Analytics-driven Sixers ride the numbers to NBA playoffs
Before each game, the Sixers receive a "deck" compiled by the analytics staff that breaks down how to exploit the opponent. Marcus Hayes got a glimpse at the deep-dive into the numbers that has fueled the franchise.
Trust the Process, if you like.
The Sixers trust the numbers.
They won 17 consecutive games before Monday's Game 2 loss, the last nine of those wins without all-star center Joel Embiid. They also won their 15th straight without Dario Saric and their 16th straight without JJ Redick. Their 17th straight win was the first playoff victory since 2012.
It took talent, chemistry and fine coaching.
It also took 10 math whizzes and computer programmers, most of them with PhDs.
When the Sixers faced the Heat to start the playoffs Saturday, Brett Brown's analytics team had broken down 1.28 billion lines of data, and that was just from this season's games. It generated a 10-page game dossier, or "deck," that not only highlighted the Heat's offensive and defensive tendencies but also projected which Sixers lineup would fare best against every Heat lineup; which Sixers pairs would fare best against each Heat pair; and which Sixers fared best, offensively and defensively, against their likely opposite on the Heat — such as, say, Ersan Ilyasova vs. Kelly Olynyk.
I got a look at a typical "deck" before the Sixers visited the Hawks last week. It's the same deck that head coach Brett Brown received before he left the Wells Fargo Center after a Sunday afternoon win over the Mavericks. He gets the next game's deck as soon as the last game ends.
The detail was jaw-dropping.
These team-by-team bibles are the core of what the Sixers get for their annual $3 million investment in analytics (about what former Sixers guard Michael Carter-Williams earned this season). It's a bargain, considering the return: 52 wins, 24 more than last season. And last season: 28 wins, 18 more than the previous year.
Certainly, adding Joel Embiid and Ben Simmons turned the Sixers around more than anything, but did deep-dive analytics really make a big difference?
"No doubt," Brown told me Friday.
Brown likes all of the trivia and minutia — who shoots best, and when, and where — but he relies most on the grades his analysts assign five-man player groups, or rotations. The Sixers have not had a stable roster, so they haven't had any stable rotations. They've undergone three major overhauls since the end of last season, when The Process began to pay off.
This season they converted Simmons, a forward, to point guard, and added Redick as a shooter. In February they signed snipers Ilyasova and Marco Belinelli. Lately, they had to subtract Embiid as he recovers from facial fractures and they also replaced backup point guard T.J. McConnell with rookie Markelle Fultz — the No. 1 overall pick who missed 68 games with a shoulder injury.
Brown stayed faithful to his rotation analytics through it all. The numbers dictate who plays with whom.
"One of the greatest assistance factors that I have is using the analytics to rotate a team," Brown said. "You can talk about how, say, the Hawks make more threes in the corners, or how Joel likes to go to his left shoulder on a jump-hook. But for me it's: 'Which groups play well together? Which don't? How does the other team sub? Historically, how can we exploit that matchup when Olynyk's at a 5?' That's the area I feel has helped us more than anything."
The analytics team demonstrated to Brown not only how his top three rotations rank among the league's top 10, but that his No. 1 rotation — Embiid, Simmons, Robert Covington, Dario Saric and Redick — flirts with the top overall spot. Since the Sixers signed Ilyasova and Belinelli, only two of the top 15 rotations project negatively.
If it sounds involved, it is — and it isn't. Brown and his staff and the commitment to analytics have been the only constants since The Process began in 2013, so he's learned to speak the language, and he's learned to ignore the noise.
"It's up to me not to be drowned by over-information," Brown said. "I'm at a stage where I see it clearly. Man, it's huge for me."
That has been true regardless of regime.
One of the more glaring misperceptions about these 76ers is that, once visionary general manager Sam Hinkie quit in 2016, the team diminished its use of analytics. The opposite it true. New GM Bryan Colangelo didn't just double down. He tripled down.
From Sabermetrics and Moneyball to the legendary Philly Special, analytics have influenced sports for almost two decades. Gabe Kapler and Matt Klentak have analytic-ed the Phillies into one of baseball's best turnaround stories this year. The Flyers this season touted their new analytics and sports science programs, and they wound up in the playoffs.
That fourth-down TD pass from Trey Burton to Nick Foles in Super Bowl LII? Well, the "Philly Special" never happens without Doug Pederson's faith in fourth-down predictors.
But no Philly franchise has an operation like the Sixers'. Every NBA team has an analytics department, but, according to one league source, most departments consist of a couple of recent college grads with bachelor's degrees in statistics. Only two — Houston and Boston — can compare with the Sixers' input and integration. (Maybe Oklahoma City, too, but no one knows; the Thunder are paranoiacally secretive.)
"I think there are other teams that use analytics as much or more than the 76ers, but their department's size is on the larger side," said one Eastern Conference scout.
The department is led by Alex Rucker, the vice president of analytics and strategy: a big, friendly former high school basketball player from Vancouver who holds degrees in business, marketing, law and education. He's the man who figures out what's most important. But that's the easy part.
