A Whole New Ball Game

MIT professor Peko Hosoi ’92 explores the underlying complexity of sports

Brett Tomlinson
By Brett Tomlinson

Published March 7, 2019

11 min read

A baseball is not that complicated, until you consider what’s inside: a rubber-encased cork “pill,” coated in adhesive, tightly wrapped by hundreds of yards of yarn, and snugly outfitted with two interlocking panels of leather, stitched together by hand. 

Alan Nathan *75, a leading baseball physicist and professor emeritus at the University of Illinois, has visited the Rawlings factory in Costa Rica that supplies Major League Baseball’s official ball. “One of the things that sort of amazed me is how uniform the balls actually are,” he says. “The various properties of the ball that you can measure really don’t change very much from one ball to another.”

Last year, Nathan chaired a committee convened by Major League Baseball’s commissioner to explore why the rate of home runs had been steadily increasing over a stretch of two and a half seasons, starting after the 2015 All-Star Game. In 2017, 4.8 percent of batted balls cleared the fence — an all-time high that was 35 percent higher than the home-run rate in the first half of the 2015 season. The committee ruled out two popular explanations: changes to the springiness of the ball, and the effects of hitters attempting to launch the ball at a higher angle. They then turned their attention to the ball’s flight. 

Enter MIT mechanical engineering professor Peko Hosoi’92, a physicist and fluid-dynamics specialist who has found a new niche in recent years as one of the country’s most prominent researchers of sports science. (It’s pronounced PECK-oh; her given name is Anette, but she’s been Peko since childhood, when her grandmother said she looked like the smiling girl on the wrapper of a popular Japanese candy of the same name.) 

Hosoi, one of 10 members of the committee, studied the trajectories of balls hit in major-league games, extracting each one’s drag coefficient, a measure of the resistance that an object encounters in flight. She identified a small but significant decrease in drag that was enough to explain the change in the rate of home runs.

A home run, Hosoi explains, is a “non-linear event.” Either the ball goes over the fence, or it doesn’t. A long home run isn’t any more of a home run than one that clears the wall by half an inch. 

“Because you have this sharp cutoff, and because a lot of the balls are close to home runs, just changing [the distance traveled] by a little sends a lot more balls over the fence,” Hosoi says.

What changed in the leather-covered, tightly wound ball of yarn? The cause is apparently very subtle because the committee could not identify it, even after analyzing game-used balls in detail through a series of lab experiments.

“Our conclusion — that it’s the carry of the ball that has changed — was a rather satisfying conclusion,” Nathan says. “It was, I have to say, unsatisfying not to understand exactly why there was a change in the aerodynamic properties of the ball. But a lot of physics is this way: You figure out something, and then it raises more questions. That’s sort of what keeps you going.” 

A bicycle is not that complicated, until you take it on a chairlift to the top of a mountain, strap on a helmet, put your feet on the pedals, and aim the front tire downhill. 

Hosoi started mountain biking as a high-school student in Corvallis, Ore., and when she and her husband first visited Highland Mountain Bike Park in New Hampshire, she gained a new appreciation for how important the right bike could be.

Zipping down the mountainside jump lines was exhilarating, but her cross-country bike wasn’t up to the challenge. “I think I went over the handlebars eight times that day,” Hosoi says with a laugh, “and even doing that, I thought, this is the most fun that a human being could possibly have.”

Hosoi, an avid downhill mountain biker, had her MIT students study the mechanics of different designs of mountain bikes in her introductory mechanics course.

Photo: John Freidah/MIT

At the time, Hosoi was making her mark at MIT in fluid dynamics and robotics. But after riding the trails at Highland, all she could think about were mountain bikes. 

“MIT’s motto is mens et manus,” she says. “Mens is mind, and manus is hands. ... The hands part is the experience: You have to experience what’s going on. And I experienced going over the handlebars multiple times. That told me that my center of mass was in the wrong place.”

The downhill mountain bikes on the market had a range of divergent components, with variations in everything from the positioning of the shocks to the design of the brakes. Hosoi set to work figuring out the mechanics of the different designs. And since she was teaching an introductory mechanics course, she got her students involved as well. “I think that semester, every exam question was a bicycle question,” she jokes. “And by the end of the semester, I bought a bike.” (Though her choice was informed by analysis, it was ultimately made based on feel — she purchased the same model she’d been renting.)

