Grade-adjusted pace on trails, and where it breaks

Grade-adjusted pace is the number your platform shows next to your real pace, claiming to know what your hilly run would have been worth on flat ground. Trail runners have a complicated relationship with it. It flatters your climbs, judges your descents, and every so often produces a number so strange that you stop trusting it entirely. Both the trust and the distrust are earned, and it helps to know exactly what the number is doing.

What GAP actually claims

GAP answers a narrow question: at what flat-ground pace would you burn energy at the same rate you burned it on this slope? Because running uphill costs more energy per meter than level running, an uphill GAP is faster than your actual pace. Gentle downhills cost less, so their GAP is slower.

The claim is about energy, not equivalence. GAP does not say you could have run that pace on the flat, or that two runs with equal GAP were equally hard on your body. Climbing and descending load the legs in different ways that a single pace number cannot carry.

Where the models come from

The classical model comes from laboratory physiology. Minetti and colleagues measured the oxygen cost of running on treadmill grades from 45% downhill to 45% uphill and fitted a curve to the results. Level running cost about 3.6 joules per kilogram per meter; the steepest uphill grades cost over five times as much; and the cheapest running of all sat near a 20% descent, where gravity does much of the work before braking costs take over again on steeper drops.

Strava took a different route. Its engineering team originally used a lab-derived heart-rate approach and later rebuilt GAP from its own data: hundreds of millions of runs, comparing pace and heart rate across grades at scale. The two approaches broadly agree on moderate slopes and diverge at steep ones; the redesign notably reduced how generously steep downhill running was credited.

Our free GAP and climb calculator implements the Minetti curve, shows its work, and refuses to extrapolate beyond the grades the study measured.

Where GAP breaks on trails

Every GAP model corrects for exactly one variable: gradient. Trails vary in several more, and each one it cannot see gets silently priced as fitness.

• Surface and technicality. A rooty, rocky 10% climb and a smooth fire-road 10% climb get identical adjustments. Strava's own documentation notes that GAP does not account for the technical difficulty or condition of the terrain. The model assumes you can use the grade credit; on technical descents, footing decides your speed long before energy does.
• Steep-grade extrapolation. The lab data stops at 45%, and models are least reliable exactly where trail races get interesting. Hiking a 35% grade also has different mechanics from running it, and the regression is a running curve.
• The average-runner assumption. The adjustment curve is one curve. Real runners differ in uphill economy and descending skill, so the same model flatters some and shortchanges others on identical ground.
• Conditions. Heat, mud and altitude slow you without touching the grade. GAP passes those losses straight through as if you had a bad day. We wrote about keeping conditions on the record in how to measure trail running progress.

How to use it anyway

Used with its limits in mind, GAP earns its place. It is good for sanity-checking effort on climbs, for comparing hilly weeks against flat ones in rough strokes, and for explaining to a road-running friend what a 9:00 summit kilometer actually was. A hill repeat is the model at its best: a short, consistent grade, where seeing that a 5:00 per kilometer rep up 8.4% prices out near 3:15 flat is genuinely clarifying. It is poor for comparing technical trails against smooth ones, for judging descents, and for any comparison where a real number on real ground is available.

Because that is the quiet lesson: the honest alternative to adjusting a pace is comparing it against your own history on the same ground, where no model has to guess what the terrain cost. That comparison is the one Trailbuddy is being built around, with GAP kept in its proper place: a well-informed estimate, and a genuinely useful one, so long as nobody mistakes it for a measurement.