"Maybe 20 percent of my job is analytics," said Rucker, 42. "Maybe 80 percent is being able to collaborate with people and communicate with the coaches."
He has degrees in four disciplines — and he's the dumb one.
His chief lieutenant, Sergi Oliva, is a tall, striking Catalan with a lush black beard and a PhD in computational complexity. He's the only holdover from Hinkie's three-man staff. A former player and coach himself, Oliva whispers in Brown's ear more than Brown's wife does. This week, he might be reminding Brown about Hassan Whiteside's preferred post-up position, or Goran Dragic's likelihood of scoring at the basket when he drives left, from the right side.
Oliva attends all of the coaches' meetings, all of the practices and all of the games, just like assistants Billy Lange, Lloyd Pierce and Jim O'Brien.
"Sergi — he's no different than Billy Lange," Brown said. "My analytics guy is with me all the time. I think that's rare, if not unique, in this league."
Ivana Seric is a 6-foot-2 Croatian forward from NJIT who also happens to hold a PhD in computational fluid dynamics. She sometimes briefs the coaches, too.
There are three Harvard PhDs who do research.
In 2013, the NBA installed high-definition cameras in the rafters of every arena to track player movements by jersey numbers (this season they switched from SportVU to Second Spectrum). Michael Lai, poached from IBM, wrote the Sixers' proprietary programs that synthesize that wealth of information culled from the league.
Grant Fiddyment is a red-headed brainiac, literally; his PhD in computational neuroscience helped him track energy in people's brains as they had seizures. That skill translates nicely when he's trying to decipher all those richly overlaid tracks of player movement.
Software engineer Danilo Orlando spends hours every day cleaning up the NBA's messy play-by-play data, so that Lai's programs can read it.
As a finishing touch, Rucker last year added user-interface expert Tica Lin, a sunny computer artist whose main job is to make the decks dynamic, colorful and personable. She generates both print and the online editions, which players can access on their phones.
Thanks to Lin, every player, friend or foe, gets his own color portrait next to his numbers.
"We hired her, specifically, because how you present something has a huge impact on how well-received it is and how your brain is activated. It's crucial," Rucker said. "I mean, I would never think to include the guy's picture."
Building the machine
This was Colangelo's vision when he replaced Hinkie in 2016, and it began when he stole Rucker from the Raptors that same year.
"When I was brought on, I set out to be considered a best-in-class organization from top to bottom," Colangelo said. "I think by hiring Alex and allowing him to make recommendations on building out our analytics team, we are well on our way there in this space."
Colangelo's a Cornell man and analytics is rife with Ivy Leaguers, so it's no surprise he was an early adopter. In fact, this was the second time he hired Rucker.
In 2010, when Colangelo was the Raptors GM and Jay Triano — a Vancouver guy — was the coach, Rucker, still in the Navy, met Colangelo and convinced him that the time was ripe to invest in data. Colangelo hired Rucker as a consultant, then brought him on full-time when his enlistment ended. Triano was done as coach in 2011. Colangelo stepped down in 2013. Rucker outlasted them both.
His star continued to rise, as did the Raptors' win totals. Colangelo had to make a lot of promises to steal Rucker: He gave Rucker a big promotion, an important title and the freedom to hire and expand the Sixers' analytics staff with top-flight post-grads who might otherwise land at Apple, or Google, or Uber.
All of that took money.
The Sixers had plenty.
"We have an ownership group that is happy to fund things," Rucker said. "It's willing to break from traditional thinking about NBA front-office pay scales. Since I got here, there's been an increase in the analytics budget of 500 percent."
If there's anything Ivy Leaguers love, it's playing the numbers.
"We firmly believe in the value of data and analytics to enhance and inform our decision-making processes," said Sixers owner Josh Harris, a graduate of Penn's Wharton School of Business and Harvard's MBA program. "We have seen the positive impact it has had for our entire basketball operation, from the coaching staff to our players and front office."
It can only impact as far as the principles believe.
‘A more truthful place’
"Numbers can only tell you so much, but they can really tell you how to guard certain teams," Simmons said. "We know what other teams are good at and what they are not good at. We usually play to that. People sometimes think we're stupid, but they don't know what we know."
To be fair, some of the information and actions are counterintuitive. For instance: when you play a team with deadly corner three-point shooters, it's smarter to stay on your man in the corner and not help when an opponent drives down the lane, and that can look foolish for the unwashed.
Analytics also can convince a player to change. And it has.
Ideally in the Sixers' half-court offense a wing player will hold onto the ball for about 0.5 seconds then pass, shoot or dribble. Earlier this season, the analytics team noticed that one wing player was holding it 1.9 seconds, which is about as long as Embiid holds it while waiting for a double-team. The wing was being a "ball-stopper." He denied it, and indignantly — until he saw the numbers
"He was stunned," Rucker said.
There was good news, too. The same player was an infrequent offensive rebounder, but when he did crash the glass he was the second-best perimeter rebounder in the entire NBA. Now, the wing moves the ball more quickly and crashes more often.
