Pages

Tuesday, May 24, 2016

The Most Powerful Man In Cycling Data?

An interview with Robby Ketchell, Chief Data Scientist at Team Sky
The world of sport has become a numbers dominated space.
Whilst athletes and coaches used to be the sole recipients of the data, today it is permeating through every part of sport, from the ways in which fans interact with their favourite teams, through to how bars in stadiums are stocked.
Throughout this transformation, some sports have moved from focussing on tradition and what has worked for decades, into powerhouses of data. The best example of this is cycling, where traditional training techniques were still being used and unfortunately improvements tended to come from the use of illegal substances rather than pure athleticism.
The rise of data use has seen this doping culture more or less obliterated as teams like Team Sky and Giant Alpecin have seen huge successes whilst openly avoiding performance enhancing drugs. In fact, the policy of marginal gains implemented by Sir Dave Brailsford at Team Sky would not be possible without the extensive use of data and data gathering techniques. This same policy was responsible for Sir Bradley Wiggins and Chris Froome winning the 2012 and 2013 Tour de France races.
With this approach now being adopted by several other teams, the race is now on to implement marginal gains 2.0. To help with this search, Team Sky brought Robby Ketchell on board to help with their analytics and data programmes. Despite admitting that they are doing this better than any other cycling team at the moment, Brailsford admits that “We are, but even the most sophisticated data-driven companies such as Google and Facebook are constantly evolving and changing,”. This makes Robby the man in charge of the data at the most data driven cycling team in the world, is he therefore the most powerful man in cycling data?
Robby’s role at Team Sky sees him working with their data to find the small incremental improvements that will hopefully bring about further improvements to the team. He has an impressive history of success with the Garmin-Sharp (now Cannondale-Garmin) team, where he worked with cyclists such as David Millar and Bradley Wiggins.
Ahead of his presentation at the Sports Analytics Innovation Summit in San Francisco, we spoke to Robby about the change in cycling, his role at Team Sky and the datafication of sport in general.
Innovation Enterprise: Do you think that cycling has now become a numbers based sport?
Robby Ketchell: Numbers have always been a big part of sports, not just cycling. Endurance sports in general have recently become more and more data dependent with new sensors that measure aspects of physiology and physical performance. Cycling has grown to become more of a numbers aware sport with similar sensors, social media and using humans as sensors, onboard devices, and software dedicated to the analysis of all of the data collected.
Team Sky’s success has been based largely on the idea of marginal gains, where do you see marginal gains 2.0 taking us and how will powerful data gathering/analysis tools help with this?
Marginal gains is the concept of continuing to improve every aspect of performance a little bit at a time. Now that cycling has become a data rich environment, we're continuing to seek improvements in the way we collect and interpret data. We try to improve our performance by using data to make better informed decisions.
The Pro-Peloton is likely to change considerably in the next few years with new technologies, such as disc brakes, being introduced in 2016 - how important will data be in the integration of these to improve performance?
Every time new equipment is introduced into the sport, sponsors and teams spend a lot of time analyzing the performance of these innovations by either going to labs like wind tunnels or testing in the field with devices like the BATbox [a box that sits at the front of a bike to calculate air resistance]. In addition, the athletes spend some time testing the equipment and giving feedback so that we can optimize performance. This is something that's important to the design of any innovation, whether it be a piece of software or a new aerodynamic wheel, getting the user's feedback helps drive the development. Using data in conjunction with some of these subjective measures is important to improve the performance as well as ensure the safety of the athletes.
Do you think that professional sports, and cycling in particular, are close to being able to utilize traditionally business focussed products like Hadoop, to help analyze performances?
It all depends on the goals and setup of each organization on whether using these tools is appropriate. Everyone wants to get to the point where they can do Big Data Analytics, but to get there we need to do a lot of setup by warehousing and cleaning our data. Without these initial steps, the analytics part is either not possible or only possible for smaller projects.
New tools are being developed every day that help with this and as long as they're used correctly they can provide a significant improvement to how teams share information and discover new possibilities.
With the proliferation of data being available in sports, do you think this has had an effect on the ability to identify potential doping cheats?
We now know so much more about the athletes due to increased data collection. Athletes now have a footprint that didn't exist in the past, which has allowed authorities to track performance gains and losses, health, and monitor events that weren't possible a few years ago. This puts authorities in a powerful position in regards to eliminating doping, but it also comes with a big responsibility. No matter how sophisticated technology gets, it is critical to take the results of any analysis within context of the sport and the environment.
Having worked within sports science, especially within cycling, for a number of years, how has the appreciation and understanding of data changed since you first began?
I think the biggest change is the understanding that data can be used to discover new possibilities. Previously, we used to do experiments with a hypothesis that something would occur, and if it did we would say we were on to something. Now we are finally getting to the point where people ask us to look at the numbers and see if we can learn something.

No comments:

Post a Comment