A guide to the 9th annual Sloan Sports Analytics Conference, happening this weekend in Boston.
The Ninth Annual Sloan Sports Analytics Conference starts Friday morning at the Boston Convention and Exhibition Center in Boston. And even with its basketball pedigree (one of its co-founders is Houston Rockets GM Daryl Morey), there is still a lot for the sabermetrically-inclined baseball fan to take in. Armed with a media credential and a few years' experience, here's my primer on this year's conference for any first-timers.
If You Have a Ticket
A scheduling conflict moved last year's conference to the smaller Hynes Convention Center, which was located in the heart of Boston's Back Bay and accessible via a number of several MBTA Green Lines trains. This year, however, the conference is back in South Boston at a more modern and larger facility. The downside is that the BCEC is a little less accessible via public transportation, especially given the amount of snow dumped on the city over the last month. Fortunately for those staying in the city, the subway system is operational again: take the Red Line to South Station, then the Silver Line (SL1) to the World Trade Center stop. For those coming from outside the city, commuter rail service has been decimated to the point where you might want to consider driving; the center has at least one garage, and parking is available for $15.
If You Don't Have a Ticket
Even at $600 for a full-price ticket, the conference sold out months ago. If you missed out, there are still ways to get your fix. The finalists for this year's research paper award are available online. A live webcast will also be available this weekend, but it looks like you can avoid the $30 fee with clever use of their two-week free trial. And anything really interesting gets splashed all over Twitter in a matter of seconds.
Panels
The meat of the conference is the panel discussions, featuring a number of owners, general managers, media members, and the occasional player. The whole schedule is available here, and bios for the featured speakers are provided here. Below I've listed the panels I'm most interested in, in descending order.
- Ask attendees the highlight of last year's conference, and your responses will most likely be split between new NBA commissioner Adam Silver's one-on-one interview with Malcolm Gladwell and MLBAM's unveiling of StatCast. This year's conference seems to have combined both into this year's Commissioner's Perspective session (Friday, 1:00), with new MLB commissioner Rob Manfred talking one-on-one with Brian Kenny. The discussion will cover "player health, pace of play, changing viewer demographics and global marketing initiatives," but don't be surprised if we get an update on everyone's favorite video-based player tracking system.
- Related to Manfred's discussion of player health is Dr. Glenn Fleisig's talk, Analytics of the Tommy John Injury Epidemic (Saturday, 9:00). Fleisig is a member of the American Sports Medicine Institute that helped design MLB's Pitch Smart guidelines released last summer. Fleisig also consults for Motus Global, the company behind the mThrow sleeve.
- Then, of course, there's the old tried and true Baseball Analytics panel (Saturday, 10:20), featuring lots of smart baseball writers trying to get a sitting general manager to divulge trade secrets. This year, Dan Brooks, Dave Cameron, Ben Lindbergh, and Jonah Keri are the inquisitors and Mets GM Sandy Alderson will be under the microscope, a position he'll find less fun than jury duty but more fun than talking to a naked guy about the Mets.
- Finally, The Economist's sports editor Dan Rosenheck will be presenting Spring Forward: Understanding Spring Training Statistics (Saturday, 12:10). If last year's talk on the importance of holding base runners was any indication, this year's talk will also be very interesting and crucial to one of the wild card games. Mariners, Indians, you've been warned.
Honorable Mentions
There are a number of talks related to wearable technology, even if none of them are specifically baseball-oriented, including Zebra's Wearable Technology Analytics and Catapult's Sports Science: Performance Analytics panels ... Franchise values are spiralling out of control, and the Valuing Franchises: How Sports Teams Break the DCF panel will feature a number of academics trying to figure out why ... If the Braves' recent dealings with Cobb County are any indication, your favorite team is probably due for a new stadium. If You Build It, They Will Come: Designing the Stadium of the Future will at least give you a sense of some of the new bells and whistles you can expect ... New to the conference is the Booz Allen Hamilton Data Science and Visualization Zone, featuring interactive demos and demonstrations as well as talks on using and presenting analytics.
Research Papers
The research papers will always have a special place in my heart: unlike the team officials, these presenters are more than happy to talk about their metrics and the thought process behind them. Think of it as scouting the next wave of baseball ops hires. Like last year, there are three baseball-related papers among the eight finalists; unlike last year, there's a separate baseball track, suggesting that a number of prizes will be awarded. Here are some initial thoughts on this year's finalists.
Diamonds on the Line: Profits through Investment Gaming (PDF)
Clayton Graham of DePaul presents a new way to model the runs scored per game by each team, accounting for park factor, opposing pitchers, and lineup strength. By comparing the model output to the day's betting lines, he then selected games to bet on and charted his performance over a season. His final result was a $14,000 profit off a $1,000 initial investment, nearly all of which came in one month towards the end of the season. It's unclear from the paper what the training set was, and so it's impossible to know how well the strategy will generalize to future seasons.
Evaluating the Effectiveness of Dynamic Pricing Strategies of MLB Single-Game Ticket Revenue (PDF)
Three students from Penn's Wharton School of Business used ticket sales and pricing data from "an anonymous MLB franchise during a recent MLB season" and created a model to predict the revenue associated with a given dynamic pricing strategy. The authors claim their model increases revenue compared to the real-world strategy employed by the team. The sample size is understandably limited by the amount of data provided by the anonymous team (who, to make matters worse, only implemented dynamic pricing in the second half of the season). But once again, this raises concerns of overfitting, especially because the anonymous team collapsed down the stretch: "Since mid-July, the home team had a sub-.350 record for the last 70 games of the year."
Who Is Responsible for a Called Strike? (PDF)
Lastly, Joe Rosales and Scott Spratt built on Dan Brooks and Harry Pavlidis' catcher framing model, dividing credit between the batter, pitcher, catcher, and umpire. Rosales and Spratt used Baseball Info Solutions' proprietary catcher target data to augment the typical PITCHf/x data sets. The result is a much smaller spread of values (Hank Conger, for instance, saved 16 runs according to the BIS model in 2014, compared to 37 runs according to Brooks and Pavlidis' mixed model). But Tom Tango has objected to the methodology, voicing concerns that Dustin Pedroia's four extra runs are due to his height, not necessarily his proficiency at avoiding called strikes.
. . .
Bryan Cole is a featured writer for Beyond the Box Score. You can find him this weekend at the conference, and the rest of the year on Twitter at @Doctor_Bryan.