Gareth Bale, Wayne Rooney — are they good deals?  Not just a matter of analytics, but also commercial value.  Longevity, sell-on value huge factors in terms of large transfer and salary deals.  Definitely the case for Gareth Bale, a bit debatable for Wayne Rooney.  Biggest issue is PR, fan relations.

What about financial fair play? “Is there an opportunity for an Oakland A’s of European soccer?”

Pallotta: Importance of investing in youth talent, developing assets for club. “I don’t see how we will be purchasing a player like Bale or anyone like that” (referring to paying huge transfer fee).  Unbalanced pay scales for team can be destabilizing.

Mustoe: Can relate to that, giving example of Middlesbrough in 90s when Ravanelli and other Italian stars were playing at large salaries

Houston: Giving the case of Freiburg, who can do great things without large salaries.  Dortmund another example.

I’d say that Oakland As type teams in European soccer happen every year — problem is holding on to players.

Twellman: How to evaluate players at different ages, leagues, competitions?

Houston: Can find knowledge of value of player…how to determine value of player to his team?  (Example: Bale was valuable to Tottenham, has a very different value for Real Madrid) It’s difficult (no kidding), struggle to find KPIs.  [Yes, but could we put one example and have the panel fight it out?]

Mustoe: Asking what kind of weight scouts apply to data?

Houston: Scouts often see similar things that numbers observe, just told differently.  Progression of technical scouts and video.  But in-person observation will always be important.

Pallotta: Gervinho purchase very valuable to Roma, but not apparent from play in Premier League.  [Sometimes player performance dependent on his surroundings and the characteristics of the league.]

Is there buy-in from decision-makers?

Neilson: Huge influence from HIPPOs (highest-earning person’s opinion) who is often hostile to data.  Do those guys have the expertise to make those decisions more effectively?

Houston: Yes, there is buy-in

Mustoe: Two different problems: decision support and recruitment.  Short and long-scale issues.  Poses question: is it a manager decision nowadays?

Houston: In football, have had a lot of information, not knowing how to do with it.  We had positional data for long time and it went straight to coaches.  [Innovation isn’t going to happen from that path.]

Pallotta: There is a role for data, but decisions aren’t going to be made by machines like quant trading.

Is (x,y)-based analytics going to be used widely in soccer?

Neilson is giving an example, saying that it’s used mostly for fitness analysis. In-match analysis very very early.

Mustoe: yes we do have access to Opta, but rely mostly on video analysis, telestrator.

Neilson: Still using Prozone the same way now that was used 12 years ago. [That’s not reassuring]  Giving example of Lewis Hamilton’s World Driving Championship win in 2008, when McLaren made driving decision on model forecast incorporating data from cars, weather, past events, etc.

Houston: Data approach very primitive, lack of resources, etc. [We’ve heard about this again and again.]

Question time from audience.  Thankfully.

Houston: Everyone wants one golden statistic.  Very doubtful about that, and about the search for it.

Houston and Neilson talking about translation of statistical performance between leagues

Talking about FFP — Mustoe saying that big clubs generally can produce big finances and thus support the best players.  Will probably entrench big clubs.

You know, I really can’t be bothered with the remaining five minutes of this panel, so I’m cutting this short.  See ya.