Bicycles proved to be an intuitive example for the students to explore, which made Hosoi wonder if sports could provide other useful engineering problems, hidden in plain sight. She began asking her colleagues and discovered that “half the faculty are these closet sports fanatics who are doing these things in their garages on the weekends but not bringing them to the students.”

“You have to experience what’s going on. And I experienced going over the handlebars multiple times. That told me that my center of mass was in the wrong place.”

— Peko Hosoi ’92

Determined to find a way to bring her love of sports to the classroom, Hosoi co-founded what’s now called the MIT Sports Lab. She also connected with Christina Chase, a former competitive cyclist and MIT’s first entrepreneur-in-residence, to develop a course, Sports Technology: Engineering and Innovation, in which teams of students work with sports professionals to research real-world questions. In the fall semester, projects included developing an app for the U.S. Olympic Committee to give long jumpers real-time feedback on key metrics such as launch angle and launch velocity; modeling optimal dribbling dynamics for the Spanish soccer club F.C. Barcelona; and testing how different ski-goggle lenses handle “flat-light” conditions, when it’s difficult to see variations in snowy terrain, for the goggle manufacturer Shred. 

The course brings together students from engineering, business, architecture, and the sciences. They tend to be comfortable with the technical aspects of their work, Hosoi says, but integrating a series of technical findings into a final product presents more challenges. At their final presentations last December, the teams presented their projects to their “clients” via videoconference, with a combination of video and PowerPoint, distilling three months of research into about 15 minutes.

Relatively few of the students will go on to work in sports, but Chase says the questions students are studying have a wider reach. The “quantified athlete” may lead to the “quantified self,” bringing improvements in fitness and health care, and innovations in sports venues could have applications in urban planning or architecture. “There are a variety of areas where sports is a really fun starting place for a longer-tail, global impact,” Chase says. 

A running shoe is not that complicated, until you take a closer look at the racks of your local running shop and start reading the tags. What exactly is a dual-density midsole? Or a unidirectional carbon-fiber plate? The race to be lighter and bouncier, while maintaining stability, is big business: One Nike model, the Zoom Vaporfly 4%, claims to give marathoners a 4 percent efficiency edge over other top racing shoes. A pair retails for $250, but according to The New York Times, the shoes fetch $400 and more on eBay.

Adidas, a partner for one research project in Hosoi’s lab, produces a shoe with a 3D-printed sole. “They can design every strut in that mesostructure to do whatever they want it to do,” Hosoi says. 

But what do the designers want the sole to do? How do you find the optimal design that, for example, minimizes the energy that a runner uses with each stride? There are remarkable computer models from the field of biomechanics that simulate muscles, tendons, and bones, but those models are complex and slow, Hosoi says. 

To create a simpler, faster model of a runner’s leg, Hosoi’s lab is looking to computer models developed for robotics. They may not be as perfect as the biomechanical options, but if you can get reasonably close, she says, you can search different iterations more efficiently. “Then the optimization problem becomes feasible,” she adds, “and you can get to a design that maybe shoe designers have never thought of before.” 

READ MORE Data-Based Coaching: Sports Science at the Princeton Boathouse

Robotics is familiar territory for Hosoi, but her path at Princeton began not in engineering but in physics, which introduced her to fluid dynamics. She remembers sitting in a Fine Library carrel, diligently working through the equations in L.D. Landau and E.M. Lifshitz’s Fluid Mechanics — the type of textbook that will say something like “it can easily be shown ...” and then launch into eight pages of algebraic equations. 

“It wasn’t that there was an assignment or that somebody was telling me I had to do this,” Hosoi says. “I was going to understand everything in that book. ... When students go from undergrad to grad, a lot of them go through that switch, where you develop some additional agency, and that was what flipped the switch for me.” 

The experience, paired with a helpful suggestion from her thesis adviser, Raymond Goldstein, propelled Hosoi to Ph.D. studies at the University of Chicago, followed by her first faculty appointment, at Harvey Mudd College in California, and a year later, a post at MIT.

