How many medals will the U.S. Ski Team win in the future? What, really, is perfect ski technique? Gus Kaeding — the one man team behind the US Ski Team’s analytics department — aims to provide data-driven answers to these questions. From his office in the Center of Excellence in Park City, Utah, he delves into the data that may help illuminate a path to success for both individual athletes and the greater cross country ski collective in the U.S.
Kaeding, whose background as both a nordic ski athlete and coach led him to his current role with U.S. Ski and Snowboard, discussed his current and future projects with FasterSkier over the summer. Talking with him, it became clear that many assumptions about the sport remain unproven, and in his perspective, “we actually understand very little of what happens in between the start gun and the finish tape.”
Through data analysis and some exciting new technologies, Kaeding has a plan to gain a deeper understanding of the fundamentals of ski performance. This interview has been edited for clarity and length.
FasterSkier: I’ve read a little bit about what you’re up to with the analytics department, but I’m not totally up to speed.
Gus Kaeding: Yeah, I don’t think anyone really is.
FS: What sort of projects are you working on?
GK: If you watched the Bend camp, a number of coaches and athletes are now using Archinisis bibs. They basically look like a sports bra with a GPS unit in the back. I tested these last winter, and you’ll see a lot of the Scandinavian countries using them or something similar.
The Archinisis bibs can not only give you the GPS [data], but they can also tell you what technique you’re using. If you’re skating, it tells you your time in V1 versus V2, versus V2 alternate—to a pretty good percent of accuracy. The coolest thing is you can compare [the metrics] to each other, and also to yourself, meaning if you set a rollerski loop, and do it multiple times, you’re able to see an overlay of previous performances and where you’re gaining or losing time, and it’s really visualized nicely. You can do the same with multiple athletes, and then compare athletes and say: “you were three seconds slower than another athlete over the top of this hill. And as a result, you were five seconds slower at the bottom of the hill,” which are insights that we didn’t really have a lens into before.
That’s still in the test mode—we’ve got six of them. We had the option of rolling it out to the whole team and taking a big financial risk, but we chose to start small, and see where there’s value. Part of the risk when you roll out something like that is: who’s managing the project? Do they have the bandwidth to do it?
Someone like Jason Cork is super excited about things like that, and then if it ends up adding value, we’ll probably roll it out further next year, to a larger squad, and have a few bibs we can rotate through.
We may also have a few set courses around the country, so that when people come in, they can say, “I want to do a time trial against previous performances”, and afterward they can really look granularly where they’ve had success or need to improve a bit.
FS: What other ideas are you working on?
GK: The second one that I’m really excited about but didn’t get funded this year, is a ski treadmill at the Center of Excellence here in Park City. The problem is, the cross country team does not come here very often. One of the weaknesses, frankly, as a team is just general technique compared to the other nations—I don’t think that’s a secret. So we’ve been looking for a way to improve that for a few years now.
What we proposed was installing a 3D motion capture system around our ski treadmill, to get a better quantitative idea of what is good technique. What are the markers we should be looking for? And what are the markers we should be ignoring? That project did not get funded, but it got positive reviews. So I think in the future, there is hopefully a good chance [of getting it].
The idea would be to have the athletes come in once a year and do a movement assessment while on roller skis. From that, we would be able to break down multiple speeds, techniques, and inclines. Then we would look at where our techniques are similar across our athletes and where they are dissimilar. Where do we think they’re improving and where do they need some work?
I would build into that proposal getting someone who’s really got some stronger chops as a biomechanist. A lot of other countries have multiple people on staff who have PhDs in skate technique, so it’s like, “Yeah, no kidding, they have better technique”. So we need to do a little bit better job there. This is a project that has yet to come to fruition, but I think is going to be promising in the next 12 months.
FS: I believe Sweden’s been a leader in adopting 3D motion capture technology.
GK: Sweden with HD [Hans Christer Holmberg] and Norway with Oyvend [Sandbaak], do a lot of research in this area. Fortunately for us, they make a lot of it public, so we don’t have to start from scratch. We can play catch up for a little while until we feel like we’re in a place where we’re able to start to innovate ahead of the other countries. We’re still a few years from that, for sure.
FS: Tell me about your research into sleep?
GK: We’ve been looking for ways to quantify recovery and sleep. Finally, a couple of years ago, we decided to take the Oura ring as our primary recovery tool. So that’s been a project for the past two years.
We could look back at athletes who had Covid and their Oura data previous to that. Fortunately, we didn’t have a large sample set of people. Were the results conclusive on anything? No, probably not, but I will say the main similarity between everyone was that three days prior to showing Covid symptoms, their [sleep] temperatures just went bonkers. Whether high or low, it was all over the place. The deviation from the standard variation was nuts. So I think, for anyone out there that has a Whoop or an Oura, anytime your temperature goes one way or the other, watch out.
FS: What is the main goal of collecting this sleep data?
GK: It gives you some indication of how stressed out you are. It’s helped us add another lens to evaluate athlete load and upcoming performance.
FS: What are the challenges to gathering the kind of data you would like to work with?
