ANALYSIS WINDOWS
Why warm-up minutes can change a drift calculation
Changing the analysis window changes the question. A lower drift result after removing early minutes is not a change in the recorded run.
Make the window explicit
A first-half versus second-half calculation needs a beginning, an end and a rule for the middle. If you remove the first ten minutes, the halves must be defined again over the remaining period. Otherwise, two results with the same label may cover different amounts of time.
For this worked example, assume constant speed, evenly sampled heart rate and no stops. We calculate simple mean-heart-rate change, not an app-specific fitted slope. The simplified recording is invented to isolate the effect of the window.
One recording, two results
The first ten minutes average 130 bpm; the remaining 30 minutes average 150 bpm. The complete run lasts 40 minutes.
| Window | Early-half average | Late-half average | Simple HR change |
|---|---|---|---|
| All 40 minutes | 140 bpm | 150 bpm | 7.14% |
| Final 30 minutes | 150 bpm | 150 bpm | 0.00% |
For the complete recording, the first half contains ten minutes at 130 and ten at 150: (10 × 130 + 10 × 150) ÷ 20 = 140. The second half is 150. Removing the opening ten minutes leaves two 15-minute portions at 150. No measured sample changed; only the included period did.
Do not search for the most flattering start
It can be reasonable to study a defined steady portion after the beginning of an outing. It is less useful to move the boundary repeatedly until the result looks good. Set the rule from the workout structure and use it consistently across comparable runs.
Do not assume every opening ten minutes should be removed. Some runs have no sustained steady section; others contain repeated stops or changing terrain. Where a defensible comparison window is absent, reporting that limitation is more informative than forcing a percentage.
Keep the full activity accessible
When using Runome's detailed charts, retain the complete activity as context even if you study a narrower interval. Record your chosen boundaries, whether the clock excludes pauses and whether both signals cover the same portion.
Use the drift and decoupling definitions to label the calculation correctly. For a structured session, review work and recovery separately. A transparent note about the window lets you repeat the analysis later without confusing selection with improvement.