Her specialty within fluid dynamics was thin-film flows, the behavior of liquids on a surface. But as she began to meet with MIT graduate students, she realized nearly everyone wanted to build robots. “And so I thought, ‘If I’m going to survive here, I’m going to have to learn how to build a robot,’” she says. 

Leveraging her knowledge of thin films, she worked with a student to develop a robotic snail — “nature’s all-terrain vehicle” — using a substance similar to hair gel to replicate the slime that snails use to crawl across a surface. More bio-inspired robotics followed, with projects that emulated the pumping mechanism of plants and the digging skills of a razor clam. 

Seeing what her students could achieve in robotics made Hosoi “recalibrate” what she believed was possible. “If I had not come to MIT, my trajectory would have been completely different,” she says. 

A glove is not that complicated, until you spread your fingers wide, reach out, and try to grasp a spiraling football firmly enough to hold on as your body crashes against the shoulder of a charging defender. 

When The New York Times wrote about the tacky silicone gloves that have become a staple for NFL wide receivers, reporter David Waldstein posed a question to Hosoi: How much of a difference do those gloves make, versus catching the ball the old-fashioned way, with your bare hands? Sarah Fay, a Ph.D. student who works with the MIT Sports Lab, studied the respective coefficients of friction for hand and glove and found a significant difference: In dry conditions, the gloves are 20 percent stickier than a human hand. In the Times story, Hosoi explained that the soft, deformable silicone adheres to variations on the surface of the ball. “Every time you get more deformable, you get a better adhesion,” she told Waldstein. 

Sports fans seem to have a growing appetite for the science behind the games they watch, and pro teams now compete for talent in their analytics departments in the same way that they compete for free-agent players. There is no shortage of questions for Hosoi and her colleagues to explore. But Hosoi’s time is a more finite resource, especially since she became an associate dean of engineering in 2017. Working in administration, she says, allows her to see the broader world of innovation at MIT, outside of mechanical engineering, and that has been interesting and rewarding. 

At the same time, the MIT Sports Lab has blossomed. Hosoi continues to team up with Chase on the sports course twice a year, which makes for a schedule that is “difficult, but self-inflicted.” As potential partners line up to be part of the next batch of student projects, Hosoi’s network of front-office and corporate contacts has grown rapidly. Maintaining those relationships takes time.

“Lots of people get to be a dean or a provost,” she says. “But there are not a lot of people who get to open up this box that we’ve opened up, where it’s like, ‘Oh, I think I want to do hockey this year; let’s call the Islanders and see if we can spend a day with them.’ Or, ‘Today I’m interested in football; let’s call the Steelers and see what they’ve got.’ ... So do I give that up? I don’t know.” 

One thing she does know: As the data collected in the sports world grows exponentially, so does the demand for smart people who can make sense of it. When the Oakland A’s adopted a more analytic approach to baseball in the early 2000s (popularized in part by Michael Lewis ’82’s best-seller Moneyball), they were relying on undervalued statistics — on-base percentage and slugging percentage — that now seem quaint. Today, through baseball’s Statcast system, every movement on the field is measured, from the spin-rate of a pitcher’s slider to the exit velocity of the ball when it leaves the bat. The data set in the MLB home-run study included more than 300,000 batted balls. Other pro leagues have similar in-stadium trackers to record movement during games, and wearable tech has simplified data collection during training, from youth sports to the pros. 

Moneyball was amazing,” Hosoi says, “but the size of the data sets we have now are orders of magnitude more than that. And so the question is, when you have these gigantic data sets, how do you pull out the insights that are actually going to be useful to the coaches and the athletes?”

Translating the numbers into something actionable requires a higher level of understanding — mathematics, computer science, machine learning. “To date it’s an underutilized resource,” Hosoi says, “because people have not extracted the real meat of the information that’s in that data.” 

Brett Tomlinson is PAW’s sports and digital editor.

0 Responses

Join the conversation

Plain text

Full name and Princeton affiliation (if applicable) are required for all published comments. For more information, view our commenting policy. Responses are limited to 500 words for online and 250 words for print consideration.

Related News

Newsletters.
Get More From PAW In Your Inbox.

Learn More

Title complimentary graphics