Edge 540 "Performance Stats" are nonsense

After 5 years of Edge 530 and now 540 I can confirm, that the performance Garmin algorithms (or Firstbeat) is nonsense

Did you have the physio true up setting enabled?  That essentially syncs data between all devices.  I recently upgraded from an Edge 530 to 540 Solar and I didn't notice anything different.  My watch and 540 have the same suggested workouts.  I'd just give it a few days before writing it off.  If you just suddenly got a lower VO2 Max reading that differs from what you'd seen previously, it could just be data from one bad ride where you were too hot, didn't get enough sleep, etc.. It should balance out.  It will also "get used to" your training load and adjust accordingly, as I'm sure you know.  It may just be that the device needs to get to know you.

I have no idea what physio true is. As far as I can tell, the 540 is supposed to be synced to the data already on Connect, but I can't, for instance, use the 540 to look at pre-540 rides that are on Connect (which is fine, of course). But when I first set it up, it did seem to use recent rides to automatically set my FTP based on my best 1hr performance on recent rides. Then a few days later, reset it to way below that. Then after I logged a few more rides on the 540 that were way better than the FTP it auto-set, it never updated my FTP to  match the performance on actual rides. So I manually reset it. Then after today's 3:30 hr ride (which, note IS NOT an FTP test, and really none of my workouts are paced to be FTP tests), the 540 again automatically reset my FTP to *much lower* than I've performed on other ~1hr rides this week (which also aren't FTP tests). The power zones on that 60-miler made sense based on the manual FTP setting, a lot of Z2 and Z3 (and more Z1 than I liked, but keeping a steady power on real roads is just impossible!). But based on the 540's automatic "correction," my zones would have been way too high.

What the *** is going on? I posted a reply, that one post was repeated 4 times, and part of the post is missing. Is there a moderator who can, I don't know, fix it? Restore the missing portion of the post?

I had written something about whether the training load score is inflated beyond usefulness because of Garmin's serious underestimation of my fitness. And about how the 540's post-workout predictions of my recovery time are usually about 4 days, but when I go out the very next day for my workout, it tells me that I am recovered and ready for a good workout (based on its "Performance Condition"). If that section can be recovered, that'd be great.

the forum's engine is a nightmare, as the server under the forum also lagging for 10-15secs very often. in the beta section you cant modify a comment or add a picture, you get "error occured.... blabla " messages all the time.  

yes, this FTP auto-estimation "bug" (Garmin told me it works as it should... :D, so it is a feature, that if you ride 1h as the golden standard is not enough for this marvellous algorhythm. why dont they call it GFP aka Garmin Fictional Power)  is exist, i rode many times (i tested this FTP auto thing on purpose)  for 1hour above the autodetected FTP value... (the 20 and 30mins were far above with 10-15%). then after 1 week , on a fully Z2 ride where i suffered as a ***, i was weak , especially in the last part  of the ride my perf. condition went to -9 or similar (it was right :D) , at the end of the ride the Edge gave me new FTP with +20watts or similar (ofc, it was still under my real FTP what i know extra accurate with closed eyes). yes, sometimes it decreases it, probably there are body types where the algorhythm falls.

Guessing FTP based on a single (non-test) training ride shouldn't be a "gold standard", but it is definitely a lower-bound. A "sophisticated" algorithm that takes in *a bunch* of data to estimate FTP lower than demonstrated, actual performance is just *broken*. If auto-detect FTP would *only* update FTP when you demonstrate higher performance than the currently set FTP, that would be a nice quality-of-life feature. If auto-detect could analyze your performance across power and heart-rate zones and give a recreational cyclist a good(ish) estimate of how you'd perform on an actual FTP test (which the recreational cyclist thinks is too much trouble to perform, for various reasons), that would be pretty cool. But the actual auto-detect, which overrides demonstrated performance to make up numbers, that you then have to manually go and correct? What is the point of that? Is it just to encourage people to subscribe to more Garmin services? (like, who would do that, given that the device is showing them that their coaching algorithms are nonsense?)

But I'm also just confused about how much all these bad "estimates" are being used as input to all the other algorithms and circling around to make all the estimates worse: is my bad VO2max estimate being used to calculate that my training loads are too high, and my bad HR zone estimates feeding being used to misestimate my FTP, which give bad power zone estimates, which are feeding back into my training stress estimates, which are feeding back to my VO2max estimates, and HR and FTP estimates?

it wasnt a single training, it was 3 days in a row where i rode every day on higher 1h avg power than the last known autodetected FTP value (lets says it was 240w as last known and i rode on 250, 260 and 270w for 1h with zeros included ofc) . and it didnt change, just 1-2 week later on a low Z2 ride... so the algorhythm can give you estimated FTP from a low intensity ride also (derivated from PWR/HR and maxHR rate i assume) just if i am a newbie, i train with zones lower from the autodetected FTP for weeks (or forever... because it is still lower with 10% than the reality, and i used it 2.5 years previously , with all range max efforts from 1sec to 2h , so my whole power curve was fullfilled and it never could give me accurare FTP estimation, and the other platforms like intervals icu or wko gave pinpoint accurate ones...) 

My Garmin has my max HR set at 180, which results in a lot of ~1hr workouts entirely in what it labels "Zone 5". I manually bumped my max HR up to 184 based on my highest HR during a workout from this season, which is still mostly in Z5, but after today's ride, 45 minutes out of 72 total in Z5, the rest in Z4, overall a *not very fatiguing* workout, my Edge decided that actually it preferred my Max HR at 180 again.

If it's going to autodetect that stuff, it needs to move my max UP. Why is it moving it down? And are these too-low zone assignments affecting my training load scores? Are my training load scores affecting how it updates my other fitness metrics? Etc etc. Can anyone tell us how this stuff makes sense?