Race time predictors: what they know and what they assume

Enter a 5K time into a race predictor and it will tell you your marathon time to the second. The confidence of that output is not matched by the confidence you should place in it.

What the formula does

Most predictors use Riegel’s formula, published in 1981. It scales a known time by the ratio of distances raised to the power 1.06:

T₂ = T₁ × (D₂ ÷ D₁)^1.06

The exponent is the whole idea. If it were 1.0, the formula would assume you hold the same pace regardless of distance, which is obviously wrong. Setting it slightly above 1 encodes the fact that pace degrades as distance grows.

Where 1.06 came from matters, though, and almost nobody mentions it. Riegel fitted it to world records, across distances from 100 metres to 100 miles. He never presented it as a single universal number — he fitted separate values for other sports, and broke the running exponent out by age and sex group. ¹ The 1.06 in every race calculator on the internet is a constant derived from the fastest humans alive, applied to everybody else.

That is still genuinely useful. It is a compact, empirically grounded description of how endurance performance falls off with distance, and for moderate extrapolations it works well.

The assumption it cannot verify

The formula’s central assumption is that you are equally well trained for both distances.

It has no way of checking this. It sees a time and a distance and nothing else. It does not know whether you have done a single run over ten kilometres, whether you have practised fuelling, or whether your longest run this year was the 5K you just entered.

For a 10K predicted from a 5K, that assumption is usually close enough — the training required for the two overlaps heavily. Empirical analysis of recreational runners’ race times has found that predictions degrade as the extrapolation grows, which is exactly what you would expect if the training-equivalence assumption is doing the work.

This is why our calculator flags confidence by distance ratio rather than presenting all predictions identically. A prediction is only as good as the assumption underneath it, and telling you which predictions rest on a shakier assumption is more useful than a uniform row of numbers.

Why the marathon breaks predictions

The marathon is where predictors fail most often and most dramatically, for reasons that sit outside the physiology the formula captures.

Shorter races are limited primarily by aerobic capacity and the ability to sustain effort near threshold. The marathon adds constraints that barely feature at 5K: glycogen availability, fuelling strategy, gastrointestinal tolerance, thermoregulation over hours, and musculoskeletal durability. Large-scale analysis of recreational marathon pacing has documented how commonly runners slow dramatically in the closing stages — a pattern driven by these factors rather than by aerobic fitness. An analysis of more than four million race records found the slowdown is more common in the years around a personal best: 36 per cent of runners hit the wall in the three years before a recent PB, against 23 per cent in earlier years. Pushing closer to your ceiling makes the collapse more likely, not less. ⁴

This is measurable, and the size of it is worth knowing. A study of 2,303 recreational endurance runners found Riegel well calibrated for races up to the half marathon — and then, at the marathon, giving predictions at least ten minutes too fast for half of all runners. A simple model built from a runner’s own prior race times roughly halved the error. ²

The important part is that this is not a problem of extrapolating too far. Predicting a marathon from a half is a short hop by the formula’s own logic, and it is still biased fast. The bias comes from the distance, not the gap. Our calculator therefore never reports high confidence for a marathon prediction, however close your input race is — it tells you the number is a floor on your finish time rather than a target.

A runner with an excellent 5K and no long-run history will not run the predicted marathon time. The prediction describes what their aerobic engine could support; the race is decided by whether the rest of the system can last.

Using splits on race day

Even splits — running each segment at the same pace — are a sound default for most runners at most distances. The most common recreational error is starting too fast, and an even-split plan is a direct guard against it.

Two qualifications. Elite performances are frequently run with a slightly faster second half, suggesting that a marginally conservative opening is not a cost. And terrain, heat and wind all justify deviating: holding goal pace up a long climb spends far more than it saves.

Use the splits table as an anchor for the early kilometres, where discipline matters most, and as a rough check thereafter. If you reach halfway on plan and feeling controlled, the plan was right.

The bottom line

Trust a predictor most when the distances are close and you have trained for both. Treat a marathon predicted from a 5K as a description of your aerobic potential rather than a time you should attempt to run.

And remember what no formula can see: whether you have done the long runs, practised taking on fuel at pace, and rehearsed the effort. Those decide the race. The arithmetic only describes the engine.