Resting Heart Rate Calculation Inaccurate

Jon said:

Today I saw on Garmin Connect that my RHR was 47, but I was able to manually scrub through a 30+ min period while I was sleeping (~1am) during which my heart rate was 44-46 and did not even hit 47.

You cannot really know, since in Garmin Connect the HR is being shown with the sampling rate of 2 minutes, while the watch looks at the live continuous HR data with the sampling rate of 15s. So the 30 minutes average over unfiltered data can be in fact slightly different than the average over the same period done with the data sampled each 2 minutes.

Thanks for the reply! Even if that is the case, though, statistically that discrepancy shouldn't be there if the 2 minute "sampling" on Garmin Connect just comes from averaging the 15-second HR data from the watch in 2 minute bins. In other words, if the 2-minute averages never went above 47, then it is impossible for the 15-second values to average above 47. And even if this 2-minute sampling is not using the 15-second values from the watch, it seems extremely unlikely that my heart rate consistently spiked between every 2 minute sample to make the watch-calculated average increase without Garmin Connect "noticing." Does that make sense, or am I misunderstanding something?

Jon said:

statistically that discrepancy shouldn't be there if the 2 minute "sampling" on Garmin Connect just comes from averaging the 15-second HR data from the watch in 2 minute bins

That's the question. Theoretically it should, but it may also depend on the device. Then there may be another condition, when calculating the moving 30 minutes average - for example that it gets reset when the HR rises above certain level, and the 30 minutes window starts anew. We cannot know, since the exact algorithm is not published.

However, personally, I do not see any similar discrepancies in my RHR data. I have a recent watch model though, so it is also possible that older models like the FR45 do not average the 2 minutes sample, and rather just log the last 15s value (or an average shorter than the 2 minutes). 

Could you post a screenshot of the all-day HR, zoomed to the maximum at the concerned 30 minutes interval? In GC Web, the resolution may be better for seeing the details.

Right, I guess also if they aren't taking a constant moving 30-minute average, then they might miss a true 30-minute minimum.

Sure, here's a screenshot of a 90 minute window. The 30 minutes around my cursor hovers around 45-46, never going to 47, but my average RHR is reported as 47.