2011 MIT SSAC: Soccer Analytics panel as it happens

It's Day 2 of the 2011 MIT Sloan Sports Analytics Conference, and it's the day of the Soccer Analytics session.  I'm really excited about today, not just because of the session, but also because I will be liveblogging it on my site.  It's the first time I've liveblogged anything at Soccermetrics (I've done it on Hexagonal Blog), so I'm not 100% sure of what to expect as to the response, but I hope enough people find it interesting.

The session starts at 10:30am Eastern time (7:30 am Pacific / 15:30 GMT).  Follow my liveblog after the jump.


1150: Soccer Analytics session is now over!  Thanks for following along!

1143: Session will wrap up with a question about Soccernomics, and whether there is more of a drive toward recruitment in developing (soccer) countries.  Aziza questions the tie between national economies and national team success, but it could be a surrogate for other factors.  Fleig says that the value of the book is motivating better questions by the soccer analytics people.

1140: I just asked a question about performance indices, with the Capello Index as an example.  There's a huge misconception about the nature of that index, and the panelists agreed that it does exist in the press and public.  Challenge is communicating that it is part of a larger picture.

1133: Is it easier for poorer teams to use analytics than richer (or elite) teams? Answer: yes, but it's harder for the poorer teams to keep them.

1130: Inevitable question about data being tied up by leagues and clubs.  Data rights is a huge issue in Europe.  Cultural resistance huge, but there are still interesting problems to be worked with limited data.

1130: Opening the panel for questions.

1127: Stein: What do you [Houston] find useful in your interaction with American sport teams?

  • Houston: One benefit — we're not competing with them!  Second — they are doing so much analysis, it's amazing to see the scope of what they've done.  Challenge is to work on the right problems in limited time frame.  (This will drive the release of data to public to motivate further development in analytics.  Begs for crowdsourcing.

1125: Graham: Video tracking is huge opportunity for clubs. "We're snooping around in the dark at the moment."

1123: Wooster gives an example of Thierry Henry's tactics in pacing his positions and runs close to the offside line.  The fans thought he was lazy, but the data showed he was actually very intelligent about his play.

1121: All Premier League teams have cameras, about 95% of Championship teams do, much rarer in League 1 and League 2.

1121: Strict video-tracking (big issue with occlusions).  No GPS permitted by FIFA.  (Really?)

1119: Wooster asked a question about what Prozone does.  Video-based analysis with eight cameras at the football match, measuring the player position, speed, acceleration, as well as ball location and referee location.  Challenge: how to use those data to develop other measurements, such as predictive modeling.

1118: Graham gives example of goalkeepers as a very high impact position, yet extremely undervalued in the market. (Corrected a typo here)

1118: Graham: Margin of impact for analytics is very high because of effect of random events in the game. 

1115: Houston: You don't need analytics to know that Messi and Rooney are great players.  They're useful to find the best role players, or the best players for their particular team.

1108: Marc Stein: Surprised by the level of public resistance to statistical analysis in football.

  • Graham: Soccer traditionally seen as sport where statistics are not important, and the data has not been made available to fans.  But that's changing, through (partially) the various performance indices.  "Traditional attitude of fans in England is anti-data"
  • Wooster: There's need in American sport to differentiate their content, to attract people to their sporting events.
  • Aziza: Borrowing European approaches to American soccer isn't going to work.  Inability to understand the sport in a statistical context is a limiting factor.

1107: Wooster: "Do you think other clubs are looking to Bolton for best practices for analytics?"  Fleig: The work at Bolton was an eight-year process, and driven by the financial limitations in the club.  Started with physical tracking data, added more systematic data collection to answer questions and recruit players with better knowledge.

1105: Hostility to analytics is generated sometimes by coaches seeing analytical approach as a threat.  The challenge is to have analytics being seen as a value-adding tool, NOT a replacement.  I've been saying that many times on this site.

1104: Wooster: "Data isn't the only thing."  I completely agree.  Analytics is a tool, but one tool among several.  Owners/managers still need to bring their brains.

1101: Fleig: Three aspects to analysis: recruitment, performance assessment, opponent preparation.  Understanding the basics about your team is very important.

1100: Houston: Next opportunity is comparative analysis of players across professional leagues of varying quality, and assessing foreign-based players.

1059: Wooster: Are we asking the right questions about the data?  (Me: Are we posing better questions that will drive analytical studies?)

1058: Wooster: There's a disconnect not just between the analytics people and the front-office staff, but also analytics and the coaching staff.

1057: Marc Stein has said very little during the session, and he's letting the panelists talk.  Very good moderator, he realizes that we're not there to listen to him!

1054: Do data lend themselves well to football? The panelists are giving the Barcelona-Inter Milan match from last year as an example.  Pass completion percentage and number of passes meant nothing to the final result.  I'd claim that passes DO matter, but in much more complex way.

1054: Aziza talking about analytics from a business organizational point of view. 

1053: Houston: Education process required to sell analytical approaches to front-office people and scouting heads.

