Correlations

The clearest actionable signal here is one you're probably already living: your step count is the single strongest predictor of where your body battery lands by end of day (r = -0.374), and that relationship echoes a full week later (r = -0.203), meaning genuinely high-output days have a recovery cost that doesn't fully resolve overnight. This lines up well with your 3-4 day training gaps — the data suggests your body actually needs that runway, and compressing it on weeks where you're also logging 18k+ step days would likely show up as depressed readiness by the following session. The HRV-to-stress link (r = -0.168, lagged 7 days) is worth noting but not overreading — it's statistically real but modest in magnitude, and given your 18-24ms baseline where small absolute swings look large in percentage terms, it mostly confirms that when your nervous system is recovered, your perceived stress load tracks lower across the following week. The most interesting finding is the deep sleep → steps relationship (r = 0.152, 7-day lag): better deep sleep appears to predict higher activity a week later, which could reflect genuine physical readiness expressing itself as natural movement, or simply that good recovery weeks tend to cluster. The honest caveat is that with 40 tests run and only 4 surviving FDR correction, these are real signals but not large ones — body battery in particular is a Garmin-constructed metric that's already downstream of HRV, so some of these correlations are partly measuring the same underlying thing through different lenses.