Skip to content


‘Data is the new doping’

The use of data in the training of athletes has only just begun. Sensors and statistics are boosting performances.

The Data Science & Sports Seminar brought people together from universities, sports and companies in a series of short updates on how data science is used in sports. The presentations fell roughly into two categories: data for training and predictions of future results based on historical data, also known as ‘Sports Intelligence’.

20160407_tudelft_ewi_sportscience_webhr

Speed skating coach Jac Orie, who worked together with data scientist Arno Knobbe from Leiden University, best illustrated the use of data in training. Orie has collected 15 years’ worth of training data from 40 different athletes. His data show the importance of what he calls the ‘periodization of training’. Simply said, training is followed by fatigue and recovery, the super compensation, a period in which the skater performs best, and return to pre-training level.

To perform best at the day of the races, it all comes down to timing and dosing the load (intensity x duration) of the training. Knobbe analyses Orie’s data like heart rate, lap times, subjective intensity scores and more to determine just the right intensity of the training. Do less and you’ll get too lazy, do more and you’ll get over trained. Kjeld Nuis, who won gold medals in 1,000 metres and for the ‘grand medal’ during the last World Cup, was trained using this method.

Ruud van Elk supports trainers at football club PSV. The data he collects on the continuous monitoring of the position of each player in the field, their heart rate, their accelerations and their interaction with the ball are all digested into performance reports for every player each 15 minutes of the match. During the halftime bread, the coach gets reports on who performed on, below or above average. The coach can use this input to give feedback to the players or to order a change of players.

Simon Gleave from Gracenote (formerly known as Infostrada) exemplified the other branch of data-usage in sports, to wit: Sports Intelligence. “Sports data are easy to get”, he said, “but difficult to manage.” Gracenote collects scores from world contests of almost every sport. They process these results into predictions of upcoming sports events.

The Virtual Medal Table’ for the Olympics is a good example. It gives a list of medals per country calculated by Gracenote. Media and national Olympic committees can hardly wait for the next monthly update. For the last Olympics of 2012, Gracenote had 22% of the gold medals correct and 54% of the medalists if you ignore the colour of the medals. Currently, the Netherlands is 13th on the list with 27 medals.

“Data science is the new doping,” Professor Joost Kok (informatics and medicine at Leiden University) said, only half joking. Now that sensors are small, cheap and almost everywhere, the main challenges for sports data scientist are according to him:

– Matching data processing to the rate of data acquisition

– Developing standards in technology and methodology to improve exchangeability

– Keeping track of a purpose while using data science in sports

– Making data relevant to the end-user.

That same call also came from Kamiel Maase, who coordinates scientific support of elite sport with the Netherlands Olympic Committee. He said: “Data is nice, but it is all about the information you can retrieve from it.”

–> View more events from Delft Data Science

Posted in Articles, Delta.

Tagged with , , .


0 Responses

Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.



Some HTML is OK

or, reply to this post via trackback.