Numbers game intrigues athletes
Analytics making more inroads in basketball and football, too
Michael Lewis is explaining why baseball players are dumb.
It is the opening session of the MIT Sloan Sports Analytics Conference in Boston, the ninth annual gathering of the statistics-minded affectionately known as Dorkapalooza, and Lewis, the author of Moneyball, is on a bit of a jag.
“The old saying is you can’t be too stupid to play baseball,” he says.
He explains that when he was researching Moneyball, he spoke with Oakland A’s general manager Billy Beane about whether the front office was trying to keep the players in the loop about what the analytics team was unearthing.
“He said, ‘We never talk to them, it just confuses them.’ ”
Lewis said he was kind of surprised to discover, when he started going into basketball and football locker rooms, that there were athletes there who were intellectually curious.
They were far fewer in baseball, he said. Not that that was necessarily a problem, given what the sport often entailed: “You have to shut your mind down and stand in the outfield.” Another panellist put it this way: “Empty head, full bat.”
The irony is that baseball, as a game of individual events, is the one sport where analytics data is the easiest to apply. It’s the one where buy-in at the player and coach level should be absolute. And, yet: “When you have an argument between a front office using analytics and a manager (who is more old-school), the manager is almost always wrong,” Lewis says.
“But if you have a sport like basketball” — where the data is less clean and the understanding of it is more in its infancy — “then I’d put my money, maybe, on the coach.”
One of those intellectually curious players was Shane Battier, the 13-year NBA veteran who was on the same panel with Lewis. He’s the poster child — literally, he was on a Sloan poster once — for a player whose study of advanced stats made him far more effective than his traditional numbers would suggest.
Daryl Morey, his former GM with the Houston Rockets and the founder of the Sloan event, said Battier was “sort of like Luke Skywalker. It’s your destiny to be here.”
Battier explained that in his first few years in the NBA, they would receive scouting reports that said things like “Carmelo Anthony is good from the left block. Watch out for him there.” Gee, thanks, Battier said.
But when he was traded to the stats-focused Rockets, Battier suddenly had reams of data on all sorts of opponent tendencies. He learned that Anthony, for example, was 10 percentage points less efficient if you forced him to go to his left instead of his right. “Why wouldn’t I take those 10 points right there?” he said.
Morey pointed to the final panellist, Jeff Van Gundy, his former head coach with the Rockets, to provide an illustration of why stats and tendencies aren’t always the only answer.
“We had numbers saying that Dikembe (Mutombo) was better than Yao (Ming),” Morey said. “But I said to Jeff, ‘Well, we’re going to play Yao.’ He said, ‘well, you either believe this s--- or you don’t.’”
Van Gundy nodded in agreement. The point was, none of these numbers exist in a vacuum. Context matters. (Though Morey also noted that when Yao was hurt, the Rockets went on a long winning streak with Mutombo at centre.) Van Gundy also had the story of the day.
He referenced Sacramento Kings owner Vivek Ranadive, who infamously claimed he could win youth girls basketball based on systems, not talent. “I coached fourth-grade girls basketball this year, and it’s all about players,” Van Gundy said. “So I call bull---t on that.”
Winning, he said, relies on two things: “Can they make a layup, and will all your girls show up? If all 10 show up, can you convince two of them they are really sick?” Play your good players more, he said.
“That’s how you win.” Ruthless, but effective.
In addition to the fancy panels, the Sloan conference includes research papers from academic types. One of them, by a Harvard PhD candidate named Stephen Pettigrew, sought to develop a metric — Added Goal Value — that determined how valuable a hockey player’s goals were based on how much they affected his team’s win probability at the time they were scored.
The essence is that goals scored early in the game, or when one team had a big lead, are less valuable than goals scored late or when the game is close. In theory, it’s the elusive clutch metric, and Pettigrew says that over several seasons of data, there was a reasonable correlation from year to year.
Two economics professors presented a study that looked at the PGA Tour’s extensive ShotLink database and measured the impact of pressure on making putts.
They found that there was a correlation between the amount of prize money at stake on the final putt of a tournament and the likelihood of a player sinking it. For every $50,000 of value a putt increased, the chance of making it dropped one percentage point. The effect, they said, was most dramatic in the tricky five-to-10-foot range, where the one-point drop was observed for every $20,000 value a putt increased.
Football, one of the Sloan panellists noted, is the toughest nut to crack analytically because there are so many players on the field and so much of a player’s performance is dependent on what other players do to assist him.
But one area where it is at the fore is in wearable technology: every NFL player in 2014 wore a chip — in games and practice — that measured his movement, allowing teams to gather data instantly on speed and distance covered. New Orleans head coach Sean Payton, and retired players Marshall Faulk and Matt Birk, were a little skeptical of its in-game applications — “if you score a touchdown, I don’t care about your speed on the play,” Payton said — but all agreed that it should have huge value in determining wear and tear on players.