The numbers help improve some granular, team-wide issues, Redick said.
The analytics crew discovered the Sixers were woefully inefficient during the first six seconds of each possession and made suggestions as to how to improve; Rucker indicated the improvements centered on getting up court faster and initiating the offense quicker. At first, the players were ambivalent but they did what they were told.
"They broke it down on film for two months, then came to us and showed us: 'Guys, here's how you're making these incremental improvements,' " Redick said. "Incorporating these advanced stats with player development coaching — it's actually really cool, if you really care about this game."
Nobody cares more than rookies. They have the most to gain.
Simmons missed his rookie season with a foot injury. Fultz missed 68 games this season. As point guards, both were expected to struggle running the team and defending the position. As they rehabbed, both spent time with Brown and Lange watching detailed film that Rucker and his troops had prepared.
"They'd be, like, 'Side pick-and-roll — these are the reads you've got to see. Gotta watch the weak-side defender. Did he step in the paint? Then the corner's open. If he's under-helping, you can drive and expect no help,'" Rucker said.
The video sessions helped Simmons and Fultz start their careers ahead of the rookie curve.
The information is even more complete when applied to veterans, and often more striking, too.
Richaun Holmes' athletic defensive style seems to make him a more logical replacement for Embiid than ground-bound Amir Johnson. However, analytics show that Holmes' energy often works against him and that Johnson's space-conscious defensive habits are more effective. They also predict which teammates work best with Holmes.
"Richaun is a good initial defender, but once he goes to contest a shot or to help, there are some breakdowns," Rucker explained. "So, he plays best with Robert Covington and Simmons, who are defenders with long wing spans who play well in space and can cover for him."
Occasionally, the numbers and the recommended strategies don't make sense even to the experts.
"Sometimes it'll refute what I see in the reports," Brown said. "And then, sometimes analytics will steer me to a different place I hadn't thought of before. Sometimes I'll disagree."
Disagreement is the entire point.
"I'm wrong all the time. We all are," Rucker said. "The numbers — they don't solve everything. They just bring us to a more truthful place."
Numbers in action
The Sixers, undermanned and weary, were ahead by nine points with 90 seconds to go in Atlanta because they had played as the analytics demanded. At all costs, they denied corner three-pointers, especially from the left side. The Hawks were 1 for 5.
Covington cheated toward Taurean Prince in the middle, who quickly passed to John Collins in the corner. Collins nailed a three. A few seconds later, Prince inbounded the ball on the sideline and dived to the corner. Neither Redick nor Simmons followed quickly enough. Prince hit a three and cut the lead to 115-100 with 42.7 seconds left. The Sixers held on, but it's almost as though the numbers taunted their lapses.
Other relevant percentages played out as well. Ilyasova projected to have a huge advantage against a certain Hawks defender; sure enough, Ilyasova tied his season high with 26 points. Simmons and Covington looked good on paper against Prince, who shot 9 for 25, or 36 percent, 6 percent worse than usual.
It was like watching basketball by script.
It was mathematics come to life.
The Sixers' system, which is synced to its video cache, lets them put a number on everything. In the last five years, Rucker said, the information available has increased by "an order of magnitude," which is a smart guy's way of saying there's 10 times as much today as when The Process began.
"I can ask, 'Let me see every pick-and-roll in the NBA this season,' and it'll spit out about 30,000 of them," Rucker said. "I can show you every pick-and-roll LeBron's run in the last three games. Or the last 30 games. If you're concerned about Cleveland's five-out [no post player] offense when LeBron gets the ball — and you should be, because they're good at it — you can say 'I want to see those clips.' "
The analytics will tell you that Hawks swingman DeAndre Bembry, the former St. Joseph's star, surrenders much worse than the league average 1.1 points-per-possession when he defends the pick-and-roll: 1.27 PPP in 132 chances before the Sixers game in the season's final week.
They will tell you that the Hawks are even better shooting threes in transition, and that they create a lot of turnovers when you're in transition, so be careful when you run.
They will also tell you that LeBron is the third-least active player on defense in the entire NBA, which, somehow is a good thing. It seems that LeBron's defensive spacing is so exquisite that he's almost always in the right spot to deter a pass or a drive.
This might explain why, earlier this month, Brown — a 57-year-old analytics devotee who watched Michael Jordan and Kobe Bryant in their primes and who grew up on the careers of Wilt Chamberlain and Oscar Robertson — said of LeBron: "He is the best player to have ever played our sport."
Analytics explains why LeBron is so good in so many areas, and not all of them obvious. Analytics help show other players how they might aspire to his level.
Rucker admits that analytics essentially tell you what most good basketball people know intuitively.
"If we didn't have this, the coaches would get most of it, anyway," Rucker said. "It's just getting all of it."
Brett Brown wants all of it.
"At this stage, for me, I've learned how to form a marriage with gut-feel and math," Brown said. "We make our decisions with the numbers. Not by the numbers."
Either way, after years of futility, they're finally making a lot of correct decisions.