To outperform better-funded rivals in events such as Women’s Team Pursuit, USA Cycling needed an edge. Coaches realized that smarter analytics of riders’ performance could be a key enabler.
The Women’s Team Pursuit team worked with IBM jStart to transform its approach to analytics, instantly delivering eye-opening performance metrics to the coaches’ tablets after each training session.
Instantinsight helps coaches reinforce winning behaviors as soon as riders finish training
8 hoursper day saved on analysis, enabling coaches to spend more time with riders
One thirdof the team’s 3% annual improvement target is expected to come from technology
Business challenge story
Analytics and track cycling: a match made in heaven
If you were to ask a group of statisticians what makes a football, hockey or basketball team successful, the debate might last for hours. Even in a sport like baseball, which is relatively amenable to statistical analysis, the complexity of the factors that contribute to a team’s performance mean that analytics is as much an art as it is a science.
Track cycling, by contrast, is a statistician’s dream sport. Races are held indoors in a velodrome, on a flat, smooth track with no weather conditions to take into account; and the bikes have a single fixed gear, so there is no need to worry about optimal gear ratios or gear changes. As a result, there are very few significant variables that determine whether a rider or a team wins or loses, and performance can be measured using a relatively simple set of equations.
Andy Sparks, Director of Track Programs for USA Cycling, explains: “The single most important factor in winning a race is the power that the riders are able to exert on the pedals. The bikes we use have a power meter on the crank that measures the power generated in Watts. The introduction of the power meter has totally transformed competitive cycling, because it lets us quantify exactly what riders need to do to achieve a certain time in events like the Women’s Team Pursuit.”
Team Pursuit is a four-person cycling event in which the riders aim to complete 16 laps of a 250 meter track in the shortest possible time. Each rider takes turns to assume the “pull” position at the front of the team, while the other three members ride in their slipstream. The slipstream effect is so significant that it requires approximately 30 percent less effort to follow than it does to pull.
When a rider’s pull comes to an end, they need to move to the back of the group as efficiently as possible. Even a smooth exchange slows the team down by around 0.5 seconds, and a poorly executed exchange can be very costly: if a rider falls too far back, they need to exert a significant amount of energy to catch up and get back into the slipstream. Such errors are known as “matches burned”.
In a sport that is often decided by fractions of a second—for example, the USA team’s victory over Australia in the semi-finals at the London 2012 Olympics was by less than one tenth of a second—it is critical to plan and execute the pulls and exchanges to perfection. Deciding who should pull when and for how long is critical to get the best possible performance from each rider and help the team become greater than the sum of its parts.
Andy Sparks explains: “The ability to measure the key factors that drive race performance gives us the ability to set targets for what we want to achieve. Actually achieving those targets is a different matter. We aim at improving our overall race times by three percent per year, of which one percent should come from the athletes’ physiological improvement, one percent from team and staff fortification, and the final one percent from technological innovation.”
Neal Henderson of APEX Coaching & Consulting works for USA Cycling as a High Performance Consultant. He comments: “From the technology side, one of the most important things we do is analyze the data that the bikes’ power meters capture during our training sessions. This helps us see the wattage that each rider was producing at every stage of the race, and also calculate W-prime depletion—a measurement of how much of their anaerobic muscle capacity each rider has used up during the session and how long it takes to regenerate.”
However, gathering and analyzing the data from these sensors has always been a challenge. After each training session finished, Neal Henderson had to plug the head unit of each bike into his PC, download the data, manually slice it into half-second intervals, match those intervals to the events that took place during the session (for example, when each rider was pulling, versus when they were exchanging or pursuing), and then calculate a variety of key metrics. This took at least one hour per rider—and even when all four riders’ data was ready, there was still the task of collating and comparing individual results to get a 360-degree view of team performance.
Andy Sparks comments: “Neal would regularly be staying up past midnight each day during a training camp. He’s an incredibly valuable and talented coach, and we share him with several other cycling teams—so while he’s with us, we want him to be interacting with our riders, rather than spending all his time crunching the numbers on a computer.”
Neal Henderson adds: “The time taken to get the data also had an impact on how effective the analysis was as a coaching tool. If you’re talking to a rider about what they did on the track yesterday, that’s much less immediate and powerful than if you can talk to them while their legs are still burning from the session!”
Finding a better solution with the Internet of Things
Working with IBM jStart, the Women’s Team Pursuit team is now harnessing emerging technologies to solve its analytics challenge. Instead of manually extracting the data from the power meters and sensors after each training session, the data is automatically collected by an Android phone in the rider’s pocket, and transmitted to the cloud, where it is stored and analyzed as soon as the session finishes. Within seconds, the results are then sent back from the cloud to the coaches’ tablets, in the form of a summary dashboard that presents metrics such as W-prime depletion and matches burned in an intuitive graphical format.
