Data helped Michigan win its NCAA basketball title. Human communications and feel for the game were the real heroes.
The Humanist’s Keyboard goes courtside
In season one of Running Point on Netflix, Kate Hudson plays a party girl turned president of her father’s pro basketball team, the fictional Los Angeles Waves. In a scene reminiscent of Brad Pitt in Moneyball, Hudson brokers a complicated three-way trade based on data analytics.
It got me thinking. My blog is about the symbiosis of human and digital analytics. Human, plus machine. In this post, I speak with Jacob Kohn, director of data analytics for the University of Michigan men’s basketball team, which won the NCAA national championship this year.
I’ve known Jacob since he was a middle schooler who played with my son on the Evergreen Tree Frogs basketball team. Needless to say, he’s climbed the ladder of American sports management, and I wouldn’t be surprised if his name becomes as well known as those in Michael Lewis’ Moneyball, which was published when Jacob was just 6 years old.
Moneyball is subtitled, “the art of winning an unfair game.” It explores the brilliance of statistician Bill James and general manager Billy Beane in building a winning franchise with the Athletics, then in Oakland, with a constrained budget. Their approach was to exploit inefficiencies in the game by leveraging little advantages that add up to victories.
The University of Michigan surprised many as an underdog, ending a 37-year NCAA championship drought.
Jacob and I spoke last week via videoconference in the midst of the NBA semi-finals. Sitting in a cluttered “Go Blue” conference room on campus, we chatted about Oklahoma City’s “Wemby Problem” and what this 7-foot-3-inch, high IQ center, would mean for the future of basketball. A few nights earlier, Wemby sunk a deep three pointer after a shooting guard passed to him from near the paint, to help bury the NBA’s best basketball team. It was an example of exploiting an inefficiency in basketball – taking the 3-pointer instead of a less efficient mid-2.
Greg
You report directly to Head Coach Dusty May. Tell me your duties at Michigan.
Jacob
When I took the job, I asked Dusty, what do you want me to do? And he said, I actually don’t care what you do. Just do whatever you think will help us win. So during the season, I write scouting reports on upcoming opponents. I’ll write post-game reports and I’ll do self-evaluation reports throughout the season. I’ll write notes on what’s going well, what’s not going well, what trends are happening. If there are things that are going really well, but we’re not doing them enough, I point out we should try this more. I also help with recruiting.
Greg
Those self-assessments are interesting to me because it’s human-led data rather than data leading humans. What’s interesting to me is you’re not just presenting
data, you’re giving a human self-assessment of the team based on data alongside what you’re seeing.
Jacob
Yeah, so I try to stay on the more objective data side. I think the most important part of my job is the communication piece. Can I present it cleanly? You can have a great idea when you see something in the data, but if you can’t communicate it effectively, then there’s no use in even having the data in the first place. It’s useless. The coaches appreciate analytics, but none of them have super strong statistical backgrounds. So if I just throw out a number with no context and no explanation, it’s not going to really do anything. It’s not going to drive our process and what we choose to do.
If there’s something I’m seeing, I might see it and then go to the data or I might look at the data and then watch film to see if the data are accurate? Could it be misleading because of the way it’s measured? There’s definitely a human spin, but I try to stay more objective because I’m not really adding anything since the coaching staff has hundreds of years of basketball coaching and playing experience. I don’t.
Greg
Yeah, that’s really fascinating and well communicated (laughter). Did you read Moneyball?
Jacob
Yes, I’ve read Moneyball and I’ve watched the movie. Working in sports was never on my radar until probably years after I watched the movie and read the book in college. I didn’t really think about it that much. I didn’t really go down the rabbit hole.
What really kicked it off for me was I took a data science class at Duke. One of the projects was to pick a data set from a list of 25 and do some analysis. I chose a baseball data set. I could have just half-assed it and been done in 45 minutes. But I ended up going really deep into it. I looked at contract years for players. I found other data sources and linked them. And that’s where it started. I was really interested. The following summer, my sophomore spring I did a Duke program in Serbia to teach English for refugees, which I was really bad at. I was basically teaching for an hour, three days a week. I had all this free time. So I went back to the baseball data and started to build my own models and predictions.
Greg
So how did you translate this curiosity into your career?
Jacob
I interned at Big League Advantage (BLA), a sports analytics firm and was assigned to Duke University basketball during the final years of Coach K. In Coach K’s final year, we made the final four and my boss at BLA went to the final four where he met Nate Oats at the Final Four, the head coach at Alabama. And Nate Oats who coached at Alabama. He was the most analytics-driven coach in college basketball. They met at a fortuitous time because this was 2022, basically the beginning of the NIL transfer portal era.
A big part of coaching became evaluating all these college players with data and then being able to basically sign them as free agents. BLA really took off as a result. We helped Alabama build their team through the transfer portal the following year, which is when Alabama made their first Final Four in program history.
Greg
And this was the gateway to Michigan?
