MODEL INTERPRETATION
Why a modeled fitness score can fall during a rest week
A training model changes as recent inputs age. Its label should not be confused with a direct fitness test.
Read the score as a model state
A training-load trend summarizes recorded activity over time. As demanding sessions become less recent, a model can show a lower value even though no new fitness test has taken place.
Runome lets you read training load alongside the activities behind it. Pair the trend with the reason for a quieter week: a planned break, a schedule change or an incomplete activity history.
A simple moving-average example
A five-day average makes the arithmetic easy to see. These invented load values illustrate a public statistical idea, not Runome’s fitness or fatigue calculation.
| Point | Daily load values | Mean load |
|---|---|---|
| Initial window | 40, 50, 30, 60, 20 | 200 ÷ 5 = 40 |
| After one zero-load day | 50, 30, 60, 20, 0 | 160 ÷ 5 = 32 |
| After two zero-load days | 30, 60, 20, 0, 0 | 110 ÷ 5 = 22 |
The average falls because earlier values leave the window and zeros enter it. This is a change in the summary, not a measurement of how much fitness the runner has lost.
Check whether zero really means zero
Before interpreting the curve, confirm that no activity is missing and that the included runs have usable load inputs. An unimported workout can look like a day without load. Profile changes or a recalculation may also alter the available history.
Do not compare this model state numerically with a similarly named value from another service until its weights, load definition and initialization are known. Matching labels do not guarantee matching mathematics.
Use the trend alongside real context
In Runome, inspect the activities behind the load history and consider the schedule that produced the gap. The model cannot by itself decide whether you are recovered, injured or ready for a hard session. Treat personal symptoms and real-world training decisions separately from the arithmetic.
Read how missing activities change a trend before explaining an unexpected drop. For a broader weekly review, compare volume and workout structure rather than trying to keep one modeled number rising every day.