Running
Every running watch gives you the same zones. They're wrong for almost everyone.
Current methods use fixed %HRmax zones and age-predicted maximum heart rate. Two runners with identical HRmax but different endurance levels get identical zones—even though their physiology is completely different.
The Problem
Generic zones ignore what makes you different.
The entire running wearable industry uses the same approach: estimate HRmax from age, divide into five fixed-percentage zones, and call it personalized. But endurance—the single most important variable in aerobic performance—is never measured. Two runners with a HRmax of 190 bpm get identical zones, even if one can hold threshold for 25 minutes and the other for 70.
Age-predicted HRmax
The standard formula (220 − age) has a standard error of ±10–12 bpm. For a 35-year-old, true HRmax could be anywhere from 173 to 197 bpm. Every zone built on this estimate inherits the error.
Fixed percentage zones
Dividing HRmax into fixed bands (e.g., 60–70%, 70–80%) assumes everyone's thresholds fall at the same relative intensity. They don't. Endurance shifts where thresholds actually occur.
The Solution
TrueZone individualizes everything from 3 parameters.
Endurance (E), maximum speed (Vmax), and HR peak (P)—extracted from ordinary running data using Bayesian accumulation. No lab test, no maximal effort, no calibration run. From these three parameters, every zone boundary, threshold, and race prediction is derived individually.
High endurance runner
Can hold threshold pace for over an hour. Heart rate rises slowly during steady runs. Training zones sit higher on the intensity scale than generic models predict.
Speed-dominant runner
Fatigues faster at threshold. Heart rate rises quickly during sustained effort. Training zones sit lower than generic models suggest. Same HRmax, completely different profile.
Recreational runner
Near the population average. Generic zones happen to be roughly correct here, but only by coincidence. Move a few points in either direction and the error grows.
Bayesian Learning
The model learns who you are.
The trend line shows the Bayesian consensus for endurance (E) across sessions, while the shaded region represents the confidence interval—narrowing as evidence accumulates. Within 10–15 sessions, the model converges on a stable physiological profile from ordinary training runs.
Try It
Predict your race times.
Adjust endurance and max speed to see predicted times across every distance from 100 m to the marathon. Click any row to fix a known race time, then explore how endurance and speed interact—and why two runners with the same VO₂max can have very different marathon times.
| Distance | Pace (min/km) | Time |
|---|---|---|
| Marathon | 5:24 | 3:48:21 |
| HM | 4:59 | 1:45:28 |
| 10K | 4:35 | 45:57 |
| 5K | 4:15 | 21:15 |
| 3K | 4:00 | 12:01 |
| 1500 m | 3:41 | 5:32 |
| 800 m | 3:23 | 2:43 |
| 400 m | 3:01 | 1:12.5 |
| 200 m | 2:32 | 30.44 |
| 100 m | 2:22 | 14.30 |
Click any row to fix a known race time. Then adjust the sliders to match your other times.
Exercise thresholds
Exercise thresholds reflect transitions between muscle fiber types during exercise. Use the sliders to see how endurance affects threshold alignment.
| T | Speed | Pace | % of T5 |
|---|---|---|---|
| T1 | 10.0 km/h | 6:00 | 33% |
| T2 | 15.0 km/h | 3:59 | 50% |
| T3 | 20.0 km/h | 3:00 | 67% |
| T4 | 25.0 km/h | 2:24 | 83% |
| T5 | 30.0 km/h | 1:59 | 100% |
Training paces
Optimal training paces and heart rates derived from your parameters. Set your max heart rate for personalized heart rate targets.
| Training pace | Speed | Pace | Heart rate |
|---|---|---|---|
| Easy pace | 10.0 km/h | 6:00 | 106 bpm |
| Steady pace | 12.5 km/h | 4:48 | 132 bpm |
| Endurance training pace | 13.3 km/h | 4:30 | 141 bpm |
| 60 minute pace | 14.2 km/h | 4:13 | 150 bpm |
| Lactate training pace | 15.0 km/h | 3:59 | 159 bpm |
| 30 minute pace | 15.4 km/h | 3:53 | 159 bpm |
| Tempo pace | 16.3 km/h | 3:41 | 159 bpm |
| VO2max pace | 17.5 km/h | 3:25 | 159 bpm |
Validation
Matched Apple's neural network with 3 parameters.
Apple trained a hybrid ODE + deep neural network on 270,000 runs from 7,465 users. TrueZone's pure physiological model, using three interpretable parameters and no global training data, matches their accuracy on the same prediction task.
Endomondo (large-scale)
UCSD FitRec, CC-BY-NC
Endomondo (benchmark subset)
Apple comparison cohort
A 3-parameter physiological model with no training data achieves the same heart rate prediction accuracy as a neural network trained on a quarter million runs. The difference: every TrueZone parameter is physiologically meaningful, trackable over time, and derives a full fitness profile—not just a heart rate prediction.
Bring real running analytics to your platform.
Replace generic zones and age-predicted HRmax with individualized physiology. TrueZone is SDK and API ready today.