“I’m not a person that frowns upon analytics,” ex-NBA player Jalen Rose said this weekend during a panel at the MIT Sloan Sports Analytics Conference in Boston. “I do think it’s a tool, but not the toolbox.”
The same could be said about the conference itself. While Sloan began as a hub for deep thinking and data-driven findings about sports, the gathering, now in its 12th year, is more than a rock concert for NBA-analytics messiahs Daryl Morey and Sam Hinkie. There are dissenting voices, like Rose and others, who rounded out panels and provided different perspectives about everything from the role of statistics to business to the human element in it all. It’s also become a networking opportunity, where students and job seekers from around the world come to shake hands and peddle résumés with the hopes of getting hired by a publication or a sports organization.
Here are the most interesting takeaways from the weekend, along with some scuttlebutt making the rounds in the hallways of the Boston Convention and Exhibition Center.
The Trouble With Public Stats
During a panel called “Take That for Data: Basketball Analytics,” Nick Wright of Fox Sports asked Rockets general manager Morey and Celtics assistant general manager Mike Zarren about the strength of a single-number advanced statistic such as real plus-minus (RPM), which currently ranks Tyus Jones and Fred VanVleet as top-14 players. Zarren said that if a stat is at odds with what you’re watching, then either the way the number is calculated or the way you’re watching the game is probably wrong. Morey pointed out that although a statistic like RPM can pick up on how some players provide a role when they’re on the floor that helps the team win, “that player could be very replaceable by multiple players with that same skill set. … Even though it’s correct that they’re creating that winning, in our roles of having to decide player to player, you have to think about how else can you fill that role.” In other words, data can’t be the be-all and end-all.
I spent the past week or so chatting with NBA executives and coaches about the role of stats, and the one thing they agree on is that they’re occasionally misused by the public. One front-office stats guy told me that he observes people making mistakes most frequently with NBA.com’s defensive rating, which is a team statistic, not a player statistic. For example, you might see the following sentence somewhere on the internet: VanVleet leads the Raptors with a 97.8 defensive rating. But that’s not true. The correct way to convey the statistic is: The Raptors have a 97.8 defensive rating when VanVleet is on the floor. There’s a subtle but important difference between the two sentences. The former implies that VanVleet is an elite defender, while the latter says VanVleet is one of five players who make up a team performing at an elite level. Fans and writers alike made this error last season when analyzing Jae Crowder, who ranked 20th in real plus-minus and shined in virtually all other advanced statistics. The reality is Crowder was a beneficiary of the Celtics’ system and the superior defensive players he frequently shared the floor with.
The problem is there aren’t many other options, and NBA executives disagree on better practices for using defensive stats. A few execs told me they like using on/off-court data to see how different units affect shot distribution and quality (like how many midrange jumpers they force or using tracking data to see what percentage of shots are contested). But a handful of other executives suggested that the complexities of on/off data are vastly underappreciated. The situation, the referees, the players on the floor, and luck all play a significant role in any number that a machine spits out.
None of the Blazers in the play above can be negatively graded for their defense against Kobe Bryant in those two possessions. They were in perfect position; Kobe just did something amazing each time. Multiple league executives say that a better way of analyzing defense is to focus on the process rather than the results—looking at whether rotations were on time or late, or the way in which a player navigated a screen to chase a shooter, or whether a player was in the right position to help. Even so, as one executive told me, it’s hard enough to get tracking cameras to accurately log actions as “simple” as a high pick-and-roll, never mind more nuanced actions like help rotations.
One executive said that while the public defensive data is bad, he also wasn’t sure how valuable his team’s proprietary tracking data is. If NBA personnel feel this way about their private numbers, how much stock should we put into what we have access to? Still, another exec said the available data can also help fans watch the game with more depth. Something is better than nothing.
