The growth of analytics in sports analysis and team construction has been explosive. The evolution of data science in the realm of sports has been well-tracked, and it seems like the biggest step teams can take to gain a competitive edge is to invest in an analytics department.
And though the word “analytics” has been getting much traction in sports coverage circles, teams have been slower to adopt the changes suggested by the rapidly advancing data science community.
That community came together in Stevens Point, Wis. last week at the Great Lakes Analytics in Sports Conference to discuss not only the steps made in modern analytics, but how to move the industry from one of theory into practice. Hosted by the University of Wisconsin-Stevens Point, over 45 speakers from a variety of backgrounds shared their insights into the data behind sports.
More than anything else, it seems as if the consensus among those presenters was that the practitioners of analytics need to also be the best salespeople for the approach.
Sometimes, that might mean literal salespeople — companies like OaSIS and Competitive Sports Analysis mean to break down data generated at the high school and college levels to allow teams to identify which players can best make the next step, while Zebra and Catapult are making sure to find new data in the form of tracking players during practice and games using wearable technology.
Football fans have become increasingly familiar with Pro Football Focus, a firm built up from the ground as a passion project to grade every player on every single play of an NFL season. They too have invested more heavily into ways to analyze the rich data they’ve collected and produce meaningful solutions for NFL teams.
And while corporate entities are doing a fine job of creating an emerging industry, they won’t be the only ones doing the heavy lifting for the credibility of statistical analysis. Individual consultants for teams or even in-house analytical personnel will also need to make pitches that justify a new approach to sports decision-making.
Much of the conversation at the Great Lakes Analytics in Sports Conference was centered not on the novel new concepts that entrepreneurs and academics have innovated, but on ways to communicate what they’ve learned to teams.
In a talk titled “Defeating the Stigma of Sports Analytics,” Johnny Carver, a consultant who has worked with NBA teams like the Indiana Pacers laid it out clearly. “If no one is listening to you, what you do won’t matter.”
That communication means making sure that analysts understand the big picture. They’ll need to not only familiarize themselves with the sport they advise for, but the coaching and front office decision making process already in place.
Carver pointed out in his talk that coaches knowingly make decisions that may inhibit the team’s chances in the short term in order to generate long-term value; criticizing those decisions may not make sense without understanding the already extant reasoning behind those decisions.
For example, younger players may be asked to take shots in a basketball game that they’re not very good at making in order to develop their skill-set, or veterans may be sitting longer than optimal in a game to preserve the long-term seasonal goals.
“You’d be surprised at the number of analysts don’t watch baseball, basketball or football. This is important because you’ll miss something,” Carver said in his talk. For example, when Carver was breaking down statistics for NCAA basketball teams, he had to account for the unique 2-3 zone that Syracuse plays and how that skews numbers going forward. Without accounting for that – something he gathered by watching the game and choosing not to solely rely on tracking data – he could better prepare the University of Arkansas’ men’s basketball team.
That understanding of the sport — more than the understanding that comes with data analysis or casually watching on the sidelines — will be the only way front offices and coaches can trust the conclusions that come from statistical examination.
“If no one is listening to you, what you do won’t matter”
This insight isn’t new to Stephen Shea, the chair of mathematics at Saint Anselm college. In consulting with basketball and hockey organizations, he’s found deep familiarity to be an enormous asset in his work.
“A couple years ago I went to an NHL team after doing a pilot study with them,” he explained, “and I sat down with them to see if we extend the project. I sat down with the GM and his team and they started asking all these questions. And I did the best I could to answer them. In the end, they decided that they weren’t going to continue the project.
“I left with this thought, ‘well they just don’t believe in this, they don’t know what’s going on.
“A year or two later,” he continued, “I embarked on a serious project in hockey and I started uncovering styles and ways teams are playing. I go back and I’m looking at the way this team’s playing at the time, and their types of strategies and then I start recalling all the questions that they asked. It all came together—it all synthesized—and I realized what they were trying to do; what they were after. I didn’t have the right answers.
“It wasn’t that they’re not savvy to analytics,” he explained. “It was that on my side, I didn’t know enough about what they were doing.”
That insight has changed his methodology when it comes to advising teams. “I think the real challenge for the people on the analytics side,” he said, “is to truly understand what’s going on. On the field, or on the court, or on the ice—truly understand the way coaches are thinking.”
“I know that my coach in the last two minutes of the game, these are the things he’s focusing on. Well, that’s going to tell me what I need to feed him, what kind of information I need to find or what kind of things I need to research. I know this coach likes to use his shooting guard in a particular way. Without him even asking for me to do a particular analysis, I can start that and I already have an idea of the types of things I’m going to report to him.”
Michael Schuckers, the Rutherford Professor of Statistics at St. Lawrence University, emphasized the importance of understanding multiple viewpoints when talking to teams. In order to overcome inherent skepticism about the approach, Shuckers said analysts need to “break it down to the most simple analysis,” that distill the data into its most important conclusions.
These advisors will have to “spend time integrating into the teams, so that they have an understanding and are talking the same language. When we’re doing the communication, it’s always in terms that they can understand and it needs to have clear objectives that they can use.”
Diane Bloodworth, CEO of Competitive Sports Analysis, thinks teams are getting to a tipping point in their acceptance of analytics. Her company focuses on recruiting analytics and determining how to find the most appropriate fit for a player and a coach in the context of school culture, individual skill set and overall scheme.
