What each prediction is actually saying
| METHOD | STRENGTH | MAIN WEAKNESS |
|---|---|---|
| Riegel 1.06 | Simple, famous, easy to reproduce | Derived from performance relationships; often too fast for recreational marathon conversion |
| Daniels VDOT | Connects race performance and training intensities | Equivalent performance assumes you are adequately trained for the target distance |
| Cameron | Correction changes with distance rather than using one exponent | Still based on a general performance curve, not your training log |
| 2× half + 10 min | Easy sanity check from a half marathon | Rule of thumb; unavailable from 5K/10K and not individual |
The original page displayed classic Riegel and then added a fixed distance buffer. That looked reassuring but was not a recognized model. The named methods now remain intact as theoretical equivalents, while the suggested planning range starts from their cautious end and moves slower when the distance jump, weekly volume, long run or training continuity leave a gap.
Formula sources: Riegel uses T₂ = T₁ × (D₂/D₁)^1.06; Daniels VDOT combines a running oxygen-cost curve with the fraction sustainable for a race duration; Cameron uses a distance-specific correction instead of one fixed exponent. You can compare the implementation with the official VDOT calculator and this formula reference. We do not label proprietary watch or app predictions as one of these formulas because their full models are not public.
Why the fast prediction needs skepticism
In a study of 2,303 recreational endurance runners, Riegel was well calibrated through the half marathon but substantially underestimated marathon time; for half the runners it was at least 10 minutes too fast. A later review reported average Riegel error around 10% in a comparable recreational marathon setting. Vickers & Vertosick study · marathon recommender-systems review.
This does not make Riegel useless. It makes it a clean estimate of equivalent speed endurance under its assumptions—not evidence that a fast 5K runner has already built the legs, fuel strategy and accumulated training for 42.195 km.
The prerequisites the formulas cannot see
- Consistent running volume. A recent race proves speed at that distance. It does not prove months of marathon load.
- Long-run durability. Muscular damage, form deterioration and fuelling become materially different late in a marathon.
- Strength and resilience. Strength work may support economy and tissue capacity, but it does not justify subtracting a fixed number of minutes.
- Fuelling and hydration. A predictor assumes energy delivery does not become the limiting system. That has to be practised.
- Health and injury continuity. Interrupted training can leave recent short-race speed intact while eroding marathon readiness.
- Course, weather and execution. Elevation, heat, wind, pacing errors and the taper can easily outweigh a small difference between formulas.
How to turn the range into a race target
Prefer a recent half marathon over a 5K because the extrapolation is smaller. Start from the slower end of the model spread, then ask whether your long runs and weekly training make even that pace familiar. A goal pace should appear repeatedly in controlled marathon-specific work; it should not be selected because one calculator produced an attractive finish time.
The confidence score describes how much useful evidence the calculator has, not the probability that you will hit the displayed time. Race distance, weekly volume, longest run and consistent weeks carry most of the weight. Age and sex are optional, low-weight context because the race result already reflects most differences in current performance; they never trigger a fixed time penalty.
The race-day chart is deliberately a scenario model rather than a probability model. It integrates a progressively slower pace after an assumed fatigue point and, in the severe case, adds possible walking or stopped time. The shaded width reflects the selected assumptions; it is not an empirically calibrated confidence interval.
Frequently asked questions
Why do marathon calculators give different times?
They model endurance decay differently. Riegel uses one power exponent, Daniels VDOT maps equivalent performance through oxygen-cost and duration equations, and Cameron uses a distance-varying correction. None can see your actual marathon preparation from one race result.
Is the Riegel formula too optimistic for a marathon?
It often is for recreational runners extrapolating from shorter races. A study of 2,303 recreational runners found classic Riegel predictions were at least 10 minutes too fast for half of the runners at marathon distance.
Which marathon prediction should I use?
Do not pick the fastest formula by default. Rayvik starts from the cautious end of the model spread, then moves the suggested range slower when your race distance, weekly volume, longest run, training consistency or fuelling leave a gap. The individual model results stay visible, but the training-adjusted range is the practical answer.
Does strength training affect marathon prediction?
A one-race formula does not include strength training. Strength can affect resilience, running economy and the ability to maintain form, but it cannot be converted honestly into a fixed number of minutes by this calculator.
Do age and sex change the prediction?
They are optional context and add only a small amount of confidence when provided. A recent race result already captures most differences in current performance, so this calculator does not add a fixed age- or sex-based time penalty.
Is the fade simulation a probability forecast?
No. It is a transparent stress test that applies an assumed fade onset, finishing-pace slowdown and possible walking or stopped time. Use it to understand consequence and range, not as a calibrated probability of what will happen.