# Aerobic Decoupling

Canonical HTML: https://effortsignal.com/learn/aerobic-decoupling/

Last content update: 2026-09-10

Aerobic decoupling compares output per heartbeat in the first and second halves of suitable recordings.

## Formula and sport-specific output

```text
EF1 = first-half output / first-half average HR
EF2 = second-half output / second-half average HR
Decoupling (%) = (1 − EF2 / EF1) × 100
```

Cycling uses each half's Weighted Power as output. Running uses mean speed in metres per second, not time-per-kilometre pace. The split is halfway through the compressed one-second timer timeline. Positive decoupling means less output per heartbeat in the second half; negative values can occur.

## Eligibility

Cycling needs at least 20 minutes, 80% power coverage, 90% HR coverage, configured power zones, and at least half of classified power samples in zones 1–3. Each half must have usable weighted power and positive mean HR.

Running requires at least 20 minutes, 80% speed coverage and 90% HR coverage. It must be an indoor run or have at least 80% of valid grade values within ±2%. Terrain-sensitive outdoor runs can retain their speed/HR efficiency while decoupling remains unavailable.

## Verified example

A suitable 20-minute ride has 200 W Weighted Power in each half. Mean HR rises from 140 to 147 bpm. EF1 = 1.428571 and EF2 = 1.360544 W/bpm, giving decoupling 4.761905%. The same proportion holds for constant running speed with those heart rates when the running eligibility gates pass.

## Interpretation

Heat, hydration, terrain, pacing, sensor quality and the chosen half split can affect this value. It is not a direct aerobic-fitness measurement or a fixed diagnostic threshold. Changing applicable zones may change cycling eligibility even when samples are unchanged. Open [Activity Analysis](/features/activity-analysis/#charts) and inspect the recording before comparing its percentage with another session.

## Sources and implementation scope

[Bourdon and colleagues (2017), monitoring athlete training loads](https://pubmed.ncbi.nlm.nih.gov/28463642/) provides background, not a specification of every EffortSignal rule. The formulas, windows, quality gates and examples above describe EffortSignal 0.1.13, checked September 10, 2026. Product-specific thresholds and rounding should not be attributed to the background source.