FTP or 5-Minute Power: Which Number Matters for Segments?
For a 90-second segment your FTP is nearly irrelevant; for a 20-minute climb it is the single most important number. The right metric follows the duration.
Duration determines the metric
Your power-duration curve falls with time. A prediction needs the point on the curve that matches the expected segment duration — not the number that sounds best.
- under 1 minute: anaerobic power, often 150 to 200 percent of FTP
- 1 to 3 minutes: around 130 to 160 percent
- 3 to 6 minutes: around 115 to 135 percent (VO2max range)
- 10 to 20 minutes: around 100 to 110 percent
The classic mistake
Feeding a 5-minute best into a model for a 3-minute segment badly underestimates the time. Conversely, FTP produces far too pessimistic predictions for short efforts.
More importantly: lab-test bests are not what you can produce at the end of a training ride, at 8 °C, into a breeze, on a farm track.
Reality beats arithmetic
If you already hold a best time on a segment, that time is the most reliable information about your actual capability there. Power data should support the prediction, not overwrite it.
KomPredict therefore anchors on your existing best time and only applies the weather difference on top. Physics takes over only where no history exists.
Getting the numbers right
Test short and medium durations separately and rested: 1 minute, 5 minutes, 20 minutes, on different days. Refresh them every six to eight weeks or predictions drift.
In short
Use the power that matches the segment duration — and always let existing best times take priority.
Frequently asked questions
Do I need a power meter?
No. Existing segment times work very well. Power data mainly helps on new segments where you have no history.
What about W' or anaerobic capacity?
It dominates below two minutes. Predictions from steady-state power values are inherently unreliable there.