Resting heart rate

It's a Withings sensor;

I am not relying on the Withings, Garmin or the Apple Watch for my measurements I only take them as an indicator. I always take HRV and RHR readings with a chest polar chest strap. And from all devices the Garmin is performing the worst and with the 20.3 firmware it got way worse if you are not wearing the device at night as they seemed to change the algorithm. 

hausii said:

always take HRV and RHR readings with a chest polar chest strap.

But as you've said, you're taken those RHR readings in ONE five minute period during the day. That has the blanket assumption that those are the most restful period of the day, and without doing a rolling averaging of the data you're not going to agree for many reasons.

So you measure during those five minutes, do you do an average of the data to get a value, or take a single value?

Here's one of my days. RHR of 40, although I had the 39's listed during sleep - but as the graph has a granularity of 2 mins in Garmin Connect, Which does lead me to suspect it's plotting a low value. But it does show that the algorithm sort of makes sense - allowing for the unknowns.

Why not wear it once overnight to test.

hausii said:

Average with an app, EliteHRV using a polar chest strap

Which suggests it's averaging over the entire period. That's quite different from taking a sustained low value, 

For example, given a hypothetical dataset, it would be easy to create one with transients much lower than a rolling average, with those transients dropping the overall average.

Hence why they use a rolling average, weighted against activity level (I guess) as opposed to a blunt average,

To doubly check there's not some hardware issue, why not wear your watch overnight for say 4 days just to see if the behaviour changes with sleep hr in the mix? Would help to say either way, I suppose?

I was hoping anyone from Garmin support who could explain the changes they made to the RHR algorithm with the new firmware 20.3 and the Firstbeat implementation? It would be great if you could tweak the algorithms also that people not wearing that watch at night will be getting useful RHR measurements. 

I agree. I also have unexplainable behavior around RHR, with my Fenix 5X. But maybe this is not so much related to model?

I used another app and consistently measured RHR at around 55 (55, 54 and 53) over a 1 minute period. Tried it 3 times or so. And still Connect is saying RHR 62 for today.

Whats also funny, is that I have seen my RHR going up today in Garmin Connect, and that should not be possible at all if you follow the definition (lowest average HR measures over 1 minute). It should only be possible for it to go down.

I passed mine off as being wildly inaccurate. Averaging 38 as my resting.

However, after collapsing with a HR of 31, I can confirm that it closely tracked the ECG at the hospital for the following 4 days.

On Firmware 20.50

I got the same as you. A couple times I saw RHR going up. Very weird.