Regrets? Big league GMs have a few
Biggest bias in many front offices is the fear of letting a great player get away, conference told
The slogan on the banners at this year’s MIT Sloan Sports Analytics Conference was “Elevate the Game,” but Daryl Morey said something they might want to consider for future posters: “Humans are really, really bad at making decisions.” Catchy!
Morey, the general manager of the Houston Rockets who co-founded the annual gathering of sports and spreadsheets more than a decade ago when he was a professor at MIT, was on a panel discussing cognitive bias. But he hit on something that underpins the analytic movement in all sports: it’s about finding better methods to make decisions.
The downstream impact of that expresses itself in different ways in different sports — baseball clubs place greater value now on hitters who get on base a lot and hockey teams want players who drive puck possession — but at the root of the data revolution is that very simple idea: People make bad choices all the time. Information should help save them from themselves.
Billy Beane, the executive vice-president of the Oakland Athletics, was at Sloan this weekend for the first time, despite being one of the spiritual leaders of the stats movement. He told his assembled disciples that in the 2002 MLB draft — which would form the basis for much of the book Moneyball — he and then-assistant Paul DePodesta essentially decided they wanted nothing to do with the human element of decision-making. People can overvalue the impact of seeing one good play in person, or one bad one, and can be swayed by a pleasant personality or even a winning smile.
“We wanted to go totally objective,” Beane said. “We wanted to go scorched earth, just to see what would happen.”
Sam Hinkie, the former GM of the Philadelphia 76ers, sitting across from Beane, responded: “I wouldn’t recommend that.” Everyone laughed. That Oakland draft ended up with some successes and some failures — a typical baseball crapshoot — but the underlying concept of putting more faith in data to guide decisions has endured, partly because putting faith in humans to make calls often doesn’t work out so well.
Teams almost always place a higher value on their own players, even subconsciously. It’s also human nature to be a little gun-shy when considering a move that includes obvious risk, and no front office wants the guy they just traded to become a star elsewhere.
Morey called this tendency “loss aversion” and Farhan Zaidi, a former Beane assistant who is now general manager of the Los Angeles Dodgers, used the term “avoidance of regret.”
The idea is that executives will be reluctant to make some deals, even if there is clear evidence for it, because they don’t want to end up feeling bad about it in the future.
Zaidi said his personnel team in Los Angeles was once considering a trade, but there was “a lot of consternation about giving up this particular player.” He suggested a way of thinking about it differently that was somewhat unconventional. Instead of losing the player to another team, “I said, ‘What if we just take him out back and shoot him?’” Zaidi said.
This made the deal feel better: the Dodgers would get assets, and they didn’t have to think about what might happen with their former asset elsewhere.
“A winning strategy,” Morey said. Everyone laughed.
“People were much more careful around me after that,” Zaidi said, smiling.
The point, though, was about overcoming natural biases. Baseball’s CBA, not to mention the laws of society, does not permit trades of prospects for murder victims, but it’s a way to compartmentalize that thing — the potential for regret — that could inhibit a useful trade.
How many deals every year fall victim to such thinking? “I used to say that we could solve the issue by never getting a good player,” Morey said. “Then we’d never have to worry about losing him.”
The challenge is that a GM who is ruthlessly objective still has to deal with all kinds of people who could have those same biases — fans, players, coaches, owners, media. Hinkie oversaw three seasons of atrocious Sixers teams as he tried to maximize the chances of finding a superstar at the top of the draft, but eventually ownership lost patience with the plan and he quit last year.
Hinkie said the problem with trying to be completely objective is that people will still always have to make choices. The data cannot actually decide. You can try to keep your hands off the knobs of the machine, he said, “but the truth is your hands are on the knobs all the time because the data sucks.”
But did Hinkie have bad data, or did he just make poor decisions with it? His problem didn’t seem to be the years of tanking — good idea, that — but the resulting roster full of big men in a league that is moving away from them.
Ultimately, it will still be humans making decisions. Data and statistics and information will help, and more and more it seems like the sports industry is finding peace with the idea. But it won’t be perfect.
“You’re in the fishbowl,” Morey said, “not knowing if there’s water.”