Race time predictor
A recent race becomes equivalent times at 5K, 10K, half marathon and marathon — with the model named, the assumption stated, and the distance gap marked on every row.
A race you have actually run, recently, at full effort. A time you hope to run predicts nothing.
| Distance | Equivalent | Pace | Distance gap |
|---|---|---|---|
| 5K | 23:59 | 7:43/mi · 4:48/km | Shorter |
| 10K | 50:00 · your result | 8:03/mi · 5:00/km | Same distance |
| Half marathon | 1:50:19 | 8:25/mi · 5:14/km | A step up |
| Marathon | 3:50:01 | 8:46/mi · 5:27/km | A long way up |
- Shorter
- Predicting downward from a longer race. Usually the safer direction, because endurance you have already demonstrated is not in question.
- Same distance
- Your own result, unchanged.
- A step up
- A modest step. The assumption of comparable preparation is doing less work here than it is further down the table.
- A long way up
- The assumption is doing a lot of work here. This number describes a runner whose endurance training matches their speed — check that describes you before racing on it.
These are equivalent performances, not forecasts. Riegel’s model answers one question: if your training supported this distance as well as it supports the one you raced, what would you run? It assumes comparable preparation across the distances, and that assumption is where it breaks — a runner with a fast 5K and a longest run of eight miles has a marathon prediction that is arithmetically correct and practically useless.
We do not publish a confidence interval or an accuracy percentage for these, because we have no source that supports one. What we can tell you is how far the model is asking of the assumption, which is what the last column marks.
Model and rounding are described in full on how we calculate. Source: Peter S. Riegel, “Athletic Records and Human Endurance”, American Scientist 69(3), 1981.
How to use it
Give it a race you have actually run, recently, at full effort. It returns what you would run at the other distances if your training supported them equally well. That condition is the whole of the tool — the arithmetic is three lines and every calculator on the internet has it.
Predicting downward, from a longer race to a shorter one, is the safer direction: endurance you have already demonstrated is not in question. Predicting a marathon from a 5K is the weakest thing the model does, because the number says nothing about whether you can cover the distance.
Common questions
How accurate is a race time predictor?
It depends entirely on something the model cannot see: whether your training prepared you for the target distance as well as it prepared you for the one you raced. Where that holds, equivalent-performance models are useful. Where it does not — a fast 5K and a longest run of eight miles — the marathon number is arithmetically correct and practically meaningless. Anyone quoting you a percentage accuracy is quoting a number they cannot support.
Why does it predict a faster marathon than I can run?
Almost always because the prediction came from a short race. The model assumes comparable preparation across distances, and most runners are better prepared for 5K than for a marathon. Predict from your half marathon instead if you have one — the assumption has much less work to do.
What is Riegel’s exponent?
1.06. It sets how much slower you get as distance rises, and it comes from Riegel’s 1981 analysis of athletic records. We use it unmodified and say so, rather than tuning it privately and presenting the result as fact.
How we calculate
Riegel’s model: T₂ = T₁ × (D₂ / D₁)^1.06, with distances exact — a marathon is 42,195 metres, not 26.2 miles. The exponent comes from Peter S. Riegel, “Athletic Records and Human Endurance”, American Scientist 69(3), 1981.
We publish no confidence interval or accuracy percentage, because we have no source that supports one. What we show instead is how far the model is extrapolating, which is a fact about the distance ratio rather than a claim about you.