Mean-maximal power curve (MMP)
For each duration (5s, 1min, 5min, 20min...), the highest average power you actually sustained over any rolling window of that length — an observed peak, never a model.
- A specialist track sprinter’s 60-second mean-maximal power point: ≈700 W sustained for a full minute.
- Covered by mainstream cycling press (UCI.org, Cyclist), not self-reported.
This is the same duration-grid extraction MyThreshold runs over your own power stream — the mean of a rolling window, never an instantaneous spike.
Source: Robert Förstemann’s press-covered effort (UCI.org, Cyclist)
Formula, provenance & citation
- Formula
- For each duration_s: slide a window of duration_s×sample_rate_hz samples across the power stream, skip any window with a gap, take the max window mean.
- Inputs
- Per-second power stream; durations_s grid (1, 5, 10, 15, 30, 60, 120, 300, 480, 600, 1200, 1800, 3600 s, config).
- Provenance
- measured
- Unit
- W
- Method
- Mean-maximal power extraction (power-duration curve)
- Citation
- Allen H., Coggan A. — Training and Racing with a Power Meter, 2nd ed.
- Licensing note
- The extraction technique itself is not licensed software; Strava/intervals.icu/TrainingPeaks paywall the FEATURE (showing you your own curve), not the underlying published concept.
- Implementation pointer
api/backend/backend/pipeline/metrics/curves.py:216 (power_curve). Tests: api/backend/tests/pipeline/test_metrics_curves.py.