2013 SSAC Preview: A look at the Research Paper Competition

This is the first in a multiple-part series of posts that preview the 2013 MIT Sloan Sports Analytics Conference.  In this post, I will consider the finalists of the Research Paper Competition.

The Sloan Sports Analytics Conference started a Research Paper session in 2010, coincidentally, my first year at the conference.  It’s been a very popular addition from the start and has grown into a more sophisticated and formalized process.  First, researchers have to submit an abstract of their work, and a selection of those abstracts are invited to submit full papers.  The finalists have posters and/or oral presentations at the Conference, and the winning oral presentation receives US$10,000 (the second-placed finalist also receives a cash award).

It should go without saying that I’m a huge fan of the Research Paper Competition.  As the SSAC transitions to more of a sports business conference (not a surprise given that it’s organized by a business school, but more apparent with ESPN as the lead sponsor), the Research Paper Competition keeps the SSAC close to its roots and keeps me coming back.

Last year I noted that the final paper list was overwhelmingly dominated by basketball topics.  Basketball represents a plurality of the final papers this year — four of the oral finalists (all NBA-related) and none of the poster finalists.  The finalists consist of three papers on baseball (one oral, two poster), two on ice hockey (one oral, one poster), one oral presentation on American football, and one poster presentation on tennis.  And for the first time, there is a soccer-related paper as an oral finalist.

Below is a thumbnail sketch of the eight oral finalists.  Hopefully these notes will motivate you to read their papers, and watch their presentations!

The hidden foundation of field vision in English Premier League soccer players (Geir Jordet, Jonathan Bloomfield, Johan Heijmerikx)

  • Examines player head movements off-the-ball, thanks to Sky Sports PlayerCam
  • High correlation between exploratory movements before taking ball, and performance on ball (pass %, esp forward pass %)
  • Effect strongest in midfielders, independent of field location…forward pass completion performance only significant for midfielders
  • Best players keep their heads like a swivel

Total Hockey Rating (THoR) for NHL forwards and defensemen based on on-ice events (Michael Schuckers, James Curro)

  • two-way performance evaluation of NHL forwards/defensemen
  • two-way: expected goals created and prevented
  • take in account every on-ice event in NHL match, teammates/opponents on ice, probability that event results in goal within next 20 seconds
  • a hockey analogue to a Wins Above Replacement rating

Live by the Three/Die by the Three? The price of risk in the NBA (Matthew Goldman, Justin Rao)

  • determine optimal strategy for 2-/3-point shot taking using game theory
  • tested a couple of assumptions about team risk as winning/losing team
  • as teams become more risk-loving, 3-point efficiency drops
  • conclusion: it’s against a winning team’s interests to become risk-averse
  • comebacks in NBA occur more often because…
    • losing teams motivated by scoreboard
    • shot selection for leading teams sub-optimal

A new ensemble of interior defense analytics for the NBA (Kirk Goldsberry, Eric Weiss)

  • popular defensive stats are misleading, not necessarily related to each other
  • who is good at preventing the other team from scoring, or reducing their scoring efficiency? (Dwight Howard effect)
  • apply spatial analytics to new defensive metrics in NBA
  • notion of spatial splits: close-range/mid-range/3-pt range
  • shot frequency and shot efficiency (percentage)
  • main case studies:
    • basket proximity — how does shot frequency/efficiency change when defender within 5 feet of basket?
    • shot proximity — how does shot frequency/efficiency change as function of distance between shooter and defender?
  • best interior defender: not Dwight Howard (Larry Sanders)

The value of flexibility in baseball roster construction (Timothy Chan, Douglas Fearing)

  • what is the most optimal way to construct a roster in baseball, given:
    • positional flexibility of players
    • injury risk
  • major contributions:
    • injury rate estimation, given changes in fielding positions
    • optimization models to assess vulnerabilities to injuries
    • quantify value of flexibility to MLB teams in 2012 season

A quantitative look at the relationship between offensive rebounding and transition defense in the NBA (Jenna Wiens, Guha Balakrishnan, Joel Brooks, John Guttag)

  • upon missed shot, player can either go for rebound, hold current position, or go back on defense
  • what is best strategy? quantify through SportsVU data (3D position data of ball, 2D position data of players)
  • Crash Index (number of players in neutral positions — 8-33 ft away — who move toward basket)
  • Retreat Index (number of players in neutral positions — 8-33 ft away — who move away from basket)
  • finding: getting back on defense early reduces opposition’s shooting efficiency, but reduces probability of offensive rebound by 50% (no kidding!)

Acceleration in the NBA: Towards an algorithmic taxonomy of basketball plays (Philip Maymin)

  • how to incorporate acceleration in graphical description of basketball plays
  • data: SportsVu optical data in half-court situations (N=233 games)
  • determine acceleration levels of teams/players in offensive/defensive situations
  • identify regions of half-court characterized by “burst points” – similar rates of acceleration/deceleration
  • use these regions to describe player motions around half-court

Predicting the likelihood of field goal success with logistical regression (Torin Clark, Aaron Johnson, Alexander Stimpson)

  • Thesis: distance is not the only factor in success rate of field goals in NFL
  • binary logistic model of field goal rate, include distance, environmental conditions, game scenarios
  • findings: distance and environmental conditions matter most, game scenarios not at all
  • compute “added points per attempt” — points added by kicker above a generic average kicker
  • rank stadiums by difficulty in scoring field goals
    • most difficult: outdoor stadiums in cold-weather cities
    • least difficult: indoor stadiums, and Mile High (Denver)