Course difficulty, finish rates and pacing, measured from each race’s own timing splits. Open a race for its profile, where the work sits and who finished.
Every course is converted to its flat equivalent. The official GPX is resampled to the same 80-metre grid for every race, and each stretch is priced with an energy-cost curve corrected for descents and fitted to real race splits (how it was measured). The result is effort-km; the effort factor is effort-km per kilometre — how much harder than flat the course is on average. D+ is measured from the same profile, so it can differ from the organiser’s figure.
What is left after the terrain is fade: how much the field’s pace drifts over the race against each runner’s own average, measured over the effort rather than the kilometres. Aid-station time is subtracted wherever the timing mats record it. Nothing is simulated and nothing is averaged across races: each edition stands on its own data.
Effort-km is the course converted to flat-equivalent kilometres: climbs and steep descents cost more than flat ground. The effort factor is effort-km per kilometre, so ×1.50 means the course is on average half again as hard as flat. It puts a 60 km and a 170 km course on the same scale.
From the races themselves. The curve starts from laboratory energy cost and adds a descent penalty fitted to clean race stretches, then is tested on courses it was not fitted to. The laboratory says a 20 % descent costs half of flat ground; the races say it costs more than flat. What the profile costs tells the whole story.
On the course alone. The three courses closest in effort-km are shown side by side, with effort, climbing, steepness, the hardest 5 km and both profiles on the same effort axis. Weather, cutoffs and the field are not part of it.
Elevation gain has no single true value — it depends on how finely the course is measured. We resample every official GPX to the same 80-metre grid, so all races are measured identically. Sample any finer and measurement noise starts to show through, by an amount that depends on how each file was recorded. Organisers each use their own method — some publish a more smoothed figure than ours, some a less smoothed one — so the two can differ in either direction.
Published results and timing splits from each event, plus the UTMB Index where the timing data includes it. No names are shown: the field appears as medians and percentiles.