1051: Fleig addressing the common claim that "the freeflowing nature of soccer doesn't permit statistical analysis".  It's true, but the response is to understand the characteristics of the team and develop a plan to make optimal use of it, driven by analytics.

1049: Graham making the point that Bolton were able to optimize their play for the squad that they had.  Interesting to note that they finished seventh one season with a squad rated in the lower half of the Premiership! 

1047: Fleig talking about his experience with Bolton Wanderers and the clubs embrace of an analytical approach to the game — developing a model of where the match is won, and what differentiates successful teams from those who get relegated.  Not a coincidence that they've stayed up over most of the decade.

1046: Wooster: Analytics much more advanced in the USA than in Europe, but soccer clubs have access to a wider range of data.

1045: Interesting to note that Steven Houston worked in the NBA.  As a matter of fact, he worked with Daryl Morey at the Houston Rockets.

1044: Aziza: Soccer clubs need to get beyond the idea of data being proprietary.  It's what is done with the data that is more valuable.

1043: Graham: Infrastructure and sponsorship very important in developing soccer analytics.  Made the point that without the support of the Times of London and Castrol, the performance index would have been difficult to develop.

1043: Fleig: Most of the heavy lifting in analytics will have to be done by the top-flight clubs, as they have the resources and staff to carry out the data collection and analysis.

1041: Steven Houston: Important to realize that the league does NOT provide match statistics.  All of that information is collected by the clubs (or for the clubs by the data collection companies).

1041: Ian Graham:  Challenge is to judge player performance in leagues of widely varying quality, using few matches.

1039: The new UEFA financial regulations for clubs ("Fair Play Rule") will drive a more systematic approach to player selection.  All of the panelists are making this point.

1036: I must advise that I am paraphrasing the proceedings of the session, not direct quotes

1032: Opening question: what are challenges in use data in sport without draft or salary cap?

  • Blake Wooster: Biggest challenge (or opportunity): how to make an impact with analytics — struck by participation of owners/general managers in other analytics panels.  Has not happened in soccer yet.
  • Steven Houston: lucky to work for one of the few clubs where salary is not an issue, yet people would be surprised by how much data is used in analysis.  Most of it is kept in-house by the clubs, however.  Getting the data is a MAJOR issue.
  • Bruno Aziza: addressing cultural gap is a big issue
  • Gavin Fleig: every club has analysis departments, but not focused on releasing data to fans.  At least in England, there's not a demand by fans for in-match data, they're more concerned with buying the ticket and watching the match.

1030: Marc Stein is moderating the panel.  I wasn't sure about it, but he's a genuine fan.  He's genuinely excited about this.

1029: Wow, it's started a minute early!  They're introducing everyone right now.

1028: Session about 70% full.

1025: Bruno Aziza broke his leg playing soccer last week.  Poor guy!  He's hopping over to his seat!

1023: I see the panelists for the Soccer Analytics session up front.  Here's a recap of the panel:

  • Bruno Aziza (Microsoft Business Intelligence)
  • Ian Graham (Decision Technology)
  • Gavin Fleig (Manchester City) – opponent preparation
  • Steven Houston (Chelsea) – scouting analytics
  • Blake Wooster (Prozone)

1019: Most of the audience is clearing out.  There will be plenty of seats available for those waiting outside.

1018: Ticketing Analytics session is now over.  Soccer Analytics will start at 1030 Eastern time (1530 GMT).

1004: Forum is opening up for questions from the audience.  Session should wrap up shortly.

0959: When I read this article on SI, I ask myself, "Why isn't Garth Lagerwey here?"

0955: Big challenge of ticketing people is secondary market – (legal) resellers.  One solution is for the leagues to bring those markets in-house.

0955: Key quote from Cole Gahagan (Ticketmaster): "We're doing a bad job of pricing our product.  Period."

0948: One discussion that I'm not hearing is ticketing issues in college athletics.  All of the panelists are from the professional North American sports leagues.

0945: I've been looking at the attendees list for the SSAC, and I am disappointed that there are so few MLS teams represented at the Conference.  I've run into a lot more officials from Premiership sides (Man City, Chelsea, Spurs, Fulham).

0944: About 45 minutes away from the Soccer Analytics session.

0934: Interesting quote: ticketing is one of the few industries where they know one out of three of their customers.  (The average customer usually buys close to three tickets.)

0928: It is interesting to hear the perspective of Lou DePaoli with the Pittsburgh Pirates on the challenges of attracting attendance to a team that has not been very good in almost a generation (the Pirates have had a losing record in MLB for 18 consecutive seasons).  Incredibly, the Pirates' attendance increased last season!

0922: Seating capacity is approximately 200.

0920: I would say that the conference room for the first session is about 70% full.  Not standing room only as was the case in other sessions (Basketball, Football, Referee).

0915: I'm sitting in on the Ticketing Analytics session to make sure I have a desk for my laptop.  It appears to be a good choice.