Boston has two kinds of people: nerds and sports fans. It just took them a while to realize how much that Venn diagram overlapped.
Daryl Morey has always had a foot in both camps. In 2006, he was teaching a class on data-driven decision-making at the Massachusetts Institute of Technology’s Sloan Business School when he got hired to run the NBA’s Houston Rockets.
Instead of letting the class die, Morey and his co-teacher Jessica Gelman turned it into the Sloan Sports Analytics Conference — which, depending on how you look at it, is either the coolest stats meet-up or the dorkiest sports conference in the world. Either way, it’s a big deal.
This weekend, thousands of current and aspiring industry insiders, students, gamblers, journalists and curious fans will descend on Boston for an event the sports journalist Bill Simmons affectionately nicknamed Dorkapalooza, where they’ll learn about the latest in research and data-driven practices across most major sports (plus NASCAR, chess and pickleball).
“Sloan, probably more than any other conference out there, has been the No 1 career launching pad for people in the sports analytics industry,” said Luke Bornn. He ought to know — he was a Harvard stats professor without much of a sports background when he co-authored his first Sloan research paper in 2014. Now he helps run AC Milan and co-owns a Ligue 1 club with Billy Beane, the analytical baseball executive played by Brad Pitt in Moneyball.
Football was slow to establish a presence at Sloan but lately, it’s one of the hottest tickets in town, thanks partly to a run of prize-winning research papers from, among others, Bornn and William Spearman, Liverpool FC’s newly promoted director of research.
Compared to baseball or even basketball, football is “a very complicated game” for data analysis, said Morey. “You can make 10 great passes and still get a zero because the goalie makes a great save. But you can’t have that be the same as when you just punt the ball from 40 yards out and hope it goes in. Those are both zeroes, but they’re very different zeroes.”
Daryl Morey and Jessica Gelman in 2020 (Photo: Blake Nissen for The Boston Globe via Getty Images)
The key to good football analytics is looking beyond goals, which are frustratingly rare and random, to probabilistic outcomes like expected goals. Bornn’s work used full-pitch tracking data to assign goal probabilities to all kinds of actions on or off the ball, the way he and his students had already done in basketball.
It’s space-age stuff that most clubs are still nowhere near doing — nor should they be, Bornn says. “For a team to extract full insights from tracking data, it’s just way too expensive and time-consuming. They’d be way better off starting with much more simple event data.”
One of the pioneers of football analytics, Charles Reep, started collecting event data by hand in 1950 and eventually used it to convince higher-ups at clubs and the FA to stop screwing around with possession and try to build scoring moves of “not more than three passes”. These days, a lot of people believe he helped set English football back decades.
Reep’s work is often presented as a cautionary tale about drawing bad conclusions from good data, but Morey thinks modern goal probability models might actually support something like Reep’s ideas. “If I ran a soccer club,” he said at a 2019 Sloan panel, “they would play launch-and-squish.” Launch the ball directly toward the other team’s goal, then squish the field to win it back with compact and maximally chaotic high pressing. It’s enough to make Pep Guardiola cry.
Rejecting conventional wisdom in favour of what the data tells him has made Morey an icon in basketball, where he helped lead a movement away from Michael Jordan-style jump shots toward more efficient three-pointers, dunks and free throws.
“I see a lot of bad analysis, like, ‘The top teams who win do this, so therefore it’s good’,” he said. “Well, we know that’s wrong from every sport.”
Bornn isn’t totally sold on Morey’s launch-and-squish revolution. “This is an old idea, right? Big Sam’s POMOs — what was it, positions of maximal opportunity?” he said, referencing the direct style of former Bolton Wanderers manager Sam Allardyce. Still, he does see signs of football clubs adopting a sort of probabilistic thinking that can override a pretty passing game.
“You occasionally see kick-offs where they kick it into the far corner, out of bounds. And then, all of a sudden, that’s the standard way to kick off,” he said. “That’s essentially people agreeing that having the ball at the centre line is worse than the opponent having the ball in their far corner.”
Allardyce at Bolton – POMO FOMO (Photo: Laurence Griffiths/Getty Images)
It’s hard to use data for clear-cut answers about how to play better football. “Depending on what you factor in, you might have different analyses that lead to, ‘Oh, this is actually a good idea’, or ‘It’s a bad idea’,” said Bornn. “Whereas the three-point change in basketball, there’s no one with a functioning calculator who would tell you, ‘No, we should keep taking long twos’.”
But analytics is a big field that covers a lot more than tactics and so do Sloan’s wide-ranging panels and research papers, which span topics from player psychology to sports business. For clubs, the most important use of data is in player recruitment. Morey and Bornn both want to see more work on player development — using data to improve how players scan the pitch, for example, or strike the ball.
Sloan’s most important function is the same as any conference: bringing people together.
Howard Hamilton is an introverted academic type with a PhD in aerospace engineering (“I get a lot of mileage out of the rocket scientist thing,” he jokes). When he started one of the first football analytics blogs, Soccermetrics, in 2009, he knew he would have to look farther than the local pub to find like-minded fans. He found them at Sloan.
For the past dozen years, Hamilton has helped organise an annual football meet-up on the eve of the conference. It’s a favourite of industry legends such as Sarah Rudd, who has been attending since before she became the head of analytics at Arsenal (a post she recently left to found a football analytics consultancy called src|ftbl).
Football will be well represented at this weekend’s big event. Rudd is appearing on a panel with Mladen Sormaz, who was Leicester City’s first head of analytics, and 21st Group’s Omar Chaudhuri. (Disclosure: I’ll be moderating.) A study of counter-attacks by a couple of United States Soccer Federation data scientists has reached the final round of the prestigious research paper competition, continuing football’s impressive showing in recent years. There will be other panels, research posters and networking galore.
More than anything, though, the weekend is a chance for fans who watch sports on a spreadsheet to enjoy each other’s company. Hamilton stopped attending the actual conference a few years ago, due in part to rising costs — Sloan has drawn criticism over ticket prices, which can run over $1,000 (£834) — but he still travels 1,000 miles every year for the yearly gathering of the football nerds.
“I look forward to going up to Boston the first weekend of March — it’s been part of my life for the last 12 years,” he said. “Maybe at some point, I’ll hand it over to someone else. But if I stopped attending, there would be a hole in my life.”
(Top photo: Damian Strohmeyer /Sports Illustrated via Getty Images)