Revamped analytics team helps Warriors steer new course
Warriors head coach Steve Kerr has long resisted embracing the NBA’s most popular play, the pickandroll, for a simple reason: He believes the best offenses are unpredictable, and worries that a steady drumbeat of high screens would be easy for defenses to dissect.
But after watching Golden State lose seven of eight games as rookie center James Wiseman struggled to find open looks, Kerr was desperate for something — anything — to help his team. Pabail Sidhu, the Warriors’ director of analytics and innovation, showed Kerr numbers reinforcing that Wiseman was at his best in
pickandrolls with guard Stephen Curry.
In Tuesday night’s 122121 win over the Bucks at Chase Center, Kerr ran a seasonhightying 71 pickandrolls. Instead of getting overwhelmed by the nuances of the Warriors’ readandreact system, Wiseman (13 points, 10 rebounds) kept things simple, setting high screens for Curry before he ran to the rim or popped out for midrange jumpers.
This was a significant concession for Kerr, who had spent much of the previous 31⁄2 months making only tweaks to his movementheavy offense despite the fact that it called for a passing big man, and Wiseman is more of a lob threat. Sidhu’s numbers convinced Kerr that, even though a heavy dose of pickandrolls might not be as aesthetically pleasing as Kerr’s motion system, it was what the Warriors needed.
Over the final 21 games, Kerr will continue to rely on the team’s analytics department as he experiments with lineup combinations and playing styles. This is the first time in Kerr’s seven years with the Warriors that he has not had a firm grasp on their identity nearly threefourths of the way through the season. As he put it Tuesday, “the puzzle has been much trickier to put together” because of injuries, coronavirusrelated absences and roster issues.
It’s a good thing for Golden State, then, that the Warriors’ analytics team has tripled in size since the start of last season. After manager of basketball analytics Sammy Gelfand left for a similar role with the Pistons in July 2018, Sidhu — a former Microsoft and University of Washington employee — was a oneman analytics department for more than a year. The Warriors have since hired data scientist Matt Holbrook and data engineer Kevin Steele, and they hope to add another person or two in the coming months.
“We don’t think analytics are the beall and endall, but we see them as a valuable tool for everything that we do,” said Kirk Lacob, Golden State’s executive vice president of basketball operations. “I think, when done properly, analytics can kind of be like having a really good 13th or 14th man on the roster. Is it going to be the difference between winning a title and missing the playoffs? No, but it sure helps.”
This is all part of a longterm plan that Lacob began to hatch more than a decade ago. When he first joined the Warriors in 2010, the analyticsbased model that the A’s helped popularize in baseball was just starting to gain traction in the NBA.
In addition to their typical daily responsibilities, Lacob and basketballoperations coordinator Pat Sund fielded coaches’ analytics questions. Gelfand, who joined Golden State in 2012 after being hired to work with its G League affiliate, soon became the goto staffer when someone wanted to know about the trajectory of certain shots or the number of corner 3pointers opponents were taking.
But given that the Warriors didn’t have an inhouse analytics infrastructure, they had to outsource much of their research to data firms. In 2017, after being hired away from Washington’s athletic department, Sidhu began to look into building a complex database that would serve as a onestop shop for any numbers query.
Four years later, the Warriors have the foundation of a
stateoftheart system that allows them to find deep analytics on everything from opponents to college teams to international leagues. This infrastructure enables Sidhu and his team to provide scouts, frontoffice executives and coaches nuanced reports on prospects, trade targets and matchups.
Each morning after a win or loss, Kerr reads a detailed rundown from Sidhu on the previous night’s game. The information on trends, passing numbers and types of defenses informs how Kerr approaches practices and meetings.
“We’re constantly asking Pabail and his staff for numbers as things come up throughout the year,” Kerr said. “It might be lineup combinations that we’re playing. It could be play types, what our efficiency ratings are depending on what plays we run, that sort of thing. It’s a big part of the game. It’s a big part of what we do.”
Never have analytics been more important to the Warriors than this season. Before Kerr replaced Wiseman in the starting lineup in late January with Kevon Looney, he looked over reports suggesting that
the starting lineup would be much better defensively with Looney at center. Two months later, after seeing data about the importance of letting young big men play through mistakes, Kerr moved Wiseman back into the first unit.
Many of Kerr’s most vital decisions require an innate feel that goes beyond X’s and O’s. But in a season clouded by a pandemic and a flawed roster, the analytics department has offered Kerr some clarity.
“In the NBA, margins matter a lot,” Lacob said. “The difference between an unbelievably great offense and a middling offense is not a ton when you really look at it. But a few points per 100 possessions are huge in an NBA game. That could be the difference between six wins in the regular season.
“That’s kind of what we’re trying to do. We’re just trying to put everyone in the best position, give them the best information to give them the best edge.”