137 Hours 50 Minutes in Aosta: Two Vietnamese Runners and the Measurement Problem of Tor des Géants

The clock stopped at 137 hours and 50 minutes. Luu Duc Toan finished with exactly 12 hours and 10 minutes left inside the 150-hour budget that Tor des Geants organisers grant to finishers. Le Thi Hang crossed roughly 50 minutes later, at 138 hours 40 minutes. Neither had to sprint against the clock. Nor did either have enough margin to absorb a major incident had one struck at kilometre 250.

This is the first number I logged when the results came in. A runner finishing a race with roughly a 60 percent completion rate is already notable. But a 12-hour cushion against the cut-off says something else: they finished in control, not in survival mode hanging on at the final hour.

Context: A race with no half-time

Tor des Geants takes place in Italy's Aosta Valley and has run since 2026. The main distance covers 330 km of mountain trail, staged as a single continuous effort with no fixed daily segments and no mandated rest. Athletes manage their own running, eating, sleeping and weather exposure across the whole journey.

The course accumulates more than 24,000 metres of climbing and crosses more than 20 passes above 2,000 metres of altitude. Seven large support stations sit roughly 50 km apart. The finish cut-off is 150 hours. This year's race window fell between 13 and 19 September. The event's completion rate hovers around 60 percent. In other words, for every five starters, about two do not reach the finish. That is the single most important figure for valuing any result here, because it turns finishing from an assumption into a minority outcome.

Before Tor des Geants, both Vietnamese runners had built substantial long-distance bases. Le Thi Hang had completed Prenn Trail 250 km, Son La Ultra Trail 100 km and VMM 100 km. Luu Duc Toan's record includes UTMB, Mantra 116 km, ULJU Trail 9 Peaks 100 km and Mt Fuji 100 km. Vietnam had three entrants this year; one, a France-based runner, did not finish.

The evidence chain: the two-times gap is the central number

The men's winner was Jia Ju Zhao of China in 65:55:22. The women's winner was Katharina Hartmuth of Germany in 67:10:09. The two Vietnamese runners finished in roughly double the leaders' elapsed time. The gap between the winner and this pair exceeded 70 hours.

That number needs to sit in the right place. In a race with a 150-hour cut-off, a 70-hour spread across the finishing field is normal, not a sign of organisational or personal abnormality. It reflects the stratified structure of ultra trail: a specialised elite tier racing at a completely different level, with most finishers being high-level amateurs.

More instructive is the time cushion. Toan finished with 12 hours 10 minutes to spare. Hang had 11 hours 20 minutes. Across a race lasting nearly six continuous days and nights, that margin equals roughly 8 percent of total elapsed time. Enough to avoid cut-off anxiety, but not enough to absorb two consecutive problems, such as a minor injury plus an unplanned long sleep.

The average speed derived from the data is worth recording too. 330 km divided by 137 hours 50 minutes gives about 2.4 km per hour. This is not a speed metric in the flat-road sense. With more than 24,000 metres of climbing, where every kilometre includes ascent and descent, it is a pure measure of endurance under terrain load.

One data point that most coverage overlooked: the gap between Hang and Toan is only about 50 minutes across a journey of nearly 138 hours. A sub-1 percent difference between a female and a male athlete at such an extreme distance is very tight. It suggests two possibilities: either the two have comparable physical bases, or they coordinated pacing and supported each other for much of the course. Public data cannot distinguish between these hypotheses.

Hai Phong taught me this: stars are not on the shirt, they are in the numbers. In 2026 I once brought a PPDA table for a young midfielder named Vu Minh Hieu into the coaching room. His average pressing index was 6.8, the highest in the academy, but he was overlooked because of his modest frame. In the match against Ha Noi FC on V.League round 17 that year, Minh Hieu won the ball 14 times, provided one assist, and the team won 2-1. The lesson was not that data is always right, but that data forces people to re-examine their assumptions.

Applying the same principle here: both Vietnamese runners' records show a deliberate distance progression. From 100 km to 116 km, to 250 km, then to 330 km. This is not a random leap. It is an accumulation chain, and the Aosta result is a logical continuation of it.

Toan once compared Tor des Geants to UTMB with a striking line: this race is many times harder and more challenging. That is a first-person technical judgment. It carries reference value because he had completed UTMB before, but it remains subjective perception, not measurement. My experience tracking ultra races shows such assessments are usually right in direction and wrong in magnitude.

The contrarian angle: finishing is not the same as competing

Numbers are a mirror. Most of the market looks into it and only sees itself.

The common framing places the two Vietnamese runners alongside the winners' names, accompanied by dramatic quotes about cliffside paths with no safety ropes. That framing is factually accurate but easily creates an illusion of competitive tier. The reports do not state the pair's finishing positions. With roughly double the leaders' time, their placing almost certainly sits in the back half of the finisher cohort.

The biggest blind spot in all the public data is the absence of split times. No checkpoint breakdown, no per-station data, no analysis of pacing distribution. That means we know they finished, but not how. A runner who held even pace for 138 hours and one who collapsed at km 200 then recovered to crawl home share the same finish time but are entirely different stories in terms of capability.

The number also needs the right reference frame. The finisher cohort at Tor des Geants is not the fast group but the durable group. A result here measures endurance under extreme conditions, not peak speed. Comparing this figure to a road marathon or a track performance is a methodological error from the first step.

Risk also deserves parity with achievement. In 2026 the race recorded a fatality when a participant slipped and fell on a mountain section. Toan describes narrow cliffside paths without safety ropes. That detail matters more than any time figure, because it establishes that finishing here depends on both capability and risk management, and neither can be inferred from elapsed time alone.

Signals for the next cycle

People call me a data monk. A monk does not need a cathedral, only the truth.

Three signals I will track over the next 12 to 24 months. First, whether these two runners repeat results at another international ultra. One finish is a milestone; two consecutive finishes at different races is a stable capability level.

Second, split data. When public checkpoint tables appear, pacing distribution quality becomes a far more analysable metric than total elapsed time. That is when we can assess real capability rather than merely record a result.

Third, the domestic talent pipeline. The emergence of multiple races at 100 km and above, from Prenn Trail to Son La Ultra Trail and VMM, indicates a domestic ultra trail market taking shape. A runner finishing in Aosta only has durable value when a sufficiently dense race calendar sits behind him or her to produce the next one.

Two Vietnamese runners finishing a race with a 40 percent failure rate is a real, verifiable result that deserves proportionate recognition. The ball rolls in one direction, but data can see in all of them. The remaining question is not how far they went, but how much more data we will have next time to measure it more precisely.