Elevation acclimation - going down while at altitude

This is an expected behavior. 

ticktakats said:

going up until a few days ago. Now it’s going down

I can't speak to the elevation algo since I'm usually stuck at 800ft.  But my temperature acclimation does drop (even if I've already achieved 100%) if I haven't trained for a few days even though the ambient temperature is above the acclimation algorithm trigger point.  I assume it's the same for altitude.

https://support.garmin.com/en-US/?faq=PQCtbgWxJ65nRatXoHCmy7

This link describes how it is meant to work.  If you have not left the Springs then doesn't look like it should have gone down.

Exactly in your link it is written that its expected to go down after a period of time...

• Altitude adaptation will decay within 21-28 days.

Interesting... thanks for the responses. Do you know why it decays if I am still at the same altitude?

AgentMax said:

Exactly in your link it is written that its expected to go down after a period of time...

I am not a health scientist or doctor, but I read that to mean that if you LEAVE altitude it would decay within 21-28 days.  Why would your acclimation to altitude decrease if you remain at altitude?

I don't know. But at Heat acclimation they wrote that it will decay if you are not exposed to heat anymore. They did not mention it at altitude acclimation. 

Logic suggests that you are fully acclimatised at that altitude so the reference point would reset gradually. In other words, if you stay there long enough it becomes normalised.

It happens to me very often during multiday hikes. For instance it can report acclimation for 1800 m, and after 10 hour hike in altitudes 2000 - 3000 m and night in 1400 m it goes down.

During skiing it goes fast to altitude of accommodation but it never goes above the level of the village where I sleep even if I ski in much higher altitudes.

In some other thread on the forum someone make a widgets for "proper" acclimation. According to him, Garmin algorithm use only one value per day to estimate it. It coincides with my experience. One value is not definitively enough for good estimate. My approach is to not to pay much attention to this metric.