Linked isn't caused
When two of your metrics move together, that's a link, not proof that one causes the other. Pajdo needs 30 paired days before he mentions one.
Pajdo looks for a few links in your data, like whether days with more exercise come before longer sleep, or whether a longer night comes with higher HRV the next morning. He checks the last 90 days and speaks up only with at least 30 paired days, a clear correlation, and an interval that clearly leaves out zero.
Even then it's a link, not a cause. A third thing can move both: a relaxed weekend might bring a long walk and a long sleep, and the walk gets credit it didn't earn.
So treat a link in your own data as a hint worth testing. Change one thing for a week or two, keep the rest as steady as you can, and see whether the other number follows. Single-person trials in research work on the same idea, with more repeats and more care, so one try of your own is a clue rather than proof.
Keep reading
Sources
- Statistics review 7: Correlation and regression, Critical Care (Bewick et al. 2003)
- N-of-1 trials, their reporting guidelines, and the advancement of open science principles, Harvard Data Science Review (2023), via PMC
- Using the confidence interval confidently, Journal of Thoracic Disease (Hazra 2017)
Checked against these sources on September 15, 2026.
This is general wellness information, not medical advice. If a number worries you, or stays away from your usual for a while, talk to a doctor.