Race Time Predictor
Enter two or more recent races and this fits a power law to your results — not a generic formula — to predict what you'd run at other distances. The more races you add, the better the prediction.
Predicted times
| Distance | Predicted | Pace | 90% range |
|---|
How this calculator works
Race times follow a power law: time grows predictably as distance grows, following the shape
time = a × distanceb
That exponent b is your fatigue factor — how much you slow down as races get longer. Most predictors (including the classic Riegel formula) just assume a fixed value of about 1.06 for everyone. This tool doesn't: it takes logarithms of your distances and times, fits a least-squares line through your results, and reads the fatigue factor off the slope. You get a model of you, not of the average runner.
That's why more races help. With two results the line passes exactly through both points — a perfect fit by definition, which means there's no way to tell whether either race was an off day. From three results onward the model can measure how well your races actually agree, so it reports a confidence range and a fit quality. Every extra race tightens the estimate.
Reading your fatigue factor
- Below ~1.05 — endurance-leaning. You hold pace unusually well as distance climbs; longer races are your strength.
- Around 1.06 — the classic average from Pete Riegel's research, and typical of a well-rounded runner.
- Above ~1.08 — speed-leaning. You fade more than average over distance, which usually points at aerobic base as the thing to build.
Track runners often sit higher and marathoners lower — the factor reflects your current training, not a fixed trait. Rebuild it after a training block and watch it move.
The honest limits
Predictions are most reliable near the distances you actually raced. Feeding in two track races and asking for a marathon is extrapolation, and the marathon in particular depends on fueling, heat, and long-run training that a 5K never tests — no formula can see those. Treat a long extrapolation as a conversation starter, not a goal time.
The model also assumes your races were roughly equally fresh, equally well-executed, and on comparable terrain. A hilly trail race or a heat-wave effort will drag the fit around.
FAQ
How many races do I need?
Two minimum, at different distances. Three or more is meaningfully better: that's the point where the model can cross-check your races against each other, report a confidence range, and stop taking a single bad day at face value.
Why do my two races give a "perfect" fit?
Because any two points define a line exactly. It isn't evidence the model is right — it just has nothing to disagree with. Add a third race and the fit quality becomes a real measurement.
My prediction looks too fast. What gives?
Usually one of three things: you're extrapolating well past your raced distances, one entry was a standout day the model is anchoring on, or the longer race demands (fueling, heat, sheer time on feet) that the math can't see. Add more races closer to your target distance.