Race time predictor
Most predictors take one race and apply a fixed exponent from 1977. Give us two or three and we fit your fatigue curve — how much your pace actually fades as the distance grows — and show you what four different methods make of it.
Your answer appears here
Put in one race time and you get a prediction. Two or three and we fit your own fatigue curve instead of a fixed one.
What a predictor cannot know
- It assumes you have trained for the distance. A marathon prediction from a 5k is arithmetic, not a forecast — nothing here knows whether you have ever run 30 km.
- It assumes a flat course and a decent day. Heat, hills and wind are not in these numbers.
- It is a snapshot. Run this again in six weeks and the answer should have moved. Nothing here notices that it has.
A number is easy. Keeping it true as you change is the work.
This page answers for the results you type in today. Noticing when your fitness has moved, and keeping the plan honest as it does, is the part we are building.
Join the waiting listPrivate pre-beta · one email when it opens, nothing else
How the four methods differ
Riegel (1977) says time grows with distance to the power 1.06, from one good effort. VDOT turns a result into a VO₂max-equivalent score and reads every distance off Daniels’ race-pace model. Cameron (1998) is an empirical curve for how sustainable pace fades with distance. The power law keeps Riegel’s shape but fits the exponent to you from your own results, weighting recent ones more.
Why the ranges are different widths
Inside the distances you have actually raced, the range is about two and a half percent either way. Beyond them it widens, up to five percent, because the curve is being extended rather than read. Fewer or older results widen it too.
Why a real result can set the number
If you have raced close to a distance recently and were faster than the fitted curve, the prediction will not be slower than what you have already run. When that happens the row says so.