The man (and the math) behind the Strava segment hunter

Oct 1, 2018
How a pet project in a garage evolved into an exciting new way to train, and perfectly nail Strava KOMs
The man (and the math) behind the Strava segment hunter

The Innovators is a series of stories we've crafted about the creative individuals and companies that make truly special creations for cyclists. Whether its brilliantly handmade apparel, or clever new technologies, these are the innovations that excite and inspire us, so we figured we'd share their stories with you.

Xert's Strava "segment hunter" is both thrilling and addictive. It's no wonder No. 22 co-founder Mike Smith has become a big fan of the technology, chasing KOMs out on country roads with his Drifter. Like No. 22, the idea for Xert started with new ways of thinking about solutions to age-old problems. Here's the Xert story:

Armando Mastracci likes to solve problems. Seven years ago, the Toronto-based engineer was working a sales job for IBM, and admits that he wasn’t even really that into cycling. Today, he’s cracked the algorithmic mystery behind how we can get the most out of an endurance performance, creating software-based training tools like Xert, and its new Segment Hunter tool. And it all started on an old recumbent bike.

I started doing some riding and got talked into going on longer rides,” Mastracci says. “I bought a road bike, but started training with power because I had this old recumbent bike. It had a power readings on it.” Mastracci started treating the bike, the data, and ultimately himself like a study, doing RAMP tests, writing numbers down by hand, while on the bike, taking metrics in real-time. With all these handcrafted spreadsheets, Mastracci discovered patterns in the data that hadn’t been published in the current research at the time.

“I’ve got this mathematical analytical bent to everything I approach.” says Mastracci. “When I reviewed the work of the renowned physiologist A.V. Hill who won a Nobel Prize in the 1920s and his “force-velocity curves,” or, how force and velocity work in the body, they were almost identical to the curves I found in my own testing.” Mastracci says that was his “Christmas tree moment,” when the lights flicked on. “When you see your results improve by an order of magnitude, you know you are on to something.”

Mastracci moved his theories online, on discussion boards about exercise physiology. He started using various spreadsheets and solving formulas with the aid of Wolfram Alpha, the online mathematical tool, to figure out a perfect power-duration model. “It’s all an algebraic formula,” Mastracci says of what became his suite of algorithms to figure out how to best attack a Strava segment, and also training to your maximal ability level at any given period. “It’s very simple.” As an engineer, Mastracci stresses that he didn’t invent the underlying science that drives his software, he merely discovered it. “I got a hold of some patterns in the data. I’m more the messenger. I just follow,” he says. “The math behind how the body fatigues has always been there.”

Mastracci has now figured out a way to take a cyclists’ power data from a regular power file, essentially random data, and determine fitness. He then wrote software to that could do this on a server, and the key moment was figuring out how to have the software work on a server so that any athlete around the world could get their data analyzed automatically. The results could then be used on their bike computer while attacking a segment. “Segment hunter is just a formula, not a number cruncher,” Mastracci clarifies. “The math will will keep you on track, and the math works. It can do that in real time. Our What’s My FTP? app works along the same way, in fact."

What the team at Baron Biosystems quickly realized is that there’s a mountain of data now available to the average cyclist, but the challenge remains drawing a connection between the data and how it’s influencing you to better perform, and understand what’s working and what isn’t, empirically. “In the past, when an athlete didn’t want to get into it, they'd hire a coach,” Mastracci says. “They'd ask, ‘what does this data mean?’ And a coach tells you, inferring what’s going on. We’re trying to create a way where you can avoid all of that. The data is all actionable by you, you see it directly. It’s boiled down. It’s suddenly meaningful and actionable.” Mastracci is careful to point out that a coach still can play a vital role. “We are trying to provide coaches and individuals with a direction on how they should be training and performing,” says. “Before, a coach would say, ‘this is what you’re capable of.’But coaches wouldn’t know that until the athlete performed it. Now they know you can do it.

Other researchers didn’t believe it at first, but were soon convinced by the results. Mastracci left IBM a three years ago for good, and Baron Biosystems, his company, became his all-encompassing passion project, and quickly growing business venture. Today he has a good relationship with Garmin and Strava, and is working on bringing his technologies to other vendors as well. Zwift has become a fascinating testing ground for Xert. “We have seen these videos from users while on there,” Mastracci says. “We’re seeing someone in the middle of a race, they are putting up their MPA numbers on a screen.” Another user in a video of an outdoor crit put MPA on screen to explain what was happening.

It’s not surprising to learn then that Baron Biosystems is also talking to professional teams, something that Mastracci is really excited about. He feels that this sort of technology could alter the future of pro cycling, both in terms of maximizing performances, but also improving the spectator experience during a streamed or televised broadcast. “Imagine the segment hunter for a pro during a time trial,” he suggests enthusiastically. “The numbers then become either an affirmation of what you currently feel, and answer the internal dispute of ‘should I go? Maybe I can go.’ Some people are surprised by what they are capable of, and what they can accomplish.”

This second-by-second, real-time reading of how close to 100% a pro rider is nailing their effort could also could become the most intriguing part of a live stream of, say, a Grand Tour stage. “The announcers and the viewers can have a way of comparing what they are seeing, instead of speculating,” Mastracci says. “When you have the data in front of you, you can see that someone is doing an amazing job, or struggling.” In the end, he points out, that’s the game. “We’d be able to see how they are really playing their cards; how they perform relative to their ability. And it would help an audience understand what’s really going on.”

But what Mastracci is most excited about with his creation is that it helps cyclists of all ability levels to see how to best train and doll out a hard effort, and that just because others may seem stronger on a climb or are posting big wattage numbers, it is actually about nailing your own individual efficiencies. “The game becomes your own fitness and improvement,” Mastracci says. “It isn’t just about ‘can I beat this person in a race,’ or worrying that ‘my numbers are so modest.’ I’ll never beat Peter Sagan, and it’s not about that. It’s about how I make myself the best I can be; how do I encourage that. What drives that behaviour—happiness and motivation.”