By Steve Shea (@SteveShea33)
June 30, 2017
Often, professional sports organizations employ the “go-for” model for incorporating analytics. Picture a group of executives seated around a large table deciding what to do for lunch. They may toss around a couple options—Italian or Chinese—before bickering on whether they should order individual entrees or share crowd-sized portions. Eventually, pen goes to paper and an order is scratched out.
Then, the young intern is called in to make the run, to execute the portion of the operation where the he has some authority. He will have to decide whether to pick up the food or have it delivered. If it’s pick-up, he’ll have to find and decide on a parking space. At the restaurant, he may make the call to toss in a few more soy sauce packets. And should the restaurant not be able to fill a portion of the order, he may have to make the split-second decision on a replacement item, provided it’s an appetizer or dessert and not something central to the meal.
In the sports world, the owners, the president of [pick your sport] operations, the general manager, assistant general managers, and coaches are the decision makers.
When they gather to hash out the basics of an offseason plan, the group might be unanimous that the team needs an upgrade at point guard, but argue over the means to make the acquisition. After some discussion, it may be decided that free agency has the most appealing and feasible options, and a short list of candidates could be assembled.
Then the analytics are brought in. The front office or coaching staff would like evaluations on the candidates, specifically focusing on their playmaking ability. The analytics staff will have some freedoms in the analysis. For example, they might decide to split pick-and-roll situations depending on if the screening teammate rolled or popped, or they might look at the point guard’s turnover rates split by whether or not the opposition switched.
The analytics will help the decision makers zero-in on a first, second and third choice among the group. They’ll help the general manager and his team understand which available option is the best match for what the group wants.
But maybe, analytics should have been part of that group. Perhaps, analytics should have been at the table before the short list of free-agent options was assembled and before free-agency was decided as the most appealing path. If in the room, analytics might have suggested that the group’s desire for a ball-dominant point guard is outdated, that the team should instead try to build a lineup that shares the playmaking. And that if anyone should be the primary creator, it’s a forward already on the roster. Analytics may have argued that the team should be giving as much consideration to any potential addition’s off-the-ball value as they do his on-the-ball skills. Or, analytics may have suggested that if the team moves away from the traditional point guard mold, they can find a player that is better defensively and with the length to switch onto bigger wings on the perimeter.
What should be the role of analytics?
Should analytics simply run the post-up efficiency of the bigs the team is considering for the upcoming draft, or should analytics have the forum to suggest that the team shouldn’t be so interested in posting up on offense and should instead judge centers on their ability roll to the rim, pass, knock down a perimeter shot, and switch screens on defense?
Should analytics stick to tracking the team’s paint touches and ball reversals and their influence on the offense’s efficiency, or should they be able to question the coach’s lineups, suggest swapping the rebounding power forward for another small forward, and push for a 1-in and 4-out formation?
The analytics
Ever notice how a journalist or TV commentator will reference “the analytics,” as in “the analytics say James Harden should win MVP.” The analytics don’t say Harden should be MVP because he has a higher true-shooting percentage. The analytics don’t say Rudy Gobert is the NBA’s best defender because he has the best defensive real plus-minus. The analytics don’t say anything, because a quantitative analysis isn’t the equivalent of punching an addition problem into a calculator.
Teams fill analytics positions based on the applicant’s degrees, the programming languages she or he knows, and to some extent, the individual’s fluency in the sport. But job ads don’t often ask for a demonstration of quantitative creativity. They might pry for problem-solving skills, but the kind that have answers in SQL code.
Sports mirror many industries in their increased reliance on data to make decisions. In this new world where numbers have power, everyone wants to be comfortable with statistics. But I’ve seen comfortable with statistics, and it’s not what anyone should be aiming to achieve.
Mathematics is taught as black and white. 2+2 is always 4. A teacher asks her students to journal about their dream day and expects a variety of responses, but every math exam comes with a fixed answer key. We teach that math is either right or wrong.
So, when a crowd gathers for a PowerPoint presentation with a few pie charts and vague references to statistical significance, we all nod in approval as if the sky opened before us and the speaker’s conclusions descended from the heavens.
Stats are facts, but every situation can be analyzed in numerous ways, and each analysis has a multitude of interpretations. We should get as cozy with one particular approach as we would a bed of skunks and porcupines.
In the go-for model, analytics are a simple stat run that leaves little for the decision-makers to interpret. It’s one approach with strict parameters. When analytics are in the room, they are not just the numeric answers to a specific query but a perspective when shaping the questions. It’s not just providing the team’s offensive rebounding rates, but it’s posing the question of if they should be considered in conjunction with the transition points surrendered. And any assortment of statistical information on that front leads to a complex discussion of alternate interpretations. There are analytics to support aggressive play on the glass, and there are stats that argue for the team’s need to get more bodies back to stop transition.
If teams are looking for a lackey to calculate a team’s efficiency on drives, then they are searching for the analytics, because there is only one correct answer. But should they be asking for more?
Illumination or just support?
At a 2012 MIT Sloan Sports Analytics Conference panel, longtime hockey executive Brian Burke said, “Statistics are like a lamp post to a drunk, useful for support but not for illumination.”
There’s a day-to-day component of analytics—for example, tracking and aggregating shots in practice, or assembling and communicating pre and post-game reports—that support the organization in its activities.
But Leicester City didn’t win the Barclay’s Premier League with pretty shot charts. They won because they dramatically altered their philosophy on how to build a team and implemented an innovative strategy on the field (built largely on an understanding of transition).
Teams need daily maintenance, but the supporting basketball activities described above don’t move the needle like finding Draymond Green in the 2nd round.
Can analytics be so illuminating? Recognizing the value of positional versatility and Draymond Green’s potential in that NBA goes beyond calculating college players’ efficiency by play type. It requires an understanding of the game’s trends and in that context, an intelligent interpretation of data. It requires not just the ability to crunch numbers, but the creativity to pose original questions.
But isn’t this precisely where analytics should be strong? At their core, analytics are a different perspective. As much as many quantitative analysts are fans of the game, numbers eventually beat the fan bias out of the observer. Analytics are fresh eyes, activity that very much lives outside of the traditional box in which sports organizations operated. If a team is looking for new ideas, true innovation that can distinguish them among their competitors, could there be a more fertile land than analytics?
Some NBA organizations are already onboard with this approach. Houston is the glaringly obvious example. But to what extent was analytics involved in Phil Jackson’s decisions in New York? And when we move to less progressive leagues, like the NHL, we find decision-makers that would more willingly welcome a plague than conversations on per possession efficiency.
Certain organizations still view analytics as glorified stat trackers, like the kind you’d see behind the bench charting shots at a high school basketball game. I’d argue that the true value in analytics is only found when teams recognize them as a fundamental mode of thinking in their own right, a unique approach to solving sports problems.
Analytics deserve a seat at the table, not to provide all the answers, but to question conventional thinking.