“I wouldn’t mind if it tipped a little faster,” she confided. Her company is one of many that’s growing to fill a need for more data-oriented strategy in the college football and basketball world. “I think [coaches are] looking to leverage analytics if it can help them win – and only if it can help them win – and I think they’re looking for some assistance; they’re looking for it to be fairly straightforward: something that’s understandable to them.”
Knowing that communication is the critical challenge, Competitive Sports Analysis made sure to design an intuitive interface for coaches to use, and hired an NFL scout to help design their product to make sure that it can speak the language coaches are using.
At the end of the day, the drive to win should encourage coaches to invest in data-driven techniques. Coaches at the college level and in the NFL have already begun embracing certain types of analytics. Sports nutrition has taken off, and teams have invested more into individualized programs that can maximize the work players are already putting into the practice field.
Not every audience for analytics is a member of a professional or collegiate sports team. Communicating new insights presents challenges for journalists as well. No one knows that better than Fox 4’s Edward Egros, the weekend sports anchor for the local Dallas television station.
A polished presenter, Egros provides more than human interest stories when covering local teams — he uses his master’s degree in predictive analytics to make sure he provides the most incisive sports coverage in the market.
In his talk, “Sports Analytics in Television Journalism,” Egros covered what he saw as the essential three challenges to bringing analytics into the broadcast world. The first is that a portion of the audience simply won’t be interested in statistics—that comes with the territory for any broad journalistic approach.
To combat that, Egros makes sure to emphasize qualitative descriptors when summarizing his conclusions. Not much is lost, for example, when describing that a team has an “extremely good chance of winning” next week’s matchup over drily arguing that they have “an 87 percent chance of success.” Both of those statements communicate the same idea, and there’s not much value in the precision one lost in moving from the second sentence to the first.
https://www.youtube.com/watch?v=cYD_gL1wWDY
Not only that, he found that it has helped to explain many of the advanced stats as he moves through the broadcast. When discussing which teams might advance in the March Madness bracket, he’ll argue that some statistics that are emphasized by most media aren’t all that helpful — he’ll start with a bold statement, like “defensive 3-point field goal percentage is useless,” and then explain: teams who are good at defending the three prevent 3-point attempts and force alternative shot selection with their defensive spacing. Looking for that statistic to provide an edge can be a fool’s errand.
The second challenge, Ergos said, was that some concepts are just too complex to describe neatly in a television broadcast. Sentences can become clunky with qualifiers when attempting to capture the many facets of a powerful, but complicated, statistic. In response, he’ll use statistics that are a little less explanatory, but much easier to communicate.
His preferred passing metric of choice in the NFL is something like adjusted net yards per attempt, which assigns bonuses for touchdowns and penalties for interceptions all while incorporating sacks into the otherwise standard yards-per-passing-attempt statistic. But on broadcasts, Ergos opts for net yards per attempt (without the “adjusted” modifier preceding it), which is easier to explain: it’s yards per attempt, with sacks.
The third obstacle is one that’s easy to identify — the tighter time limits for television. In some sense, planning can overcome this problem. With enough polish and practice, one can condense a broadcast and meet stricter timelines. But with so much left to communicate, Ergos also chooses to expand his thoughts elsewhere, like on social media or his website, insidesportsanalytics.com. He’ll promote those platforms on his broadcast and direct viewers to more thorough explanations of his concepts.
These difficulties were well worth it to Egros, and it may be something critical to the industry of television sports journalism. “I thought to myself, ‘for TV to keep up with the times, it’s gotta do something that other media has embraced,’” he told me. “We’ve talked a lot about how newspapers are dying, but sometimes newspapers can be the most innovative. The business model says ‘you have to go in a certain direction,’ and they are. They understand that.”
It isn’t just a case of innovation meeting needs, either. For Egros, he can tell a more complete story with the data we have available. “You can talk to the right people and still get a robust story. I think it really adds a dimension in the right cases. [Analytics don’t] have to be used every time — sometimes there’s stories where there really aren’t stats to get into. But, when you do find something, it can really sing if it’s done properly.”
He thinks that this model of sports journalism in television will grow. “As much as I like doing something that almost no one else is doing in TV,” he said, “I hope more people do it, because I think then it improves the tools and I think it improves the conversations that we have.”
Just as an analytics consultant needs buy-in from the general manager, coaching staff and players, Egros needed buy-in from his colleagues and superiors at Fox 4 in Dallas. He found it by being upfront about his approach and being fortunate to have a business environment that embraced a new approach.
“The reception at Fox 4 has been very positive and that’s been great,” he said. “It’s one of those things where I wasn’t exactly sure how it would go. This wasn’t something where there was a slow evolution where I’m like, ‘OK I’m going to tiptoe around and do this and you know eventually I’m going to jump on in.’—no I jumped in immediately and told them what I wanted to do and the reception has been incredible. I think that’s why I’ve been able to come here and talk about this and branch out and work on the website and things like that.”
Shea may have said it best. “We on the analytics side have to go out and meet them. We really have to understand what they’re doing and sometimes I think we believe we’re doing it more than we are.”
Catch Arif every week on The Andy Luke & Arif Football Machine