From a technical perspective, IBM Watson™ Internet of Things Platform acts as a cloud integration hub, receiving the data and directing it to other components of the solution. For example, the raw data from the sensors passes through a Node-RED storage flow to an IBM Cloudant® database, which is used to supply the summary dashboard, and also feed a Jupyter Notebook for more complex analysis by the team’s data scientists.
Other components, which will be coming online soon, include the use of IBM Analytics for Apache Spark to calculate metrics in real time. This will allow the team’s coaches to monitor performance not only after the training session, but while it is still in progress—for example, during each exchange, the coaches will be able to see whether a match was burned. The team even plans to introduce smart glasses, which will provide a personalized head-up display of whichever key metrics are most useful for each of the riders, while they are actually on the track.
Andy Sparks comments: “We always had a vision that this kind of thing was possible, and IBM jStart is helping us turn it into a reality. We have been so impressed with the jStart team’s ability to orchestrate all of these emerging technologies to build a solution that delivers exactly what we need, in seconds.”
Neal Henderson comments: “The ability to get hold of the data immediately after the training session has finished has completely changed my relationship with the team. I’m spending much more time with the riders and the other coaches, and because we can all see the data instantly, it’s much easier to identify problems, make adjustments, and reinforce winning behaviors that they can take into the next session.”
On track for victoryThe team began using the solution shortly before its victory at the 2016 World Championships in London, and will continue using it at training camps in the lead up to the 2016 Olympic Games in Rio.
“This year has already been one of the most successful in our history—at the World Championships, we won the qualifier by four seconds, beat the previous US record by six seconds, and took the gold medal in the final,” says Andy Sparks. “Although we only started using the IBM solution a few weeks before the event, we immediately saw its potential to help us identify and fast-track tactical and technical improvements. It’s the most important technology project we have worked on this year, and we see it as a key tool in our preparations for Rio.”
Neal Henderson adds: “We always aim to train as hard as possible, to make racing as easy as possible. The solution helps us show our riders exactly how effectively they’re working, so they can see that what we’re asking them to do in competition isn’t impossible—it’s what they’ve trained for, and what they’ve achieved 100 times before in training.
“The ability to instantly quantify and reinforce the positive gains made during each session really helps lower the stress of competing in a big race, and helps the riders focus on executing the performance that they already know they are capable of.”
Although certain elements of the solution cannot be used during competitive races, USA Cycling believes that the dashboards could become even more valuable during the intense time-pressure of a competitive event.
Neal Henderson says: “When you’re at an event, there’s only a very short window between the races—there just isn’t time to spend four or five hours pulling data together, and even if there were, the riders need some downtime instead of worrying about what happened in the last race. The big advantage of instant analytics is that we will be able to give the riders a quick debrief after the first race, advise them on tactics for the next one, and then just let them relax and recover.”
Andy Sparks concludes: “The whole engagement with IBM jStart has been in line with our culture of excellence at USA Cycling. The jStart team have the same principles—everything they do is delivered to the highest possible standard, and we’re proud to be working with them to push the boundaries of what is possible in cycling technology.
“Compared to other big cycling nations, our budget is very tight—we rely 100 percent on sponsorship, whereas many teams receive large amounts of government funding. Yet we’re showing that we can compete successfully at international level. IBM deserves credit for helping us train smarter and free our riders to execute successfully at the major events. Of all the technology projects we’ve worked on this year, the IBM jStart project has made the biggest contribution to achieving our goals.”
About USA Cycling
USA Cycling’s mission is to achieve sustained success in international cycling competition, grow competitive cycling in America, and deliver an exceptional customer experience to the nation’s cycling community. Its athletes have won medals in numerous cycling disciplines at major international events; and in 2016, the Women’s Team Pursuit team won its first gold medal at the World Championships in London.
- IBM Analytics for Apache Spark (Managed Cloud Spark Service)
- Cloudant SaaS
- CBDS: Watson IoT - Emerging Technologies
Take the next step
To learn more about the work of jStart and the IBM Emerging Technologies team, please visit ibm.com/jstart. IBM Analytics offers one of the world's deepest and broadest analytics platform, domain and industry solutions that deliver new value to businesses, governments and individuals. For more information about how IBM Analytics helps to transform industries and professions with data, visit ibm.com/analytics. Follow us on Twitter at @IBMAnalytics, on our blog at ibmbigdatahub.com and join the conversation #IBMAnalytics.
View more client stories or learn more about IBM Analytics