On a Monday in mid-November, Arnulf ran 25 kilometres with 760 metres of climbing: two and a half hours, steady, 197 TSS. The next morning he ran a 5K time trial in 23:24. That is 91 seconds faster than any 5K he has on record, run on legs that had done the hardest session of the week less than a day earlier.
His goal race is ten months away. On 17 September 2027 he lines up for TOR130 Tot Dret, the final stretch of the Tor des Géants route through Italy's Aosta Valley, from Gressoney-Saint-Jean to Courmayeur. The organisers list it at about 130 kilometres with roughly 12,000 metres of climbing. A race like that is decided on the climbs and through the night, not on a flat 5K. Which is exactly why the 5K is worth writing about.
The race behind the race
TOR130 is not the end point. When I asked him about the plan, he told me where it actually leads. In his own words (his quotes here are translated from German):
The TOR will probably be my first mountain ultra over 100 km. I may do one more race before it. My big goal, though, is the TOR330 in 2028, and to get there I have to qualify through the TOR130.
TOR330 is the full Tor des Géants loop, roughly 330 kilometres around the Aosta Valley. So next September is two things at once: his first mountain ultra past 100 kilometres, and the entry ticket to the race he is really training for. Everything below is the first year of a two-year build.
A base block that made him faster
There is a familiar idea that base training for a mountain ultra means months of slow running, with speed saved for later. Arnulf's autumn is a good argument for the other reading: build the aerobic engine properly and the short efforts get faster on their own, long before anyone asks them to.
Open his calendar and the sessions are plain, and in German, because IntervalCoach coaches him in German: Grundlagenausdauer, Ausdauerlauf, Regenerationslauf. One quality session a week, everything else easy, on a Polarized model. Since late August he has completed 109 of the 121 sessions his plan put in front of him.
The load built in steps. In late September his weeks sat between roughly 420 and 610 TSS. At the turn of November he ran back-to-back weeks of 932 and 833, then took the planned recovery week at 577, almost exactly the 582 the plan asked for. His fitness (CTL) has gone from 59 in late August to 93 today.
He is carrying fatigue right now, and that is the honest context for the 5K. His form (TSB) sits around -31 at the start of a new loading block, and the plan has adapted his sessions 44 times so far, most of them swaps that moved a session to a better day. He has the coach set to show him a change and ask before it rewrites his day. A best time set in the middle of that, rather than off a taper, says more about the engine than a rested one would.
He described his use of it simply:
I use IntervalCoach every day. It means I no longer have to build my training plan in Intervals.icu myself, and I have all my training data clearly in one place.
A week in Courmayeur
One week in late October in his calendar was nothing but trail running and hiking: more than 18 hours on his feet, the closest thing in his autumn to the race itself. The reason sits in the town where TOR130 finishes:
The training block at the end of October, when I only hiked and ran, came out of a family holiday in Courmayeur. I wanted to look at some sections of the course. I am especially proud that I got into the top 10 on an uphill KOM. I am actually prouder of that than of my 5K best from last week.
That last sentence is the whole athlete in one line. The 5K is the number that opened this post, and it is the number that is easy to measure. The one he cares about is a climb in the Aosta Valley, on ground he will see again next September with a lot more kilometres in his legs.
Where the bike miles come from
His calendar carries a lot of riding for a runner, and there is a straightforward reason: this year he raced two ultra-distance cycling events, one in France and one in Austria. The riding has a second job too. In his words:
It also takes some load off my body and brings variety into my training.
You can see the summer in his numbers. His cycling eFTP peaked at just over 300 W in July, in the season of those two races, and has eased to 293 W now. That is what you expect once riding moves from the main event to the supporting act: the watts are holding, the miles have moved to his feet.
The other supporting act is the gym. He has trained strength several times a week for years, and he runs it as a circuit rather than in sets with rests:
I always combine leg/pull or leg/push. That means I do not take classic rest between sets, I alternate between the exercises on the machines.
His files this week show it: two Krafttraining sessions on back-to-back mornings, logged alongside his runs and an easy ride.
What the score noticed
| Week | Total | Fitness Base |
|---|---|---|
| Aug 23 | 65 | 67 |
| Sep 13 | 75 | 68 |
| Oct 30 | 77 | 71 |
| Nov 20 | 74 | 76 |
Fitness Base has climbed from 67 to 76 through the autumn, most of it in the last few weeks, while the total moved with the rhythm of the plan: up through the big weeks, down through the recovery week, back up as the new block starts. His consistency subscore has not dropped below 77 since the end of August: he turns up for the easy days as reliably as the hard ones.
The ten months ahead
The plan now runs on four-week cycles through the winter, three loading weeks and a lighter one, with the loading weeks set at roughly 870 to 960 TSS. The climbing-specific work, the long days and the nights, are still ahead. So is the part that matters most to him: finishing TOR130 well enough to stand at the start of TOR330 a year later.
Here is the prediction. The next time his plan asks for a running time trial, I expect his 5K to start with a 22. If it does, that is the base block doing its job long before the mountains test it.
You can see your own version of this on your Analytics page. And if you would like to be featured in a future spotlight, reply to any IntervalCoach email and say so.