Petter Engdahl's pacing strategy for the Mont Blanc Marathon

?? How do you achieve a podium in the Mont Blanc Marathon ?

Well, there are no secrets: you have to train. A lot.

?? Data doesn't work miracles. It's the icing on the cake, the one that allows you to gain a few percent at most.

But a few percent is already a lot.?

The question I'm going to answer is: how do you define your pacing strategy? How do you use your energy in the most interesting way possible? And that's all thanks to data!

Petter Engdahl , an amazing athlete from 阿迪达斯 Terrex ran the Mont Blanc Marathon last month. With Enduraw , we mapped out his race briefing, published on Linkedin a few days before the race. He finished the race 3 minutes ahead of our forecast (1.51% difference).

This article will be divided into two parts: first, how I determined the race time and the pace breakdown, and second, whether our predictions were right.

I) How was Enduraw's pacing calculated ?

Over the 3h45-minute race, Petter finished with a 3'26 difference. A small margin, a tiny trifle when you consider all the hazards of a race at this level and all that can happen in the mountains. 1.51% difference, to be precise. Everything had been taken into account: the temperature (revised upwards the day before), the slight lengthening of the course, etc.?

So what's our secret sauce ?

First of all, there are two ways to do a race briefing:?

A) The first and mathematically purest way is based on energy consumption.

The athlete has a certain amount of energy (CHO and fat) which we're going to burn as cleverly as possible. This means adapting the speed to the athlete's strengths and weaknesses (gradient, technicality, temperature, luminosity, sprint, etc.) so that he goes as fast as possible while being as sparingly as possible.

A quick reminder of my article on nutrition and hydration: the body has two energy reserves (fats and sugars) which can be used to move around. Their quantity, speed of use and waste products are different, enabling us to develop optimization models like this one, with the athlete's speed and ability to assimilate the feed.

This is the most interesting and fairest method, as it takes only the athlete into account. It's particularly suitable when the objective is the best possible performance to go for victory. The Iron Man format is particularly effective for this method, as drafting is forbidden and pace management is key. Increasingly, we're seeing that in a race like the UTMB, athletes are racing for themselves rather than against others. In my opinion, this is the most intelligent way of racing, as it adapts to the athlete in question down to the last millimetre.

However, this method requires extreme knowledge of the runner so that the algorithm can come as close as possible to physiological reality.

B) The second depends on the expected time calculated beforehand.

On the other hand, we can look at things from another angle: we can set ourselves a time and calculate the best energy expenditure to achieve that time, and thus the breakdown of the pace. This will be the case for an FKT, a world record or an Olympic qualification. This method is easier to understand and, above all, it enables us to calculate the performance improvement due to pacing using a protocol that we have developed (in triathlon in particular).

For Petter, I used the second method because I had a good idea of his overall level, but with his somewhat truncated preparation, I found it hard to estimate his downhill abilities. I knew that he was capable of transcending himself on his training runs, but Rémi Bonnet, author of the world's best performance of all time on the KV, seemed untouchable.

So I estimated Petter's performance using previous years' performances, Petter's abilities, the distance (the course had been lengthened), the temperature, ...?

Many people and athletes ("Casquette Verte" being the latest) tell me that predicting trail performance is mission impossible because there are too many parameters. There are too many parameters, that's a fact, but that's what makes it all the more exciting. And if you don't start one day, you'll never get there. start with the altitude difference, the technicality, the temperature, then I add nutrition and hydration, the wind, the competitors, then again, the sensations, the HRV, the equipment, and in real time, fatigue, the facts of the race...

I came up with a fairly ambitious forecast of 3h43'18, which he could achieve if he didn't start too fast.?

That's the idea I've been advocating from the start with Petter (CCC 2022): you can't start too quickly. On the CCC (100 kilometers from the UTMB) he was 8th in the theoretical ranking. According to the UTMB index, he needed 20 minutes more than Jonathan Albon to complete the race. My insight at the time was: don't fight for 1st place, fight for second. All your rivals ahead of you will make the mistake of following Jon and will go off in a pack over the first 25 kilometers. You, theoretically, are 20 minutes slower than Jon over 100 kilometers, so over 10 kilometers you MUST lose 2 minutes. He'll pay 3 times more for those two minutes if he over speeds at the start. If he feels fit, he can then accelerate.

Petter set off calmly and stayed on the pace (with a difference of less than 50 seconds over the first quarter of the race). Just before Champex (km54), he suddenly accelerated and caught all the competitors who had gone too fast, including Jon Albon. A last finish of anthology and he wins in Chamonix, with the event record under his belt.