Jacob
Right, the following summer was when Dusty got hired at Michigan, and he and Nate are friends. BLA became part of the conversation and then BLA got in at Michigan. I supported Michigan for a year before Dusty asked me to come work for him directly.
Greg
Can you give us an example of data analytics being used well in recruiting?
Jacob
Yeah, so you have to really consider, what a player is being asked to do. And then you have to think about when they’re on your team, what are they going to be asked to do? Alabama recruited this guy who from Cal State Fullerton, Latrell Wrightsell. He was a really good three-point shooter, but his “shot diet” was a really bad. He was taking a lot of like really difficult mid-range shots. We asked, what was he doing well and what could he improve on? He shot 38 percent from beyond the three-point line. But he was taking a lot of tough mid-twos, so his effective field goal percentage
was 51 percent. We looked at that and projected him as a really good three-point shooter who was taking all these bad shots. At Alabama they decided they could coach him out of taking these bad shots. They coached him on this and his effective field goal percentage jumped from 51 to 61 percent the following year.
Similarly, when we bring someone to Michigan, the goal is to find guys who do specific things well and use them for that as much as possible. Like with Elliot Cadeau last year, he was at Carolina. He was an incredible passer, just extremely talented. He came to Michigan and we’re like, okay, you’re gonna have a bunch of guys around you who you’re gonna make them better.
Greg
We talked the other night a little about Dean Oliver, author of Basketball on Paper. He produced the old Journal of Basketball Studies. Is he the Bill James of basketball?
Jacob
I think so.
Greg
Are basketball data analytics as robust as baseball analytics? I suppose some of the answer to that question are functions of data availability or a function of algorithms and that sort of thing, but are they on par?
Jacob
I think basketball is behind. I think it’s getting there, but I think baseball just as a sport is easier because the events are, you know, there’s just very defined events, like a pitch or, you know, an at bat. Whereas basketball is just so much more free-flowing, it’s harder to measure, I think.
Greg
Leagues are now putting sensors on everything to measure and calibrate everything imaginable. Is that going to happen in basketball?
Jacob
Yeah, in every NBA arena, they have this sort of tracking system called Hawkeye, which is super proprietary. I don’t have access. I haven’t worked in the NBA, but basically it takes where every player is on the court and maps 29 different body points on each player at a rate of 60 frames per second. There’s only 30 arenas in the NBA, far fewer than the 365 we have in Division 1 NCAA.
Greg
You understand AI and technology. Are we headed toward having enough sensors and cameras that a coach won’t really need a data analytics person? The coach will just ask an AI questions?
Jacob
I don’t think so. I think it’ll automate to make a lot of things easier. But I just don’t know if I really see that happening.
Greg
Really?
Jacob
I don’t know if I really see that because I think at the end of the day, they’re going to want the data to be interpreted by a trusted person or team of people.
Greg Shaw
Right, there is that sports phrase, “that’s why we play the game.” What’s behind the phrase is that it is after all a human game. Sure, data can inform you but there are a lot of unknowable and unpredictable human factors that must play out in any given game.
Jacob
I think I think there’s a lot of feel involved with coaching basketball, and I think that is something that like won’t be replaced, probably ever.
Greg
If we go back to one of your first points about the importance of communications, the data says pull the starting pitcher. And so the pitching coach goes out to talk to the pitcher. Does communications become more important to both the player and coach? Does a pitcher find a way to better communicate: “Hey, I know you’re looking at the data, and it says this, but let me tell you how I’m feeling.”
Jacob
Depends. A rookie player or a rookie coach might struggle. But a seasoned player and manager won’t. Max Scherzer was pitching in the playoffs last year and effectively communicated, “nope, you’re not taking me out.” That’s a feel thing. But if you trust the player, then you might go against the data. I think data is meant to be used as a tool, not as the end all, be all of decision making.
Greg Shaw
I asked an AI agent to tell me the story of Michigan’s 2026 championship. It says instead of chaos and improvisation, 2026 was about modern roster building, dominance, and finishing the job. Do you agree with that?
Jacob
If you want to win a championship, you have to have a roster that’s good enough to win a championship. More than that, you need to have the right people. I think, you know, Yax (Yaxel Lendeborg), who was the best player on our team last year, was the number one transfer in the country. He probably could have stayed in the draft last year and been a first-round pick. But he transferred to Michigan. There’s probably a lot of basketball players of his pedigree who could have come in and, you know, been a jerk. “You know, I’m the number one transfer. I’m going to take 20 shots a game. I’m going to come here and get mine and then I’m going to go to the NBA next year. But it wasn’t like that at all. He was incredibly unselfish. He was a great teammate. He was really liked and respected by everyone.
Greg Shaw
The AI notes go on to say that when Michigan last won in 1989, it was really Glenn Rice who was the superstar. He went nuclear or something the report said. So this past year, Yax becomes the Big Ten Player of the Year, but then you had Elliott Cadeau, who was the primary playmaker.
Jacob
I think this is the most important part of what we had last year. I mean, you can measure it in assists.