Adjust the parameters to see how the comforter decision changes. Based on a Bayesian regression of 57 nights of Eight Sleep data.
| Comforter | Price | Score delta | Predicted score | Annual value | Payback | 1-year ROI |
|---|---|---|---|---|---|---|
Current (waffle knit) ~0.5 CLO · baseline | — | — | 79.0 | $0 | — | — |
Quince Ultra-Warm 700fp · +0.5°C | $240 | +1.4 | 80.4 | $913 | 96d | 281% |
Brooklinen Ultra-Warm 750fp · +0.8°C | $430 | +2.3 | 81.3 | $1,461 | 107d | 240% |
Hamvay-Lang Winter 800fp · +1°C | $885 | +2.9 | 81.9 | $1,827 | 177d | 106% |
Methodology
Bayesian linear regression (conjugate normal-inverse-gamma prior) on 57 nights of Eight Sleep data. Default coefficient: 2.88 ± 1.07 score/°C (t = 2.68, p = 0.010). Value per point derived from Eight Sleep subscription cost. Temperature deltas are conservative estimates accounting for Pod 5 thermal compensation.
Built with data from health-data · Read the full blog post