Example of garbage HR

There is a lot of Garmin garbage HR data about these days but the type of garbage can vary in no small part based upon the software version in use. Just in case you don't know, 

Hold the left middle button, when the screen changes to  show "change sport" release and then press the left middle button once 

That should take you to "system". Press the right top button to select system, then the left middle button once which should take you to "About"

Press the right top button to select "About" and that should show the Software Version Bundle which will tell you what build you are on, mine is 21.20. I get garbage data but a little differently than yours which could be due to the software version, or my application of the sensor. Garmin has been in the constant fix mode but the fixes haven't really worked from what I've seen and based on the feedback that has been logged the problems vary from release to release.

I'm on the latest version which is 22.24. Unless you can reproduce the issue consistently on the same OS version, you can't say for sure it's the OS version that's causing this.

Very True. And even for me on the same version with pretty consistent results the garbage data doesn't come out exactly the same each time. I am pretty certain of the cause through and the following is my opinion, nothing Garmin has said.; In a vacuum the  FR265 / FR255 Optical HR sensors are probably technically superior to the earlier version of the same in the FR245. In practical use Garmin has been trying to use software to compensate for the finickiness of the new sensor and it hasn't worked out well. Some of this even went on in the FR245 where the response of the sensor varied with software releases, to the point  that less than a week after purchase, an automatic update killed the accuracy of the HR sensor, so I took the watch back, got another and disabled automatic updates, life was good.

Problem with the FR265 is I've yet to find a release that works as well as the what I allowed for the FR245.

Garmin is no doubt doing a bunch of filtering on the data. Typically if I do a treadmill run, 8-10 minutes into the run my HR will go from reading too low to reading what I believe is correct, in the form of a vertical jump in the HR value.That is pretty consistent but the time frame isn't exact. Likely that is do to variations in my starting HR and how fast I ramp up the pace and with the pace the HR. Interestingly if I follow up the TM Run with a session on the stair stepper, sometimes I get the vertical rise, sometimes not. I suspect that it's again dependent on the time between the sessions (how long have I been at a relatively repeatable HR before ramping up and what HR did I start from vs full intensity on the stair stepper.

I also have to factor in am I wearing the watch the same. I've gotten to be pretty habitual on snugging up the watch before starting an exercise. For a given version the software processing should be consistent, but the variables being fed into the software less so. Next release, who knows. Did Gamin tweak the optical sensor processing, or even make some general change to the overall sensor interface that had unintended consequences. Unless there is something noted in the version change logs, it's impossible for us users to know; Garmin isn't particularly chatty about the HR sensor problems 

It bothers me A LOT that the FR265 is the 2nd generation of the FR series to use the newer optical HR sensor. Effectively one generation has already been orphaned. Turnabout will be fair play when it's time to consider a new watch should this problem not get addressed.

in my case the wrist OHR malfunction seems to correlate with cold skin. when i get cold hands during running, the HR measurement is often unusable. there are two scenarios for garbage data:
(1.) the HR slowly ramps up , until it reaches (for me) absurd high values like 185bpm
(2.) it shows steadly a too high HR like 155, but my real HR load is e.g. 110-125