It may be your watch, or how you wear it. How tight? It works by measuring the amount of "green" light (a very specific frequency) is reflected back to the sensor. Anything that cause a change in the reflectance CAN be picked up as heart beats. Running is typically bad because as you bounce your watch bounces, and causes changes in the skin contact and reflected green light. With run cadence overlapping HR frequency by a good bit, it can be hard to "ignore" the foot step effect. Changing the location, or tightness of the wrist strap can help decrease this effect. Sometimes people have better luck with the watch on the opposite side of the wrist, or even on the upper forearm instead of the wrist.

The Garmin tri-strap is equally if not worse. I've had 5 of them. The HRM-Run and HRM-Pro are marginally better....but, I've killed one of each of those also---all under warranty. So, its a unreliable source of truth. The other hard part is that you can't directly compare the HR output of the chest strap with the optical measurment....because you have to choose one or the other...and can't record "both". I've been down this road with the Tri Straps, and optical myself. Luckily, I had an old Garmin watch. So, I wore my old watch paired with the strap, and used my Fenix6 optical and recorded simultaneously. I uploaded both of these to Garmin Support and eventually got 3 new HR straps under warranty from that data.

You are trying to do the reverse....which may be harder. And I don't trust Garmin straps as much as I trust the optical these days.

YMMV. Just some thoughts.

PS: The Sleep detection uses the accelerometers to determine how still you are as well as HR. So, a low resting HR, lower respiratory rate, and laying very still can easily be detected as light sleep.