Intriguing Trade-Reform Ideas
Morey said he thinks the fundamental challenge with trades is that draft picks are the only way to bridge the gap between how two teams value one player. “They’re like cigarettes in prison. That’s the only currency you have,” Morey said Saturday at the “Take That for Data” panel. “The value changes up and down all the time, and it makes for a not-very-liquid market.” Morey said that the league should allow teams to put player-performance conditions on draft picks during trades, so that a pick’s ultimate position could be affected by the success or health of a traded player. But there’s been resistance to the idea from the league office, Zarren said, since it would become difficult to keep track of all the conditions.
Would it really be that hard? Every single draft-pick exchange is tracked online, so the league’s well-staffed office should be capable of adding it to their list. Teams are already allowed to protect draft picks based on team performance; for instance, the first-rounder the Cavaliers sent to the Lakers this month is top-three protected. Morey poked fun at Zarren and the Celtics, saying no one wants to trade with them after the 2013 Brooklyn blockbuster. But had the Nets been able to attach conditions to the first-round picks they sent to the Celtics—that is, based solely on the health of Paul Pierce and Kevin Garnett rather than their own team’s performance—perhaps they wouldn’t be digging their way out from a basketball dungeon.
Zarren later floated an idea that would allow a percentage of a team’s lottery ping-pong balls to be tradable, which would create another form of currency. The union would have to be involved for such a drastic change, but it’s another intriguing idea. The trade deadline, the draft, and free agency are three of the most exciting times on the NBA calendar. Allowing more creativity could lead to even more trades that shake up the league and drive more excitement for the game.
Jimmy Butler’s Injury Opens Up Contract Questions
Butler collapsed to the floor and clutched his right knee in the third quarter of the Timberwolves’ loss to the Rockets on Friday night. Yahoo Sports’ Shams Charania reported that Butler suffered a torn meniscus and is expected to miss four to six weeks. Butler underwent surgery, and the Wolves maintained that their All-Star swingman is out indefinitely.
An NBA executive explained to me that the Yahoo Sports timeline suggests that the torn part of the meniscus will be “shaved” via arthroscopic surgery, which often leads to a faster recovery, but can lead to pain or complications down the line. The alternative was for Butler to undergo a full repair, which involves sewing the tear together like you might mend ripped jeans. That route would have produced fewer potential complications, but also would have sidelined him the rest of the season.
No one knows how Butler, 28, will look after returning from the injury, but NBA executives I chatted with are already curious about how both he and the Timberwolves will handle his contract situation. Butler will make $20.4 million in 2018-19 and has a $19.8 million player option for 2019-20. If he declines his option and signed an extension this summer, Butler’s deal would start at “only” $24.5 million in 2019-20 but provide immediate long-term security. If he waits until next summer, he could earn 30 percent of the 2019-20 salary cap—or roughly $32.4 million (based on a projected salary cap of $108 million). Long-term security is nice, so that’s not a bad deal for a player who has missed a lot of games with injuries in the past and whose aggressive style doesn’t exactly bode well for a long, sustainable career.
Butler might prefer to just decline his player option for 2019-20, hit free agency with the ability to earn more money, and review his options then. It might not even make sense for the Timberwolves to try to extend Butler now. Not only does the injury cloud Butler’s future, but their roster is about to become very expensive. They signed Andrew Wiggins to a maximum extension in October (which he has so far failed to live up to). Gorgui Dieng is overpaid through the 2020-21 season, and so is Jeff Teague through 2019-20. Karl-Anthony Towns will look for the max when he’s a restricted free agent in 2019. Butler is an absolute monster two-way player who is pivotal to Minnesota’s success, but availability is the best ability. The Timberwolves need to figure out how often they’ll get pre-injury Butler going forward before inking him to a big-money extension.
Bhostgusters!
The coolest research paper at Sloan this year is titled “Bhostgusters: Realtime Interactive Play Sketching With Synthesized NBA Defenses.” In the paper, a five-person team from the Technical University of Munich aims to approach “play sketching from a data-driven perspective.” In layman’s terms, you can draw up a play on a touch screen using their system, then simulate how a defense might respond to it. Seriously. We’re basically living in an NBA 2K sim.
If this system makes it to NBA benches, Gregg Popovich might draw plays on an iPad instead of a whiteboard—and then see how the opponent might counter, all with the push of a button.