GK: One of the major problems is, when the coaches are at races, they are busy—really busy, and you really can’t ask them to do anything more. You know, “go collect this data, collect that data”. And if you do, it better be able to return immediate, impactful results.
I think if we’re looking at the model that works in other sports, like soccer and basketball, you really need to have someone traveling with each team who’s just in charge of collecting and analyzing data. We’re just not at that point, as a sport, or as a program yet. Certainly one of the aims, probably in the next quadrennial, is putting a research/data person with each team: cross country, alpine, snowboard, just to gather data, analyze, and then innovate. That is in our long term planning. Until then, we are just not going to be able to even collect the data necessary to do any type of really interesting analysis. Because it does take having someone there, and that being their sole priority.
FS: That makes sense. It brings me to my other question of what sort of data do you wish you could gather?
GK: We do a lot of video analysis, but we don’t do a great job of storing that video. To me, it would be really helpful to have a video library of Jessie Diggins’ V1 technique through the years, to overlay year after year, and look where improvements have been made and then look at results on sections of trails. Have these V1 technique changes correlated to actual faster split times on V1 sections?
FS: Are you able to get enough video from World Cup races to effectively analyze those questions?
GK: We use the video for, say, qualitative analysis. I think the future is in computer vision and pose estimation [technologies] that would be able to look at technique just from the FIS feeds available now. They’re just starting to hit the sport level, in terms of price and scale. Those are technologies we’ve all got our eye on, and a lot of other sports do too. The NBA has started to dabble in it, but [the cost] is in the millions of dollars and the math is also hugely complicated. That’s the future, but it’s still a ways off for us.
FS: How do you envision data helping the program as a whole?
GK: We know, essentially, what results athletes need to get at each level. But we don’t do a good enough job of helping athletes figure out what is important to get those results.
The big goal of this next year, for every sport, is going to be to formulate a list with the coaches on what are really the most important factors to creating a good skier, and then figuring out ways to measure each of those.
Is it fitness? Is it power in the gym? Is it technique? Is it sports psychology? There’s so much on tap there. We need to do a little bit better job of delineating how important each is, and then being able to help the community and lead them a bit, showing what’s important at what age.
I think we do a pretty good job of that, but it would be nice to say: if you want to be one of the best skiers in the world, you should be achieving these results and here’s how to do it. We haven’t quite gotten there yet.
FS: That seems like a large project.
GK: Yeah. The goal over time would be to have all of [the categories] with objective measures. That’s another area ahead of us: innovative ways of measuring all those elements of a skier’s development and what the opportunities are to get better.
One of the things we’re working on is a sensor project that allows you to look at all these different metrics that we haven’t really gotten before, in order to understand what goes into an alpine turn. The technology is 100% applicable to cross country skiing as well, and now we can understand a bit more of what goes into the push phase.
Whether it’s cross country, or alpine, we actually understand very little of what happens in between the start gun and the finish tape besides time, pace, and heart rate. So it’ll be good to start to understand that with some more physical markers. So that’s where we’re at. The whole thing comes back to what matters.
FS: You mentioned that you can predict, to some degree of accuracy, future medals for the US Ski Team. How do you quantify that?
GK: We can break it down to: “we need this many athletes, at this level, to expect a medal in five years”, which is cool. I did a pretty quick analysis the other day — it was kind of back of the napkin math, but I figured out that with the implementation of the bigger development team that we started in 2018, and by really putting more of an emphasis there, we can expect another medal. [This is] simply from that team being established, and the level that it’s brought not only those junior skiers, but the rest of the skiers around them.
Once you start to get into expected medals, and expected performances, then you can do a better job of evaluating how well you’ve done as a staff, as a coach, and as an athlete. That’s something we’re a little bit lacking right now. Not that all those [calculations] are going to be correct, but you can continue to improve those over time. So far, they’ve played out pretty well.
FS: You have a few interns this summer that are nordic athletes, what are they working on?
GK: One of our interns is Annabel Hagen, who is currently at Harvard on the cross country ski team. She is working on a project that is near and dear to my heart. She is looking at a spike in USSA membership in 2012, and how it correlated to 2015/2016. Are we seeing a greater influx on the national team if we had 30% more participation at the 13 year old level in 2012? We break that down by numbers, but also by region, by state, and by club. We’re really looking at doing a better job with our demographics, and looking at where our athletes are coming from.
There’s also Jackie Garso, who’s a training biathlete right now and is a recent graduate from St. Lawrence College. She’s working primarily with our women’s alpine team, looking at our force plate data. Another is current US Ski Team athlete Sophia Laukli. She’s at Middlebury College and is a math major like myself. She’s been looking at alpine skiing when you start a World Cup at the back. She’s trying to figure out when it is beneficial [for an athlete] to go down a level into the Europa cup, and try to score their best points in order to get back to the World Cup with a better start spot.
FS: Gus, thanks for speaking with FasterSkier. Is there anything you’d like to add?
GK: We have a really smart community and a really innovative community, so if there are people out there who want to help or are interested in one subject or another, I am definitely willing to utilize their expertise.
If you’re interested in getting involved with the analytics department at the US Ski Team, you can reach Gus Kaeding by email at: email@example.com
Pasha Kahn writes and coaches in Duluth, Minnesota.