Sloan Recap Day 1

By: James Eby, Suvrath Penmetcha (@savvysuv), Arvind Pendurthi (@arvindpendvrthi), Achyuta Burra, Maksim Horowitz (@bklynmaks), and Steven Silverman (@Silver_Stats)

An incredible first day at the Sloan Sports Analytics Conference in Boston left us eager for more sports discussion and exposure to groundbreaking research.  Talking and listening to sports professionals about their opinions and insights into the use of analytics in sports was engaging and thought provoking and allowed us to look at analytics in ways we have not before.  Below we shared some of our major takeaways from the panels/talks we attended yesterday:

Moneyball Reunion

The morning panel, we had the opportunity to hear from some of the more important catalysts of sports analytics’ mainstream recognition, including Bill James, Paul DePodesta and Michael Lewis. Each panelist offered advice to those in the audience, ranging from simple recommendations such as ignoring appearances when assessing player value, to the complex task of building flexible systems to adapt to the ever-changing professional sports environment. The most important takeaway, however, was James’ assertion that sports journalists should always have the self-confidence to put out their beliefs – even if they aren’t wholly supported by the data.

Analytics in Action

In this session, Shane Battier and Sue Bird claimed that there is a stigma toward advanced stats in NBA/WNBA locker rooms. Battier, who was one of the few players well-versed in advanced stats, said he often worried that certain analytics placed him in a negative light, which could have depressed his market value.

Jeff Van Gundy and Dean Oliver discussed the 76ers current state. Van Gundy claimed that there is no way of knowing whether the Sixers’ ‘process’ is actually one that will lead to success, due to the complex variables involved in the outcome.

Leveraging Digital Strategies and Analytics in Sports and Media

Professor Russell Walker (not the one from CMU) discussed the relatively new phenomenon of mass customization in media (e.g. Netflix, Amazon, etc. tailoring recommendations based on customer history). He suggested that sports teams could also use similar passive data capturing methods to learn more about the fans watching and attending sporting events.

Research Paper: The Pressing Game

Iavor Bojinov summarized his paper, in which he used spatial soccer data to catalog so-called ball-retention and defensive ‘disruption’ statistics for teams and players in the English Premier League. His paper suggested that by using this information, teams could determine positions of strength and weakness for themselves and their opponents.

Soccer Analytics: Finding New Passing Lanes

The main takeaway from this panel was the relative infancy of analytics in soccer compared to analytics in other major sports. In addition to on-field analytics, there is also a frontier for player acquisition analytics, due to the massive number of worldwide leagues and players. We also had a chance to ask a few question one-on-one with Paul Carr of ESPN after the panel. Here is what we took away from that:

Finding a career or internship in sports analytics can be an overwhelming task.

Although it sounds trivial, it needs to be said that you should find what’s right for you in the industry; that can mean finding the right sport, the right research problem, or the right team, you should be fully interested in that topic. Once you find your field, the next step is to explore it; delve into different questions you might have. Finally, you find a position that interests you and learn the specific skills and requirements for it.

Every analyst has to be able to communicate complex ideas to a client; it is an imperative skill in the industry.

When communicating to someone who lacks the technical knowledge you have, you need to boil it down to the basics. Usually jargon you use in your field will require some explanation. One good strategy is to start from the top and work your way down. For example: you can introduce a distribution, then a variance, and then an average.

Your analysis is not guaranteed to be accepted or used.

Many coaches and managers today may believe analytics is not useful in sports, or they might not trust an analyst. Therefore, analysts’ findings may often be ignored or unused. Trust can often be the issue in these situations; the management and coaching staff might not trust you and therefore reject your conclusions. One way to establish trust is to explore the issues coaches and managers are interested in, as opposed to a project you might find more interesting. Confirming staff’s beliefs through rigorous and valid analysis will likely create trust that will give weight to your analysis. It will also be helpful when explaining an outcome that invalidates preconceived ideas, as opposing ideas are much more easily ignored.

Modern Sports Finance

With franchises like the Cowboys, Patriots, Cubs, and others investing in non-sports ventures like housing developments, shopping complexes, and sports medicine centers, many investors and executives see sports franchises as investment conglomerates rather only sports. As franchises continue to add to their brand portfolios, their values will continue to surge upwards. The panel also discusses the implications of declining cable TV subscriptions. Game streaming options through companies like Google and Yahoo may be the way of the future.

