The biggest single thing wrong with most consumer biometric apps is that they compare your readings against a population average. That's the wrong reference class.
A "normal" resting heart rate is 60-100 bpm. That's a clinical range covering most healthy adults. For you, the actual range is much tighter — probably a 10-15 bpm window — and it shifts dramatically based on context: time of day, what you've eaten, whether you've moved, whether you've had caffeine.
Comparing your live heart rate against "60-100" tells you nothing useful. Comparing it against your typical afternoon-sedentary- post-caffeine baseline — that's actionable.
So Verge maintains separate baselines for every combination of:
- Time of day (5 buckets)
- Activity state (5 buckets — sedentary through workout)
- Recent context (4 buckets — none, post-meal, post-caffeine, post-nicotine)
That's 100 possible buckets. Most users only ever fill ~20-30 of them in everyday life. Each one accumulates samples over time, weighted to favour recent data — so the baseline tracks fitness and lifestyle changes without abrupt jumps.
The result: when Verge tells you "your HR is elevated +1.8σ above your usual afternoon-post-coffee baseline," that's a statement with meaning. Versus a generic "stress alert" — which fires whenever your HR crosses a population threshold that has nothing to do with how you actually function.
We'll publish more on the technical side of this as the beta progresses. For now: this is why the app takes a couple of weeks to settle into your real patterns — it's learning your fingerprint across all the contexts you live in.