Why one race result isn't enough
Most race time calculators take a single result and run it through the Riegel formula (T2 = T1 × (D2/D1)^1.06). That works at the population average, but you are not the population average. A 5K time produced on a hot day off bad sleep is a different data point than a half marathon you nailed in cool weather. Treating them as equally valid throws the prediction off by minutes.
How this predictor weights your data
You enter up to three past results. Each one gets a Riegel projection to your target distance, then the predictions are blended:
- Recency weighting: a race from last month counts more than one from a year ago.
- Distance similarity weighting: a half marathon predicts marathon better than a 5K does.
- Training volume adjustment: under 30 km/week adds a 3% time penalty (insufficient endurance base); over 60 km/week applies a 2% bonus.
The output is a single predicted time plus a confidence score. Confidence is 60% with one race, 80% with two, 95% with three or more across different distances.
Where the predictor breaks down
Marathon and ultra predictions are softer than 5K to half marathon. Pacing, fueling, and terrain dominate beyond 3 hours of effort. The confidence score drops accordingly. Use the prediction as a target band, not a guarantee.
Race Performance Predictor
Enter up to 3 past race results and your training details. The predictor blends multiple data points with recency and distance weighting. More accurate than a single-race Riegel formula.
FAQ
Why is more than one race result better?
A single race result forces the model to extrapolate from one data point. Multiple results at different distances let the algorithm cross-validate predictions and produce a confidence score reflecting how well your performances agree.
How does training volume affect the prediction?
Runners training under 30 km/week get a +3% time penalty (less endurance base), while those over 60 km/week get a -2% bonus. This reflects real-world aerobic development beyond what race results alone show.
Is this accurate for ultramarathons?
Predictions become less reliable beyond marathon distance because ultra performance depends heavily on nutrition, terrain, and pacing strategy. The confidence score will drop accordingly for ultra distances.
Methodology
Core model is the Riegel formula (T2 = T1 × (D2/D1)^1.06) applied per race result, then blended using recency weighting (more recent races count more) and distance similarity weighting (races closer to your target distance are more predictive). Training volume applies a small adjustment factor.
Want to verify the math?
Explore 170+ reference calculators built by engineer-athlete Thomas Prommer. The technical foundation behind our AI.
Browse the Library →