From this race, I had recovered Petter's GAP (gradient-adjusted speed) profile.

?

It's a concept that enables us to estimate the speed Petter could have had on the flat at the same energy expenditure. I started working on it in 2021 with the help of Simon Gosselin from the Team SIDAS x MATRYX Trail Running , and I had released a more precise algorithm in 2022. Strava had developed a formula but it wasn't precise for two reasons. The first was that it took into account gradient rather than technicality, and you know very well that technicality reduces speed (especially downhill). The second aspect overlooked by Strava is that this GAP is individual. Indeed, a mountain professional like Petter or Kilian Jornet will expend much less energy to achieve the same ascent speed. COROS has understood this need for effort estimation and has released a native GAP calculation integrated into the watch. ( Derek Dalzell )

Thanks to this GAP profile, I knew exactly how much Petter could climb and descend. As far as technicality was concerned, the data came from the CCC and therefore from the Chamonix alpine trails, the same as those used for the Mont Blanc marathon and at the same altitude. Had we used data from the Hardrock or the Western State, the calculations would have been distorted.

Applying this GAP profile, the temperature versus altitude for each point on the course and the wind direction, I deduced the following pacing plan:

II) How good was Enduraw's pacing?

Now let's see how good Enduraw's predictions were. With a time gap of 3'26 over 3 hours 45 minutes, the results speak for themselves! But let's take the analysis a step further.?

The following curve shows paces kilometer by kilometer

We can see the closeness of the two curves, and also the impact of the elevation gain. Around kilometers 15 and 16, we can imagine the climb up to Les Posettes (15.5-18 km) and at kilometer 30, the climb up to Le Béchar (29.5 km).

If you look a little more closely at the curve, you'll notice that the very beginning was faster than predicted (with speeds of around 3:35min/km). It's the start, and the adrenalin of the start makes you go much faster than you feel. But I still say that this speed is too high and will be costly in the end. We can see that the pace of the climbs is generally well respected, but the end was a little more difficult for Petter, as the red curve is higher than the blue curve. If the start had been smoother, would Petter have finished stronger? It's worth noting that his start was already calmer than that of competitors such as Anthony Felber or Maximilien Drion , who both reduced their speed (compared with Rémi) on the Flégère climb.

Here are the trajectories of the different runners. You can see that they stayed together until kilometer 10 (Manuel Merillas and Simon Paccard made the right choice to set off at their own pace). Rémi first accelerated at the Planet, then put in a violent attack on the Flégère climb. Petter made a fine comeback on the Flégère, moving up from 9th to 3rd.?

With no hope of seeing the second ( eli hemming ), this may well explain his slower than predicted descent.?

On the following curve (gap for each kilometer), we can clearly see his very fast first part of the race, his middle part in the paces and his smoother finish.

For the UTMB briefing it would be interesting to add a slight drift to the predictions. With this in mind, Enduraw carried out a scientific study on this Mont Blanc marathon to measure the impact of fatigue on performance factors. 7 hours of running in the mountains with a dioxygen analyzer and 13k€ of sensors! The natural drift (good resistance to fatigue, "diesel" athlete, 30km wall, sprint ability, etc.) of each athlete can be determined in this way and linked to the race strategy.

To conclude it was a legendary race on the Golden Trail Series circuit and a fantastic comeback to claim the podium! As far as data predictions are concerned, it just goes to show that after a few practice races (CCC and Transvulcania), you can be super accurate, and that data can provide the hyper-personalization that is the key to top-level sport.

Next step for me: the legendary Norseman Xtreme Triathlon , an XXL triathlon in Norway with Margaux Gressé , a half-marathon in Singapore with the Human Telemetrics team, then the UTMB? Group . This time we're taking it up a level with real-time data capture (HR, glucose, body temp, stride degradation) to influence the pacing plan and feed stations throughout the 170 km.?

This time, it will be with a new athlete. Here's a hint: he's a very fast runner!

Thanks for reading, I look forward to your questions in the comments section!?

Feel free to contact me for your race plans or if you're looking for performance.

Our sponsor : Salomon Polar Electro Oy CORE Supersapiens VO2 Master Moxy Monitor Stryd Nix Biosensors On Maurten RunMotion Coach Upside Strength COMPRESSPORT

--?

?? I'm a data scientist and I optimize the performance of high-level athletes.

?

My dream is to become an Olympic champion by coaching an athlete or to win the UTMB scratch race with a female athlete thanks to data analysis.?

?? Follow me on my journey !