Research Paper: Accounting for Complementary Skill Sets of NBA Players

One of the most interesting presentations of the day was “Accounting for Complementary Skill Sets of NBA Players” by Joseph Kuehn, a professor at CSU East Bay. The lecture talked about how to quantify a player’s impact on their. Professor Kuehn discussed how certain lineups can positively or negatively impact different players. The lecture showed a slide on players that have had the most positive impact from the other complementary players. Interestingly, three of the top five were Los Angeles Lakers players and Professor Kuehn went on to explain that since ball-dominant such as Kobe Bryant were not in the lineup last season the other players had more freedom on offense to score. Furthermore, the talk sparked future conversations of measuring how players fit within an NBA team.

Predicting Athletic Injury & Performance

Phil Wagner, president of Sparta Science, offered a unique predictive assessment of player performance, based explicitly on jumping ability. While this may not initially seem like a satisfactory measure of holistic performance, it makes more sense when segmented into the physiological movements involved in the act of jumping. Sparta breaks this down into a three-step process: Load-Explode-Drive. Based on the research they have conducted with clients, Wagner determined that the key components of player performance are their load and explode abilities, as well as upper stability, sleep, and nutrition.

Turning Data Science into Sports Science

According to Ray Hensberger of Booz Allen Hamilton, sports science is simply the culmination of technology developed to take advantage of raw data. The main goals of this field are health optimization, performance evaluation and game strategy information. In fact, a case study of the Golden State Warriors’ 2014-2015 season shed more light on how sports science can impact the game. Thanks to daily questionnaires, SportVU technology and wearable’s, Hensberger was able to determine that fatigue is a significant, important injury predictor. Of course, this is particularly valuable evidence for resting starters, in that injuries impact not just the players themselves, but the team as a whole.

1-on-1 with Gary Bettman

As Commissioner of the National Hockey League for the past 26 years, Bettman has seen some significant changes introduced to the sport. Surprisingly, he offered a progressive glimpse of the league’s future, asserting that technological advancements allow the executive office and league to function better as a whole. According to the Commissioner, the league is operating smoothly, with fan attendance at record highs, as well as “the best competitive balance in our history and in all of sports.” However, Bettman would not commit to analytics eventually resulting in a Moneyball-esque NHL team, stating that, “there is a human element, an emotional element, a chemistry,” that is unaccounted for in straightforward quantitative analyses.

Serving the Sports Fan with Analytics

Too often, in this day and age, casual fans get too caught up in the numbers to register the significance – or more likely insignificance – of the information they are being given. For example, with the Next Gen Stats available to the public through innovative player tracking technology, fans can now observe exactly how fast players are running during NFL games. While it may be interesting to know that Cam Newton ran 20 miles per hour on a particular play, did it tell us anything insightful about the game? This is the issue with modern analytics; there is an oversaturation of useless statistics. Rather, the key to providing meaningful data is contextualization, whether comparing the numbers to averages, teammates or opponents. As Nate Silver, founder of FiveThirtyEight suggests, “Focusing on the analytical process will yield better results.”

Started From the Bottom: Building a Business with Big Data

As Carnegie Mellon students, we were thrilled to have the opportunity to hear from a successful alum in Nik Bonaddio, CEO and Founder of numberFire. Bonaddio and his team offered the following insights to young statisticians hoping to venture into the field of sports analytics. Firstly, it is important to take risks at a young age; working for a startup requires you to learn new skills to adapt, and subsequently offers more opportunities to make a direct impact. Keith Goldner, the company’s chief analyst, stressed the importance of learning to scrape websites, as well as visualize data. Echoing the sentiments of the previous panel, Bonaddio asserted that, “We need to stop thinking about data and content as separate ideas; your data should be your content.” While the raw data itself should be supplementary, the data analysis itself should comprise the bulk of your argument.

Analytics in the NHL

Despite the presence of analysts from professional organizations, the panelists came to a consensus that it is never safe to go “all-in” with analytics. That is to say, a team’s decisions should never be based purely on what the numbers say; there are qualitative elements to the game that we have yet to quantify accurately. This statement was supported by Boston Bruins forward Chris Kelly, who alleged that team chemistry is a significant factor for players mentally, stating that there is a noticeable difference between a locker room with good chemistry, and ones without. However, there is always going to be a “feel” component to the game – try to quantify it anyway! For example, high probability decisions are a good measure of “instinct”; likewise, statistics can be manipulated to attempt to account for player contributions, which may not necessarily appear in the box score at the end of the game.


Our first day at Sloan was incredible and we are just as excited for day 2!   Once again we would like to thank all of our donors for helping us attend this event.  Follow us on Twitter and Facebook for constant updates throughout the event, and stay tuned a day two recap tomorrow!

Leave a Reply