Breathing affects HRV greatly.
I coded bit of software to read HRV with PC, using Polar H7, so that I can see time between hearbeats changing realtime.
So if my normal relaxed rMSSD runs between 20-30, after workout/eating rMSSD can get to 6 and if I take deep breath, variability jumps from 6 to 100 or even more.
It does show clear drop if eating too much sugar/chocolate though...
This shows milliseconds of change between hearbeat, dataset is 2161 heartbeats long (unfiltered), HR between 56 and 85 unless I remember wrong:
rMSSD is root mean squared value of chosen time interval, which according to papers I have read does tell much better changes from normal than some other measures of HRV.
Mobile apps can then use their own formulas and choose different time intervals, filtering data can affect some too etc. Value is not too useful, if it is not known how they are getting the number.
Also individual differences are quite big, your own moving average is what you can really compare it to, but with enough data, it can show if load is a lot more than optimal.
So for me, if rMSSD of 5 minutes is 15 or less at the morning, I can tell there is need to take easy, if it is 30, I can push harder.
As I coded program to monitor HRV, it has been much easier to see if my value is high or low.
Simple measurement with Polar just is bit random as it is just one measurement supine and one measurement standing, noise vs data issue, at least with my case.
With constant realtime measurments for 1 hour at the morning, I can see much more, I would not trust much to measurements that are done just one 5 minutes at the morning, as you can see from sample data, variation of variability is rather big
If some programmer wants to experiment, here are some valuable sources, some basic information how to interpret Bluetooth HR sensor data:
C#, pairing and receiving data stream from LE Bluetooth device Polar H7 [NEED HELP] | Brackeys Forum
Sample program, which helps greatly as it is ready HRM with Polar H7 using Bluetooth, all that is needed to change is to read HRV data that sensor sends after HR, very easy modification to do.
https://github.com/Ryan311/BluetoothLEForDesktopDemo
You just need to reference few assemblies, namely System.Runtime.dll and System.Runtime.InteropServices.WindowsRuntime.dll
After getting data from sensor, it is easy to do all sort of analysis, there are lot of scientific papers that have great information how to process data, best practices etc.
But, HRV is just one indicator, use it in combination with HR and general feeling, it is not going to be absolute truth and salvation, just sometimes it can be somewhat good indicator, at least that is my experience.