Researchers at the University of Massachusetts Amherst tested a prototype marathon shoe that PUMA created with computational design, and found it improved running economy by up to 3.6%. This overcomes a 1 to 1.5% improvement plateau in the advanced footwear technology (AFT) space since these so-called “super shoes” hit the road in 2016.

Image

“From the outside, [the PUMA prototype] is a regular AFT shoe,” says Wouter Hoogkamer, assistant professor of kinesiology at UMass Amherst, senior author of the paper and director of the UMass Integrative Locomotion Lab (UMILL). “But [PUMA] mixed the ingredients in a way that it’s a substantial step up.” He notes that the only other time he has seen such robust results on running efficiency was when AFT was first introduced to the running world.

For this study, funded by PUMA, Hoogkamer and his team evaluated running economy of four AFT models, without further involvement of PUMA: the Nike Alphafly 3, Adidas Adios Pro Evo 1, PUMA Fast-R 2 and the prototype of the PUMA Fast-R 3.

“Running economy is basically equivalent to your car’s gas mileage,” Hoogkamer explains, adding that some cars can drive more miles per gallon. “If I go out there and run at a 7-minute-mile pace, I would be using a specific amount of energy. If I train for years, I will improve my running economy, and at that same 7-minute pace, I will use less energy.”

However, in 2016, marathon running changed when Nike introduced the first AFT footwear. Hoogkamer’s previous work found that this design improved running economy by 4%, essentially bypassing years of training simply by swapping footwear.

These days, AFTs are ubiquitous in the marathon space, and Hoogkamer says that, on average, there is about a 1% difference in running economy between different AFTs. “We haven’t seen huge improvements recently, which made everybody in the field think that we sort of have reached this plateau,” he said. “And that’s what we also thought until PUMA mailed us this shoe.”

They found that the PUMA prototype is 3.15% to 3.62% more economical than the other popular AFT models. “These are substantial differences,” says Hoogkamer. “You’re talking about minutes off a marathon time, and I train years to shave a minute off of my marathon time.”

Consider the performance of Kenyan professional long-distance runner Edna Kiplagat at the 2024 Boston Marathon: She finished third with a time of 2:23:21. A 3.15% running economy improvement would have allowed her to finish at 2:20:13, more than 2 minutes faster than the winning time.

Image

Monty Bertschy, a doctoral candidate at UMass and research assistant at UMILL, explains that running economy improvements are typically small or inconsistent between participants when comparing existing AFTs.

“On a group average, we [normally] see maybe 1 to 2% differences, but that average is a result of some people responding very well, and some people not responding very well,” he says. “However, with this shoe, we see that, universally, everybody responds positively to this new footwear condition.” Unlike other comparisons between AFTs, none of the 15 participants ran with a worse running economy. Even the worst responder still saw a 1% improvement in running efficiency. 

PUMA used a data-driven computational design process. Instead of building a shoe, testing it on a person and tweaking the design, they created a digital version and used data-driven finite element analysis to play with the properties to see how that affects the shoe’s performance before it ever sees real feet. This process identified the optimal foam placement, material properties, carbon fiber layering and overall geometry to minimize weight, maximize energy storage and return, and achieve the desired carbon plate stiffness — all properties known to improve running economy and running performance.

“However, it’s still a little murky on which features we need to turn the knobs higher on and lower on, but the process is really what has gotten them there,” says John Kuzmeski, a doctoral student at UMass and research assistant at UMILL.

Image

While the FR3 is 80 grams lighter than previous PUMA model, weight alone could not explain this pronounced improvement. Using an Instron E10000 at UMass’s Institute for Applied Life Sciences, the research team was able to mimic some of the peak forces in the shoe and look at some of the foam properties over the full bed of the midsole.

Through this, they suspect one of the most meaningful parameter changes was about a 3% improvement in energy return compared to other AFTs, which means the shoe acts more like a trampoline and less like wet sand. 

Hoogkamer also notes that this demonstrates the feasibility of creating computationally optimized custom footwear. “This shoe was developed based on average data for several runners at a specific range of speeds,” he says, but it theoretically would be possible to put in an individual runner’s data to generate a shoe that meets their specific needs.

The full report of their findings can be found in a pre-print article. The study was funded by PUMA and Hoogkamer’s UMILL research team has previously received multiple research grants from PUMA and Saucony. Laura Healey, another author on the paper, is an employee of PUMA. The authors declare that PUMA did not have any influence on the views presented in the article.