LyonSaintéLyon Finish Time Predictor

The predictor uses the Riegel model with TrailMath's km-effort elevation adjustment, applied to the full 156.74 km and 4,239 m of climbing. The estimate it gives is ELAPSED time from Lyon back to Lyon, including the rest at Saint-Étienne - which is the number the 31-hour limit is measured against.

The organizer's published times are not finish times. Only the return leg is ranked, so the 2025 winning chrono of 7:05:05 covers 78 km rather than 156. Elwan Mehl's actual elapsed time was about 21 hours 36 minutes. Comparing your estimate to the published ranking will make you look four hours faster than you are.

Budget the stop. The model predicts moving time on the course. Add your planned rest at Saint-Étienne - most finishers take 45 minutes to two hours - and check the total against the 22:30 outbound barrier and the 00:40 departure barrier.

Attrition is the honest guide. 420 of 566 starters finished in 2025, a 74% rate, against 89% for the same terrain run once on the 80 km. The difference is the night half, run on legs that already have a full day in them.

Once you have a predicted finish time, use the LyonSaintéLyon Aid Station Planner to split it across both legs.

The classic Riegel formula (T2 = T1 × (D2/D1)^1.06) was derived from road race data. It assumes distance is the only factor in fatigue - ignoring elevation, terrain friction, and ultra-specific physiological degradation. The exponent is the whole model, and 1.06 is a road number.

This tool used to patch that with a stepped ladder - 1.06 up to marathon, rising to 1.15 beyond 100 km - stacked on top of ITRA km-effort and a terrain multiplier. Three heuristics on a road model. Nobody had fitted any of them.

What it does now. Your result becomes a performance index on a 0-1000 scale, and that index is mapped onto the target course - the same calibrated model as the ITRA / UTMB Index Estimator, so the two tools cannot contradict each other. The model is fitted by ordinary least squares to public UTMB Index profiles, each anchor recorded with its source and the date it was read. It is a transparent approximation of a proprietary system, not an official ITRA or UTMB number.

It is still a power law - that is worth being honest about. Solving the fit for time gives T ∝ km-effort^1.24. So the change is not that we abandoned Riegel; it is that the exponent is now fitted to public trail results instead of assumed from road ones. That single change is worth over three hours on a UTMB projection, in the direction of what the field actually runs.

Where it stops. Below 50 km-effort - roughly a road marathon - the calibrated model has too few verified results to score honestly, so the tool scales with that same fitted exponent instead and labels the answer as the rougher estimate it is. There is no terrain setting, because the model has no terrain variable: a global average of technicality is inside the curve, but nothing in it can distinguish one course's ground from another's. There is no altitude, heat or night term either. Those are named limitations, not hidden ones - run your number through the Race Conditions Simulator before you commit to it.

km-effort, twice. The index is scored on ITRA's public definition, distance + gain/100, which ignores descent. The km-effort figure shown beside your prediction is TrailMath's own, distance + gain/100 + loss/150, which does not - steep descent causes eccentric muscle damage (Minetti, 2002) that is not free. For UTMB those are 274 and 341. Both are labelled on screen so neither is mistaken for the other.

ITRA. km-effort formula. International Trail Running Association. Riegel PS. (1977). Athletic records and human endurance. American Scientist. Minetti AE et al. (2002). J Appl Physiol. 93(3):1039-46. Hoffman MD. (2014). Pacing by winners of a 161-km mountain ultramarathon. Int J Sports Physiol Perform. Millet GY et al. (2011). Neuromuscular consequences of an extreme mountain ultra-marathon. PLoS ONE - a pre/post study of neuromuscular alteration, which is what it is cited for here.