How is the HRV 7 day average calculated?

Surprised? Garmin have their own values, and they are not often in line with the real world.

HRV , i am 100% sure that the algorythm has been changed in the last update . 

But don’t expect any meaningfull answers from Garmin.

I've tried to figure it out as well and got close to some of the numbers, but others don't work out. https://forums.garmin.com/sports-fitness/running-multisport/f/forerunner-165-series/389412/hrv-7-day-average

It should be a fairly simple calculation as you described, but it seems like they do something different. I might just keep track of the daily values and make my own calculations / graphs based on the values given for each day.

Just averaging the 7 days gives 51 (if I calculated it correctly), so it's fairly close. But Garmin could weight each day based on sleep length, for example. That would be logical, because then it would become the true sleep HRV average over the 7 day duration.

Bitti I have no clue what “true sleep HRV” is, but I do know if (and that is an if, so not sure) garmin adds sleep lenght/score to the calculation it would only make things worse. Sleep tracking isn’t very accurate.

It’s just garmin marketing trying to sound scientific with mumbo jumbo calculations. HRV trends are somewhat informative, but I haven’t read anywhere that a 7 day average would be important to keep an eye on.

I meant "true average" messing the average you would get by averaging out all hrv values for 7 days. Calculating the 7-day average of daily averages weighted by the size of their measuring period gives the same result, but just simply averaging nightly averages doesn't (unless the same amount of sleep is detected each night).

It doesn't depend on whether sleep detection is accurate. Quite the contrary, it would be bad if one day only 1 hour of sleep was detected, and its hrv would affect the 7-day average as much as 8 hours of hrv on another day. I have no knowledge on how Garmin is actually doing the calculations, but I'm just saying they might be doing the right thing.

I agree with following hrv trends, and the 7-day average makes it easier since it smoothes out outliers (which occur naturally).

I think we both agree we are just speculating. (Which is enjoyable)

So, you want a really bad night with only 1 hour of sleep, resulting in low hrv, to not count as much as a good night? Why? If the bad night did happen it should count just as much as other nights.

Let's put it this way: If you want to know your weekly average running hr, do you want the Monday 10 minute run you had to stop early to count as much as your 1.5 hour run on Thursday? Or do you want to calculate the average hr of all your weekly running minutes?

I don't want to know my weekly average running hr, because it is absolutely meaningless. 

It might be that 7 day hrv average is just is meaningless, but I think a 1 hour of low hrv due to poor sleep is just as important as a 8 hour good sleep high hrv. 

But I still think it is quite useless to calculate a weekly average hrv. 

Well, then there we disagree. I don't care about a single night's hrv that much (since it fluctuates so much), I prefer the 7-day rolling average. A single night might tell me to take it easy the next day, but the average tells me about possible overtraining etc. (That's why Garmin's balanced/unbalanced HRV uses the 7-day average, and that's what's shown on the HRV glance on the watch.)

And my running average hr was an example of how combining averages works in mathematics, not an example of real use. But the same rules apply to any average. Also, I often have short sleeps with high hrv (someone/thing woke me up early). And I still think 3 hours of good hrv doesn't show recovery as much as 8 hours.

And of course your body still recovers outside of sleep, no matter how much you slept. So I think where we agree is that the longer term trends are more important than individual nights.

We don't have to agree.