Impressions From the 2017 Sloan Sports Analytics Conference
The 10th Annual Sloan Sports Analytics Conference recently wrapped up, and it couldn’t have come at a more interesting time of convergence of new analytics findings and proliferation of analysis tools. The panels with the typical star power provided the usual slate of memorable quotes, but for me, the most interesting part was picking up on the trends that are emerging across different sports and organizations, and how analytics is an increasingly large umbrella of services. Here are some of my favorite observations I took away:
· Cloud computing is starting to skyrocket among even the smallest of analytics groups. Nearly every analyst I talked to mentioned either Amazon Web Services or Google Cloud Platform as their go-to resource for larger-scale database and computing power. This is a pretty sizable departure from older setups, where database management and IT procurement was someone’s full-time job and capital resources were required to build up an organization’s data capabilities. Now, analysts need only the most cursory training to spin up as much storage and computing power as they need, which will be a lot as tracking data continues to make its way into every sport. With procurement no longer an issue, the analysts and data scientists have become even more of the focal point of analytics groups since they can effectively provide their own infrastructure.
· Storytelling had probably its biggest emphasis yet. The last couple years, you could reliably predict the mantras and truisms at every panel that were repeated: focus on process and not outcome, avoid results-oriented thinking, small sample sizes, etc. I don’t know if those words fell on a lot more deaf ears in front offices, because there were entire panels devoted to communicating and storytelling with analytics, and how they can drive engaging and compelling stories. As more of the data-driven strategies become adopted and make their way into common wisdom, it makes sense that the higher-hanging fruit will come from more complex data sources, making communication the real limiting factor in how any of these findings get adopted. A panel of journalists was a pretty natural group to explain this importance; I would love to see some more stories from adopting these findings inside front offices.
· The range of what can be quantified is expanding. The hackathon switched from football to basketball data this year, with the explicit goal of using SportVU tracking data to try and quantify concepts that previously fell under more intangible-type measures. The presentations were great from top to bottom, with a lot of proxy metrics being developed for things like hustle, determination, and mental toughness. I expect a lot more proxy metrics to come out of these richer data sources going forward.
Now that technology infrastructure has become easy to assemble for pretty much all facets of sports, the crossover technology I feel like is coming any moment is the application of deep learning and/or AI to a lot of these decision-making processes. As teams get more aggressive in acquiring new data sources that grow in size and scope, they will have the capability, but not necessarily the implementation practice, of putting some of the more cutting-edge black-box algorithms to work. Everything from automated film study to tactical decisions will probably get tried on an AI module at some point, and the first group to figure out how to bring the right application from other fields will probably have an edge for a while. I doubt anyone will spill the beans any time soon at a panel or a research paper, but the flip side of infrastructure being easy to spin up is that approach is just as accessible to amateur enthusiasts to do it themselves as well.