Week 3 is the third active Sleepgenic Weekly Report. After Week 1's adaptation pressure and Week 2's clear recovery, the purpose of this report is to read the third week against the established pattern and the Week 0 baseline — and to determine whether the body continued recovering, plateaued, or began to accumulate strain beneath stable scores.
The training week followed the same structure as Weeks 1 and 2: a Rest day, a Summit, two Recovery days, a Rucking day, a Run, and a Walk. The stimulus rhythm did not change. That consistency matters, because it means any change in sleep response this week is a change in the body's reaction to a familiar load — not a reaction to a heavier one.
At the score layer, Week 3 looked like a continuation of Week 2. The average Overall Score was 68.1, compared with Week 2's 70.7 and Week 1's 60.2 — essentially flat against the prior week and still well above Week 1. Quality Score averaged 73.3, and Duration Score averaged 69.9. Recovery Score averaged 82.1 — the highest weekly Recovery Score recorded so far, above Week 2's 78.6 and far above Week 1's 62.0. Sleep Stress averaged 14.9, also the lowest of the run, slightly below Week 2's 15.8. Read on the headline numbers alone, Week 3 was the calmest, most recovered week to date.
But the physiological layer did not agree.
The clearest signal was REM. Average REM sleep fell to 0.67 hours, down from 1.10 hours in Week 2 — a drop of roughly 39 percent. REM was below 0.60 hours on four of the seven nights, and the week's feedback tags reflect it directly: "Not Enough REM" appears repeatedly, alongside "Poor Structure." Deep sleep moved the same direction. It averaged 0.99 hours, down from 1.38 in Week 2 and 1.46 in Week 1 — the first week of the run where average deep sleep fell below one hour. Total sleep was not the problem: at 6.48 hours, it was the highest weekly average so far. The body was spending adequate time asleep. It was the composition of that sleep that thinned.
Resting heart rate continued its slow improvement, averaging 58.9 bpm — the lowest of the run, down from 60.6 in Week 2. On its own, that reads as a positive recovery marker. But it has to be set against the most important limitation in this week's data.
HRV is largely missing for Week 3. Only two of the seven nights — May 2 and May 3 — recorded an HRV value. The remaining five nights returned no reading. The aggregate HRV figure of 42.5 ms is therefore calculated from two nights, not seven, and should not be read as a weekly average. It is reported here only for completeness. This is a consumer-grade wearable limitation, not a physiological result, and it means the autonomic picture for Week 3 is effectively dark. For a week that otherwise shows thinning sleep architecture, the absence of HRV data is itself the signal worth noting: the single most sensitive recovery biomarker went unrecorded precisely when it would have been most useful.
There was real variability across the week. The strongest night came on May 7, after the Run, with an Overall Score of 78, Recovery Score of 85, and the week's only REM reading at or above 1.0 hours — tagged "Positive Continuous." The weakest came on May 6, after the Rucking day, with an Overall Score of 53, REM of just 0.23 hours, and a "Poor Structure" tag. The Summit night on May 3 scored a respectable 73 but carried elevated Sleep Stress (19.5) and the week's highest Restless Moments count (60). The pattern is not collapse — no night dropped into the 40s, as happened in Week 1 — but no night was architecturally clean either.
The interpretation is that Week 3 was not a recovery week and was not a strain week in the way Week 1 was. It was something quieter and, in longitudinal terms, more important to catch: a week where the score layer stayed stable while the physiological layer weakened underneath it. Recovery Score reached its peak, Sleep Stress reached its low, total sleep was ample — and at the same time REM fell by nearly 40 percent and deep sleep dropped below an hour. Those two facts do not contradict each other. They describe a body that is still producing good-looking scores while the restorative structure of its sleep is thinning.
Sleepgenic classifies Week 3 as latent strain under stable scores. The headline numbers did not warn of anything. The architecture did. This is exactly the divergence the Sleepgenic methodology is built to surface — the reason a weekly report reads the physiological layer separately from the score layer rather than trusting the composite number.
The watch item going into Week 4 is explicit: REM sleep has now declined for two consecutive weeks, from 1.10 to 0.67 hours, and deep sleep has fallen below one hour. If the next week converts this thinning architecture back into REM and deep-sleep recovery, Week 3 will read in hindsight as a brief dip. If it does not — if scores follow the architecture downward — then Week 3 was the early warning, and Week 4 will be where the strain becomes visible at the score layer.