New max heart rate vs. max heart rate recorded

6988996 said:

Is it essentially writing it off because it was such a short spike? 

Yes, it was written off as non-qualified data.

And yet I had a non qualified 209 today and that’s now my max so my actual level 3/4 workout is now a level 2 with only cardio benefit and a weak one at that. I know I can manually set my max but I’d like to be able to edit my data and then have garmin put the max back to whatever I was before. Sigh. Forerunner 965 fwiw (which is little I know).

Edit: garmin connect is rational and rejected the 209 but the watch took it as my new max. I don’t know which firmware it was - my watch updated on the way home!

I'm in a similar boat and I don't know what to make of it. I went on a run yesterday and my max HR for the activity was 165. It set a new Max HR at 212. Now, I'm in my 30s so that seems like a bit much. I wouldn't mind that so much if daily suggested workouts didn't seem to base threshold on strictly Max HR so my threshold runs are like 185. Maybe I'm out of shape, but I can't maintain a 185 heart rate for 15 minutes. 

Is this happening with or without a chest strap?

Without a chest strap, I think the wrist HR inaccuracies might overcome the filtering and qualifying algorithm...

With a chest strap, there is still a risk of spurious data: dry strap, cold weather/wind, low battery, worn out strap. Straps have a life duration of about 12 months if you train several days of week. Your mileage will vary with volume, how well you rinse it, etc. Batteries might tap out twice as fast. Again, your mileage will vary.

This is with a chest strap on an indoor trainer - a very clean and quiet environment. And I'm surprised that straps are only good for a year. But the spurious data was less than one minute but was completely bogus.

thisIsMyUser said:

And I'm surprised that straps are only good for a year.

Again, your mileage will vary. If you train a couple of hours a week, they will last longer. If your train a couple of hours a day, they will wear out fast: oxydation, stretch/rupture of the sensors, etc.

thisIsMyUser said:

But the spurious data was less than one minute but was completely bogus.

There is not much you can do about it. The strap might react to a shift of position, strap adjustments, or heavy breathing that move the electrodes around for a few seconds. The algorithm that qualifies the information data segments will do what it can, but nothing is perfect and it might too hard to distinguish a brutal increase in effort from a totally false data segment. From the patent filed, the watch is evaluating 5s data windows.

Well, I hear what you're saying. But the online algorithm (garmin connect) is smart enough to reject the spurious data. The beta my watch was on (two betas ago!) wasn't. So... I get it. I work with spurious data and spikes and filtering. There is/was a bad filter on the watch and/or incorrect logic applied by the watch to update the max hr. It should have gotten the update info from garmin connect and not from what it thought. At least, I'm going to assume that there are many more users on garmin connect and the algorithms are better, there.

All these algorithms are run on the watch. The graphing on Garmin Connect might hide some daily spikes? Maybe some sync issues/conflicts? Maybe you have several devices?

It is hard for us users to do data forensics on what the watch sees, qualifies, calculates, which data is saved in which file at what point, how the sync performs, ...

If you are not happy with the way the technology works for you, you always have the alternative to do lab or field testing for Max HR and LTHR. These are hard tests, so some people might not be willing to trade the extra effort for the extra accuracy.

I’m less concerned with the actual value than I am with the (bad) algorithm that chose the number for me based on a momentary lapse of reason. 

As for the algos being the same then explain that the watch shows a spike and Connect doesn’t. I’m going out on a limb to say all of the data get transferred from the watch up. But if they don’t and filtering is happening on the watch then clearly that part of everything on the watch knows the figures are bogus. In any of these cases there is the opportunity of using shared code that is not being taken. Omg! Is Connect written in monkey C???? That would be something!

The data are obviously (to a human) bad. What some of the programs do with the bad data is bad, too.

thisIsMyUser said:

As for the algos being the same then explain that the watch shows a spike and Connect doesn’t

This is a question for Garmin support.

thisIsMyUser said:

I’m going out on a limb to say all of the data get transferred from the watch up

I don't know for sure, but I doubt it. The results data is stored in files that are synched with Garmin Connect. The web side is only displaying the info in (maybe) more convenient way. I am thinking the fitness age is a back-end algorithm given the user experience to get the result, but all the core training metrics are calculated during workouts and wearing the watch, then filed away.

thisIsMyUser said:

In any of these cases there is the opportunity of using shared code that is not being taken

I don't know what that means. It sounds like you are an engineer working at Garmin with detailed information about how the code base is managed :-)