The 11th annual MIT Sloan Sports Analytics Conference (Boston, March 3-4, 2017) brought together 3500 people from around the globe, and therefore can be seen as the must-attend event in this area of sports technology and innovation. A first blog post on this site discussed 11 characteristics of sports technology where current researchers, entrepreneurs and sport professionals in various countries are working on right now. But…
Does technology and figures really matter? The general answer on this question is becoming more and more: yes! From a variety of resources such as scientific papers, stories from the field, etc. it can be derived that a more data driven approach leads to higher effectivity than a pure intuitively one. Nevertheless, there are some boundary conditions that should be taken into account when trying to gain a more competitive edge with (data) technology, for sure in team sports.
People from teams already embracing this modern approach such as Midjylland, Manchester City, FC Barcelona and Aston Villa clearly stated that the main success factor is to have the management team working with a clear vision on the role of tech within their organization. The use of data should also fit the club´s and coach´s philosophy. Once this is the case, a culture of leadership and innovation can be gradually installed, ultimately resulting in an improved way of training and approaching the game. According to Daniel Stenz (former Vancouver Whitecaps FC analyst, now working with the Hungarian National Team), a strong commitment of different stakeholders in and around the club should also be there, in order to let people stick to the plan. Clubs as Borussia Dortmund and Sevilla for example keep on going on the same track, despite changes are also happening in those clubs. Unforeseen things happening, such as a new coach joining the club (and another one leaving), could have profound consequences for the club´s intellectual property that has been gained by collecting and analyzing data over years. Teams should be prepared for that.
Other success factors for implementation include the presence of role models in the team (‘if they use it, I will too’), proper player education, and a lot of hard work by people on the pitch e.g. helping players in the collection of information from questionnaires or the use of GPS trackers.
Besides what companies have to offer and regardless of the work to be done within the club, some innovation is coming from the scientific community. So what to learn from this year´s research competition? First of all, it´s clear that this is not a classic scientific abstract and paper competition, since almost all papers are based on data gathered from prestigious collaborations with companies, clubs, leagues, etc., mainly based in the US. Is this bad? It shouldn´t be so. It makes the papers not only interesting from a scientific point of view, but also relevant for sports and business. On the other hand, there appears to be some discussion about the MIT Sloan paper´s submission and decision process, with issues of irreproducibility in the selected papers, as discussed here. Nevertheless, you can view and download all papers on the conference website. Three attracted special attention from myself and many others:
- “BodyShots”: Analyzing Shooting Styles in the NBA using 3D Body-Pose Information: in this paper, 1500 labelled NBA three point shots were analyzed using markerles motion capturing techniques.
- Data-Driven Ghosting using Deep Imitation Learning: a paper on the Toronto Raptors´ software to predict what a defensive player should have done instead of what he actually did.
- Possession Sketches: Mapping NBA Strategies: in this paper, topic modeling techniques are used to identify types of plays in a basketball game, organizing game possessions accordingly.
Who won the paper competition? The winning paper was on the use of basketball analytics to increase interest in STEM (Science, Technology, Engineering and Mathematics) among young athletes. I think this approach could be an inspiration for many institutes in general, and ours in particular.
What about the startup competition? For your interest, I listed up below the presenting startups relevant in the area of athlete and team analytics.
- Decachord is a team’s playbook analyzing the interaction of players via passes and coordinated movements.
- Optimal Player Solutions provides Artificial Intelligence to optimize Big Data and enable GM’s to make more informed mission-critical decisions faster at less cost.
- RightBlue Labs builds software that accurately forecasts athletes’ injury, illness, burnout risk before any observable deterioration in performance
- RSPCT Basketball changes the way basketball professionals and fans understand, improve and enjoy shooting.
- Springbok Analytics visualizes muscle quantitatively, enabling teams and athletes to enhance performance, speed recovery from injuries, and assess injury risk.
- statUP is skill tracking software helping athletes earn stats that matter to coaches. Athletes take skill tests, earn stats, and level up.
- Kinexon Sports & Media develops cutting-edge solutions for precise localization and motion sensing of professional athletes. The portfolio includes sensor-based real-time detection of precise performance data and analytics.
- Noah Basketball is a data-service provider using 3-D computer-vision technology to provide real-time data and feedback to improve shooting accuracy.
- SciSports is the first to combine mathematical models, machine learning and fully-automated 3D Voxel Generation (BallJames) to create new insights about the world of football.
Overall, MIT Sloan is a very inspiring and interesting conference, with a strong focus on the business, management and application side of sports tech, however lacking a bit (and in some sessions a lot of) scientific body, which is surprising for an established organization as MIT. Nevertheless, it was worth visiting the conference given its large amount of sessions and its great networking opportunities.