Eindhoven University of Technology: This is how our speed skating stars ride their perfect Olympic race

Take the first question from the intro above. It is clear to the researcher at a glance what De Jong’s battle plan should look like. She will have to ride a considerably flatter race to keep Schouten from an Olympic title. Willemsen: “De Jong can only go faster if she starts slower. If she wants to finish in three minutes and 55 seconds, just like Schouten, the first two laps will have to be considerably slower, so the last laps can be faster.”

Willemsen developed the profile below, with the distance on the horizontal axis and the lap times on the vertical axis. The yellow line represents the ideal, fictitious 1,500 metres of De Jong. She starts significantly slower than during her races at the Dutch Championships (blue line), the European Championships (grey line) and the Olympic Qualifying Tournament (orange line), but her final sprint could be significantly faster. Willemsen, laughing: “But maybe her coaches have already concluded this.”

Willemsen is also a passionate skater, and certainly not without success. For instance, he competed in the 1,500 metre event at the Dutch Masters Championships, for skaters over 40 years of age. His children skate as well.

He has been doing research into the prediction of personal records of skaters at all levels for some time. Two years ago, he wrote a scientific publication about this with colleague Barry Smyth, who had previously applied this to marathon runners. The idea: via so-called case-based reasoning (reasoning based on similar cases from the past), it is possible to determine accurately and in a personalised manner how skaters should build up their lap times in order to achieve a faster end result. Ideal for talented skaters who want to take the next step, but also the top skaters can learn from this.

The principle is simple: every skater has his or her own level. In order to raise your level, you can learn from the lap times of your peers in the same age category. A requirement is that these peers have a similar race structure and have improved themselves significantly in the years since. Did these peers train for a faster opening? Or did they ride flatter laps – with less decay – in order to achieve better times?

Especially on the longer distances a good build-up of lap times – the so-called pacing – is of great importance, Willemsen explains. “From the 1,500 metres onwards, pacing becomes relevant. A lot depends on what kind of rider you are; are you more of a sprinter who starts fast and hopes he is still fit enough for the last lap or someone who starts slower and comes on strong in the final phase?”

He developed an algorithm that links riders with similar lap times on the basis of case-based reasoning and applied it to the large amount of race data he obtained from Osta (which keeps track of skating statistics). He is currently developing an app, which is in the testing phase. In this app, athletes can enter the laps of their personal records. At the push of a button, the app will produce a prediction showing how skaters can improve their 1,500 or 3,000 metres.

Martijn Willemsen
“Ideal for talent development in skating,” thinks Willemsen, who is in contact with the Innovation Lab of skating stadium Thialf about how these insights can help coaches train the Olympic champions of tomorrow.

“There is often only a one or two percent margin of error in the model’s predictions,” Willemsen clarifies. “You don’t even need the lap times of a particular distance to make an accurate prediction. You can therefore predict, based on a 1,000 and 1,500 metre race, how someone should approach his or her race in the 3,000 metre event. This is useful if someone has (almost) never raced a certain distance, because based on shorter distances you can still come up with a plan and predict the end time.”

“However, it is important to correct for the type of track. Semi-open tracks that are more exposed to weather often have slower times than closed tracks, especially if the humidity and temperature is conditioned, as in Heerenveen.”

There is also a drawback to his future app, Willemsen says with a laugh. “Skaters like myself, at master level (over the age of forty) for example, will not see their lap times go down, but up. The times are of course based on the times of their peers, and at a certain point the deterioration really begins.” Then, with a wink: “But if the data shows that the deterioration is actually not that bad, that can actually motivate again!”

Comments are closed.