In early March, Turnkey attended the annual MIT Sloan Analytics Convention in Boston. Highlighting some of the large analytical and fan-centric panels were most of the usual familiar faces including Daryl Morey, Nate Silver and Jonathan Kraft.

This being the 10th anniversary of the conference, many of the event’s large panels focused on the origins of the conference, and how it – and sports analytics – have evolved over the past 10 years. General consensus seemed to be that both collection and application of advanced data has advanced significantly; however, there is still room for improvement when it comes to data application and analysis.

One of the many sessions I attended was the “Moneyball” Reunion panel featuring Michael Lewis (author of “Moneyball”), Bill James (sports analytics writer and sabermetric pioneer), and replacing Billy Beane, Paul DePodesta (recently hired Chief Strategy Officer of the Browns, formerly Assistant General Manager of the Mets and Athletics). Like the book, much of this panel was spent recounting the MLB Draft of 2002 and reflecting on the analyses that went into the Oakland A’s draft day thought processes. Some interesting takeaways:

  • “In order to get from a question to an answer, you need a structured way of looking at the problem.”
  • The fact that analysts now need to evaluate how well 15-year olds playing in Venezuelan leagues will eventually do against Major League pitching illustrates how difficult their jobs are.
  • In 2002, the average MLB team ended up with 1 everyday player from their 50 picks, so the A’s goal was to get 2.
  • In examining player value, the A’s considered that a player’s physical “defect” (such as being overweight, short or – really – ugly) would erroneously deceive other teams into a lower evaluation of that player, which the A’s would rightfully ignore, when appropriate.
  • Kevin Youkilis was a player the A’s really liked a lot (not just for value), but he projected as a 10th round pick (in 2001), which led to a re-evaluation of their strategy. The Red Sox wound up drafting him in the 8th.

Some interesting points from the “NBA Analytics” panel, which featured Tom Thibodeau (former head coach of the Bulls), Brian Scalabrine (ESPN Analyst and former player), Mike Zarren (Celtics Assistant General Manager), and Brian Kopp (developer of the NBA’s player tracking technology, SportVU) included:

  • The analysts wrongly assumed that the new advanced data coming in would “flip everything on its head”.
  • Much of the current focus is on proper training to play the proper number of minutes, to find and replace at the times when players start slowing down, but most panelists believe that in 10 years, they will look back and laugh that they were using something as simple as “minutes played” to evaluate this. That’s because not every minute is equal – there are too many extenuating factors in player fatigue besides simply number of minutes.
  • The wearables that track physicality data ares only currently allowed to be worn during practice, which is misleading because the data captured isn’t the same as the data that would be captured during an actual game.
  • 82 games in a season is not good for the game if it results in fans seeing 3 starters sitting, even if that rest is what the players need.

The research papers and smaller solo panels were interesting as well. I may be biased, but I particularly enjoyed Chris Zeppenfeld’s (Senior Director of Business Intelligence for the Charlotte Hornets) discussion about how he tracks a wealth of datapoints about his customers (including Turnkey’s custom Lead Scoring Models, Surveyor data, retail behavior, touchpoints, reasons for not purchasing, etc) and utilizes them, each individually, in order to “score” his leads to predict likelihood of purchase.

#####