2017 Sloan Sports Conference: Day 2

Photo courtesy of the MIT Sloan Sports Analytics Conference
Photo courtesy of the MIT Sloan Sports Analytics Conference /

The 2017 MIT Sloan Sports Analytics Conference is being held in Boston Friday and Saturday. Several Nylon Calculus contributors are there covering the event and we’ll be sharing our thoughts and reactions to panels and research papers. Check back throughout the day for updates. Also, catch up on Day 1, along with Q&As with Nate Duncan, Neil Paine, and Chris Herring.

Data-driven storytelling

Communication has been the backdrop of the Sloan Conference for roughly as long as it has existed. How do we communicate meaningful insights to players and coaches? At the Data Driven Storytelling panel, the focus was on how we can use data to tell stories.

Basketball wasn’t explicitly represented on the panel — it featured Bill Barnwell, Cris Collinsworth, Brian Kenny, Nate Silver, and Chad Millman — but almost everything discussed was applicable to what we’re doing here at Nylon Calculus. Collinsworth did a nice job of summing up the goal, saying, “Nobody wants to just hear the data. For me, it’s, can we take the helmet off the guy?”

The panel did a nice job of circling through the different types of data — narrative, predictive, evaluative — and spent some time circling the chaos that can surround the increasing prevalence of predictive data in the sphere of sports coverage.

Silver talked about how statistics and statistical models are interacting with history, and it’s always an open question on how well that historical context will work as a predictive base for the evolving future. Kenny also pointed out that internal struggle — acknowledging the inherent possibility for error can makes for a crappy column.

“We’re at a different point in the ability to gauge the effectiveness of people speaking as experts,” said Barnwall. “You have a significant level of accountability that you didn’t have 30 years ago.”

In the end, everyone seemed to agree that variety was the key. Consumers of sports media content are not a uniform block and offering variety is the best way to serve everyone’s needs, tell better stories, and offer the richest set of experiences and opportunities when interacting with sports stories.

538’s Chris Herring spoke to this exact point in my Q&A with with him yesterday:

"It can’t be one thing. You can watch a lot of basketball but if you’re not talking to guys, you might not know why unless you hear someone explain it like that sideline reporter did. If you’re at the games but you’re not talking to players you’re not going to know why. And if you’re just looking at the numbers and you’re not watching the film to back that part of it up, it could be off-base for a number of reasons. Or just irrelevant. There’s a lot of stuff out there that’s just noise. You have to blend at least two of the three, and ideally all three if it’s really worth investigating."

When in doubt offer more, cast a wide net. Let people find what works for them.

“That’s one of the advantages of technology, you can now experience the same game in different ways,” said Silver.

— Ian Levy, @HickoryHigh

Cuban speaks

One of the feature panels today was Mark Cuban talking one-on-one with Nate Silver — Shark V. Fox: Politics and Forecasting in the Time of the Hedgehog. The topic was ostensibly political but wandered in several different directions, including to basketball.

Cuban shared an interesting tidbit about how the Mavericks had noticed a potential advantage and prepared for the 2016 NBA Draft.

“We now have like five guys that were undrafted this year. But when we went in, because the Celtics and the Hawks had so many picks, in just two teams, they couldn’t keep everyone. So for us, it was more of a 70-player draft than it was a 60-player draft.”

The implication is, I believe, that because the Hawks and Celtics might not be able to keep all those picks on their rosters they were more likely to reach for foreign draft-and-stash players (which turned out to be true in the case of the Celtics and Ante Zizic and Guerschon Yabusele), leaving behind some NBA-ready contributors who might fall completely out of the draft. It’s a nice window into how logic and rational thought, could manifest an analytic advantage divorced from actual statistics.

The topic of referees came up and Cuban, of course, had plenty to say. He shared his thought that the one tangential benefit of the Last Two-Minute reports is that it has inspired more fan trust in the game.

As to whether it had done anything to improve officiating, let’s just say Cuban was…aggressively skeptical.

— Ian Levy, @HickoryHigh

What lies ahead for basketball analytics

In “Ball Don’t Lie: The Future of Basketball Analytics,” Zach Lowe got right down to the point, asking Dean Oliver why the latest player-tracking research matters when we’ve already learned, for example, how the Houston Rockets generate corner 3-pointers. The esteemed author of Basketball on Paper promptly responded: “Defense.” The implication is that, up to this point, we’ve leveraged box scores, play-by-play logs, Synergy, SportVU and other data to measure player and team performance. Yet there’s an entire dimension of basketball that remains largely unsolved, in part because it’s driven by positioning and coordination. If granular spatial and motion data are available, perhaps analysts can advance our defensive understanding.

Lowe shifted the conversation to sports science and psychology, wondering about the value of the mental, interpersonal and emotional sides of the game. Luis Scola shared a story about how players are sometimes subjected to tests that attempt to measure them. In light of the high-stakes nature of the assessments, he said, players are ultimately incentivized to perform well on them. Educators describe this dynamic as “teaching to the test”: the metrics end up capturing the player’s test-taking preparation rather than the desired skills themselves. Oliver emphasized the gravity of this issue, so he advises teams to vet data providers rigorously.

The panel also discussed wearable devices, how they are codified in the collective bargaining agreement and how teams might act on key findings from them. Sue Bird drew a connection to shootarounds. She considered them useless and, given all that we know about the importance of rest (especially on the road), she suggested that forward-thinking franchises should eliminate them. In general, her insight shed light on how our takeaways from biometric data should prompt teams to reengineer their processes and, for lack of a better term, tweak “workplace issues” for optimal performance.

Lastly, Lowe asked the panel if there are reliable ways to evaluate coaching performance. Oliver said that we can arrive at an estimate by determining how players improve or stagnate under certain coaches, but such a metric would not be able to tell us how they contribute to the improvement or stagnation. Is the effectiveness (or lack thereof) a function of their teaching, communication, tactical or other skills? The two streams of data would have to be married, and even then the result would be an “indirect” measurement at best.

Admirably, the panel covered as much ground as an hour-long program can possibly allow. Their insights prompted me to reflect on other tangential questions that might be the subject of future SSAC proceedings. Most notably, how might teams evolve in the way they structure their analytics operations?

A few weeks ago, the Miami Heat hired Shane Battier as Director of Basketball Development and Analytics, bringing an interesting type of expertise to the front office. As a former player, Battier was an early adopter and power-user of data-driven insights. He had to think about interpreting the numbers from a practitioner’s perspective, determining what’s relevant and figuring out how to implement them. This background makes him uniquely positioned to bridge the gap between analysts and basketball staff, and perhaps we’ll see such appointments more frequently as the strategic landscape in the NBA changes.

— Positive Residual, @presidual

Catching up

Most of the basketball-specific programs at the conference were front-loaded yesterday so things should be a little lighter today (catch up on yesterday’s panels and papers here). The basketball analytics panel runs a little later on and should be as bombastic as usual, featuring Dean Oliver, Sue Bird, Mike Zarren, moderated by Zach Lowe. I am also looking forward to the panel on Data-Driven Storytelling which will include Nate Silver, Chad Millman, Chris Collinsworth, Brian Kenny, and Bill Barnwall. Should be of particular relevance to what we’re trying to do here at Nylon Calculus.

In general, the conference has been extremely strong this year on the basketball side. The research papers were particularly interesting and the presentations involving NBA decision-making (both Evan Wasch’s presentation and Adam Silver talking with Nate Silver) were a comforting peak at how thoroughly the league vets and analyzes the possible ramifications of even their smallest decisions.

Wish you could all be here with us.

— Ian Levy, @HickoryHigh