For a time, after sports teams became obsessed with using data analysis to inform decision making and improve performance, there was a shortage of data for analytics teams to get their teeth into. This, clearly, did not last long, and we quickly had a situation in which a plethora of new and established companies were creating hardware to record just about everything that happened on the field. From the velocity of an athlete's sprint to the angle at which a baseball strikes a bat, the events of any given sport are now quantifiable.
The problem with this, however, is that analytics teams were very quickly swamped by a sea of data that was without adequate structure and subsequently lacked any real meaning. What good, for example, is knowing the amount of pressure applied to a basketball player's knees when they take a three pointer, if there is no long term injury data to assess it against. Because of this, there is a premium on data scientists that can make these connections and present them in a way that makes sense. Perhaps most important in all of this is to take complicated data and simplify it into actionable insight.
Simplicity extends to knowing the limits of any given analytics program, and understanding that analytics is really just about providing better information for coaches and other staff to work with. Speaking at the Sports Analytics Innovation Summit this July, Nike Performance Specialist Keith D'Amelio discussed the use of technology in sports analytics. One of the key points to his presentation was that too many teams are trying to find the analytics programs that can be rolled out across their entire organization and revolutionize all working practices. Each program is far too unique for this. 'This is also in part why, when teams are very reluctant to share their information, and keep everything a secret - why?' Keith asks. 'I could never recreate your situation. It doesn't matter if you tell me the secret algorithm that you guys use. I don't have your players, I don't have your coaches, I don't have your management. I could never recreate what you've created.'
Keith also goes into the problem of obsessing over new technology, particularly when that technology is useful only for data collection. 'In 2017, ballpark, wearable technology... is a $6 billion industry. It's projected to be a $25 billion industry in 2019. If you think there's some sh**ty products out there right now, wait 18 months.' There are constantly new products appearing on the market to measure key (and often not key) metrics in innovative or more convenient ways.
The tech is often impressive but, as Keith notes, analytics teams are adopting new products with little regard for both their necessity to the team, or their accuracy. 'We have to understand the error in it. Because if it has a measurement error of 4% but I'm running and telling my coach when my athlete has changed 3%, what am I doing? If I don't understand that, it's very difficult to make meaningful decisions.' Keith insists that he would take a person over a piece of technology any day of the week; technology alone won't solve any problems, its intelligent users that have to know how to contextualize the information the technology presents.
Validity and reliability are, of course, important elements of sports analytics. Meaningfulness, though, is an area that is not given its fair share of the conversation. 'Just because you can collect the information,' Keith says, 'doesn't mean it actually matters.' There is a huge movement within sports analytics to find the best ways of taking all the myriad metrics the technology can collect and them down into the things that actually affect performance. To be meaningful, Keith posits, information should fundamentally help coaches, managers, and athletes make better decisions.
Ultimately, it comes down to doing the basics well and ensuring that everyone involved in the data process from those collecting it to the coaches putting it into practice, simplicity is key. 'If I'm an owner,' Keith says, 'and I'm going to invest financial resources (amongst others) to hiring a sports scientist, to hiring a performance director, do I want to hear someone who's going to tell me 'we're going to do the basics and we're going to do them really, really well'? Or is it better, more sexy, for me to hear that we're going to do everything under the sun? And I think that gets lost. And then it becomes this constant circle of 'we have to do more, we have to do more' to justify it.'
This is why effective data visualization software is so necessary. When presented in a clear and digestible manner, even the most complex sports analytics reports can be simplified enough to be useful. Technology companies have a responsibility to ensure not just that their hardware is reliable and that it collects a worthwhile metric, but also that its software can present the information collected in a meaningful way. This, in turn, affects how coaches can present information to the athletes themselves. Just because an analytics team can understand a complex set of data, it doesn't mean the team's linebacker can - data visualization software can make the process of disseminating the findings far easier.
So meaning in sports analytics comes down to three key points. Firstly, teams should understand the limitations of the data, creating analytics programs for specific tasks and outcomes rather than looking to overhaul the way an organization works. Secondly, analytics teams should dampen the obsession with new technology, particularly when that technology adds nothing more than a new metric to consider. And, finally, teams should always be looking for ways to present the information collected in a simple, digestible way so that coaches can actually act on it. Keith's presentation revealed a frustration with an industry that he finds both incredibly exciting and often problematic. But, if teams are geared towards simple outcomes, we will see the industry move forward in a positive way.