Student creates new way to rank NBA coaches

Win-loss records and championship rings usually define the best NBA coaches, but Carolina student Shane Faberman looked at different data to see which coaches get the most out of their players.

Faberman, who has a double-major in data science as well as statistics and operations research, created a metric called Box Plus Minus Over Expected, which he abbreviates to BOE.

Box Plus Minus is a widely used basketball statistic that estimates a player’s on-court performance using traditional box score stats like points, rebounds, assists and steals. Faberman’s twist was to calculate how much a player over- or under-performed relative to expectations — and to attribute that difference partly to the coach.

“I’ve seen analysis on NFL coaching decisions, but I haven’t seen as much on NBA coaches. Coaching in basketball is hard to quantify. It’s not easy to separate a coach’s influence from the talent of the players,” he said.

Constructing the “expected” BPM wasn’t straightforward. “I went through 10 to 15 versions,” said Faberman, who earned an analyst spot in Carolina’s Sports Analysis Intelligence Laboratory.

His formula considers a player’s previous BPM, age-based improvement curves and performance trends. “I couldn’t just multiply last year’s BPM by a percent change,” said the Philadelphia native. “Negative and positive values don’t work well together mathematically. I had to get creative.”

Faberman built a “scraper” tool in R, a programming language, and captured NBA statistics for seasons 2015-16 to 2024-25 from the public website basketballreference.com. He compiled the data into a format he could analyze.

He presented his findings at Carolina’s 2025 Celebration of Undergraduate Research. Using nearly 40 reference sources, Faberman peppered his research paper with analyses of players such as JaKarr Sampson, Malik Beasley and Luka Dončić and coaches such as Steve Kerr, Tyronn Lue, Jason Kidd and Chris Finch.

Ime Udoka of the Houston Rockets was among the most influential coaches with a +0.668 BPM Over Expected. On the opposite end was Brian Keefe of the Washington Wizards, who scored – 0.616, although he was excluded from the final graphs due to limited data. (2024 was his first season.)

Joe Mazzulla of the Boston Celtics was an interesting case. Despite his team claiming the 2024 NBA title and 61 wins in 2025, Mazzulla had a low BOE of -0.182. Faberman thinks that may be because he inherited a roster stacked with elite shooters. “It’s hard to tell if success is because of his coaching schemes or because of the talent,” he said.

Miami Heat coach Erik Spoelstra stood out with a +0.177 BOE. “Players who joined Spoelstra’s team tended to outperform expectations in their first year,” Faberman said. “After leaving, they underperformed. That suggests his system elevates players.” Spoelstra appeared in a quadrant of Faberman’s graph that indicates a coach’s strong positive influence before and after a player’s departure.

Faberman looked at unique data to see which coaches get the most out of their players. (submitted photo; Graphic by Gillie Sibrian/UNC-Chapel Hill)

Surprisingly, many recent championship-winning coaches did not score high. “Some of the coaches who people think of as elite weren’t near the top in my metric. That suggests talent and roster construction may matter more than coaching when it comes to winning titles,” Faberman said.

“Shane’s methodology stands out through his novel creation of the BOE metric. It’s an innovative solution,” said adviser Kendall Thomas, a graduate teaching fellow in the statistics and operations research department.

Faberman doesn’t claim to identify the NBA’s best coach but offers a data-driven glimpse into who is making a measurable difference. “This new metric is not gospel,” Faberman said. “It’s a useful way to see which coaches really help players get better — and which ones might just be along for the ride.”