PREDICTION METHODS
How sensitive is a race prediction to its input?
A precise-looking prediction can be sensitive to a small change in the input. Check what the model knows before giving the final seconds too much weight.
Start with the explicit web formula
The Runome web predictor uses a basic Riegel-style extrapolation: new time = recorded time × (new distance ÷ recorded distance)1.06. Its inputs are a distance and a time. It does not read your training history or adjust itself to an individual runner's preparation.
This public calculator is distinct from Runome's in-app race-prediction engine. A similar-looking answer does not mean the two methods used the same evidence. Keep the method name with any result you save.
Change a 5K result by one minute
The examples below use invented recent results and the same exponent in both columns.
| Target distance | From 25:00 for 5K | From 26:00 for 5K |
|---|---|---|
| 10K | 52:07 | 54:12 |
| Half marathon | 1:55:00 | 1:59:36 |
| Marathon | 3:59:47 | 4:09:22 |
The second input is 4% longer, so every unrounded output is also 4% longer. The absolute difference grows as the predicted time grows. Displaying seconds does not imply that the model can establish an outcome to the nearest second.
Distance increases the assumptions
A short race and a marathon require different preparation. In a study of recreational endurance runners, the common Riegel approach underestimated marathon times for many participants. That finding is a reason to inspect the limits of extrapolation, not a fixed correction to apply to every runner.
Check whether the input was a recent, relevant race result or a different kind of effort. An easy training outing, a shortened distance or a moving-time total with substantial stopped time answers a different question. The formula cannot detect those distinctions from the two entered numbers.
Use the estimate as one reference
Try the race-time predictor with more than one relevant input and inspect the spread. Keep the source dates and distances visible. In Runome, use the available race-readiness context and actual activity history rather than treating the simple web result as a personalized guarantee.
For a marathon, review recorded volume preparation separately. Once a target is chosen, the splits tool can divide that target across the distance without claiming to validate it.