Skating faster thanks to data science
The Netherlands has a lot of world-class speed skaters and footballers. Clever statistical analyses could help them perform even better. Leiden University is working on this together with Delft and Amsterdam.
Image: The skaters of the LottoNL-Jumbo team train in the summer on their racing bikes. Left: coach Jac Orie.
The three cities recently started to work together in the Sport Data Centre, a network of data researchers. Leiden University's Centre of Data Science (LCDS) is also a partner. The researchers are examining how data can contribute to better performance. They are also looking for medical applications, in physiotherapy, for example. The partners presented their first results during a seminar on 7 April in Delft.
The Leiden data scientists examined the training plans and competition results of the LottoNL-Jumbo skating team under coach Jac Orie. The team – that includes world champion Sven Kramer – is one ofe the top teams in the world. Nonetheless, they may be able to make even faster times. One of the conclusions reached by the researchers was that sprinters and mid- to long-distance skaters performed less well if they did too much endurance training.
Researcher Arno Knobbe: ‘All speed skaters have to have a good basic level of fitness, but the data shows that trainers can sometimes go over the top with that. They have their sprinters do too much endurance training, and that has a negative impact on their explosive power. It could be worthwhile changing the training routine to take this into account.'
Improving an excellent recipe
Knobbe believes that speed skaters like Kjeld Nuis and Sven Kramer could probably improve their speed by 1 to 2 per cent. That may not sound like much, but in the world of speed skating it can easily mean the difference between a gold medal and just missing a place on the podium. Or, as one of the data scientists put it: 'We're trying to make a good formula even better.'
Knobbe: ‘We used statistical models to look at the link between competition results and preparation. Does heavy training have a positive or negative effect on the competition? And should athletes rest the day before a competition or do a brief but intense workout?'
Wealth of information
Thanks to the cooperation of the LottoNL-Jumbo team, the data scientists have a wealth of information to work with. Speed skater Kjeld Nuis has been a member of the team for six years. In that period he has done around 500 workouts and 30 competitions a year. Everything is recorded, from competition results to the length and intensity of the workouts. Similar data are available on the other team members.
Eindhoven football club PSV is also a partner in the study. The club's competitions are all filmed with three cameras from different angles. That gives the exact X and Y coordinates of the players and the ball. These coordinates can be used to study all the movements on the field from a bird's eye point of view, and a huge range of analyses can be made of all kinds of technical aspects.
Leiden scientists are now studying the game situation. They are looking at such things as which movements eventually lead to opportunities, such as shooting for a goal and actual goals. Knobbe: ‘Which defender passed to which mid-fielder? Did the goalkeeper kick the ball far and high enough, or did he take the calmer option and give it to the defenders? If you measure the success of each of these actions, you find patterns that are more and less successful. What you're doing is showing clearly the chance of success. And once you have that information, you know where you need to focus in the training sessions.'
Sport and data in de National Research Agenda
Joost Kok, Professor of Computer Science and Professor of Medicine, is seeing increasing attention being paid to the combination of data and sport. The Sport and Exercise route has now been submitted to the National Research Agenda. As part of the preparations for this Agenda, Kok recently led a brainstorm session with representatives of the business sector. Their findings will form the Sport Research Agenda to be written on behalf of the Topteam sport. Big data is the link between the three themes: the value of exercise, better performance and life-long exercise. Innovation in sport and exercise can make an important contribution to a well-performing society that is typified by a dynamic economy, resilient institutions and vigorous citizens. Leiden University's strong position in data science and the research carried out at its Faculties of Social and Behavioural Sciences and Medicine mean that the University has a key role to play in this initiative.