Running Time Predictor: How Far One Race Result Stretches
A running time predictor turns one race into times for other distances. Riegel's math, why the marathon estimate drifts 10 minutes, and which input to use.
Kristian Hoffmann
SaaS founder and operator
A running time predictor takes one recent race result and projects your finish time at another distance. Most calculators use Riegel's formula, T2 = T1 × (D2 / D1)^1.06. From a 25:00 5K, that works out to about 52:07 for 10K, 1:55:00 for the half marathon and 3:59:47 for the marathon. The rule for using it: the closer your input race is to the target distance, the more weight the prediction can take.
What the formula actually does with your 5K
Riegel's formula scales your time by the distance ratio raised to an exponent. The exponent, usually 1.06, is the fatigue factor: how much your pace slows each time the distance goes up. At an exponent of 1.00 you would hold your 5K pace for a full marathon. At 1.06, every doubling of distance costs roughly 4% of pace.
Here is the full ladder from a single 25:00 5K. The paces are calculated here from the formula, not taken from a table:
| Target distance | Distance ratio | Predicted time (exponent 1.06) | Implied pace |
|---|---|---|---|
| 5K (input) | 1.00 | 25:00 | 5:00 /km |
| 10K | 2.00 | 52:07 | 5:13 /km |
| Half marathon | 4.22 | 1:55:00 | 5:27 /km |
| Marathon | 8.44 | 3:59:47 | 5:41 /km |
Tools that sit on top of this math look different but do the same scaling. Momentum Sports' race predictor is built for a distance you don't normally race. Garmin describes its race time prediction as an input for decisions on pace, nutrition and what to wear. Which inputs a watch feeds its model, and how it weights them, is for Garmin to document. That is a different question from the formula.
VDOT, the fitness score from Jack Daniels' tables, is the other common engine. It also starts from a race result, but it maps that result to a fitness level instead of applying a single exponent. The Daniels VDOT running calculator explains how the score becomes training paces.
Why the marathon number is the least trustworthy one
The exponent is not a law of nature. It is an average, and runners with little endurance training slow down faster than 1.06 allows. Nudging the exponent shows how much the long-distance prediction depends on that single assumption:
| Exponent | 10K from 25:00 5K | Marathon from 25:00 5K |
|---|---|---|
| 1.06 | 52:07 | 3:59:47 |
| 1.07 | 52:29 | 4:04:57 |
| 1.08 | 52:51 | 4:10:13 |
Moving from 1.06 to 1.08 adds 44 seconds to the 10K prediction. It adds 10 minutes 26 seconds to the marathon. Same runner, same 5K, and the uncertainty grows fourteen-fold because the distance ratio is 8.4 instead of 2.
Use the input closest to your target
Feed the same runner's 1:55:00 half marathon into the formula instead. At 1.06 the marathon still comes out at 3:59:47. At 1.08 it comes out at 4:03:07. The spread between the two exponents shrinks from 10:26 to 3:20, simply because the half is only half the target distance.
That gives a practical ranking of inputs for a marathon prediction:
- A half marathon run in the last eight to ten weeks
- A 10K or 15K from the same period
- A 5K, treated as an optimistic upper bound
The half marathon to marathon predictor goes further into how far apart the common half-to-full methods land. If a 10K is your most recent race, the 10K to marathon predictor compares Riegel with the "5× minus 10" rule.
Where the prediction fails in practice
The typical miss has one of three causes:
- Endurance gap. A fast 5K from a runner doing 25 km a week says little about kilometre 35. The formula assumes the endurance matches the speed.
- Distorted input. A hilly, hot or windy input race makes your fitness look worse than it is. A downhill or short-measured course makes it look better.
- Different target conditions. The prediction describes a neutral day. Course elevation and race-day temperature come on top.
An early warning sign: if your long runs at the predicted marathon pace plus 30–45 s/km already feel hard after 25 km, the prediction is running ahead of your endurance.
Turning a predicted time into a race plan
A predicted time is a ceiling for a well-trained day, not a pace to start at. TrainingFlow computes pacing from VDOT and then adjusts for course elevation and the race's climate baseline. The same order works if you are doing it by hand:
- Predict from your closest recent race. Note the result at both 1.06 and 1.08.
- Pick a goal inside that range. Choose the faster end only if your long runs and weekly volume support it.
- Adjust for the course. Climbs and descents shift splits even when the total stays the same.
- Adjust for weather. A warm race day slows the whole plan, not only the final kilometres.
- Convert to checkpoints. Cumulative clock times at 5 km intervals are easier to check mid-race than a single average pace.
For the 25:00 5K runner, that could mean a marathon target of 4:02–4:05, starting around 5:45 /km, rather than chasing 3:59:47 from the gun.
Using November and December results
Late autumn is when many runners set targets for a spring marathon. A 5K or 10K run in cool November conditions tends to be a clean input, since heat is not distorting it. Two caveats apply. A result from after an autumn marathon may still reflect fatigue from that race. And if the goal race is four or five months away, recalculate from a new race in late winter rather than training to a December number all the way through.
FAQ
How accurate is a running time predictor?
Accuracy depends mainly on the gap between the input and target distances. From 5K to 10K, a change in the fatigue exponent from 1.06 to 1.08 moves the prediction by under a minute. From 5K to marathon, the same change moves it by more than ten minutes. Use the closest recent race you have, and treat long-distance predictions from short races as optimistic.
Is a running time predictor app better than an online calculator?
An app or watch can update its estimate from your recent training instead of a single race, which helps if you rarely race. An online calculator shows its formula and inputs, which makes the result easier to check. Whichever you use, compare its marathon estimate with a Riegel calculation from your most recent half marathon; a large gap tells you which assumption to question.
Can a running time predictor be used for cycling?
Riegel's formula was fitted to running performances, so its exponent and pacing outputs are not calibrated for cycling. Cycling speed depends heavily on terrain, wind, drafting and equipment, which a distance ratio does not capture. For bike legs, use a power- or speed-based model built for cycling, and keep the running predictor for the run.