Are there at least 2 streams of data responsible for what you see in Connect?

I agree, "streams of data" was a misleading and inaccurate choice of words. I was trying to describe this user's experience of seeing what was previously displayed in the Android app get rewritten minutes later and slightly altered, presumably after some additional processing.

Venutianiii said:

For example, I see that I have climbed/descended many flights of stairs on days when I have not really done so. However, I do see correlation with higher overall activity levels on those days. 

In this case, I would assume that the both the app and the watch report the same amount of stairs climbed/descended (whether it's correct or not).

https://support.garmin.com/en-CA/?faq=LJ6vYF2phv7JUAOlsRRjC8

I Am Getting Too Much Credit

Situations that can cause excessive credit include a significant change in pressure such as a strong pressure system or walking into a warm building on a cold day. Walking near exit doors when it is cold outside may also trigger extra credit. Walking outside on a windy day with the watch exposed to the wind can also result in false credit.

You will also get credit for stairs climbed/descended if you go up and down a hill (assuming the hill is steep enough). Garmin is fairly transparent about this:

This feature uses the built in barometric altimeter sensor to detect a change in air pressure which occurs when you climb or descend a flight of stairs or a hill.  

Venutianiii said:

Also, I will see my real time stress graph (for instance, on Android app) in the low blue zone when resting and inactive and subsequently after activity it will of course spike into the orange. What I do not understand is why some of the prior blue areas seem to get re-written (retroactively) on the graph later on to reflect additional stress if I have been particularly active minutes later

Interesting.

Is it possible that this data is retroactively changed on the watch as well? 

Speaking of this, Garmin claims that stress should not be measured during "physical activity" (because high stress levels are supposed to be normal and somewhat beneficial in this case, if I'm understanding them correctly).

www.garmin.com/.../

But people have been saying that they're seeing high levels of stress during a timed activity:

https://www.reddit.com/r/Garmin/comments/1gpvkic/garmin_showing_activity_as_stress_whereas_it/ 

As far as filtering and processing of data goes, this definitely does happen on the watch itself. If it also happens in the Connect app (or, actually, the backend, since you can also view the same data in the Connect website), would that really be a problem? I still think that the majority of the data which is gathered/calculated by the watch should be presented in Connect without modification (except for things like rounding and aggregation).

FlowState said:

But people have been saying that they're seeing high levels of stress during a timed activity:

In that particular case on Reddit, the watch was not recording any timed activity (Edge was used for that), hence it was registering the stress as usually. So it still stands - watches do not log stress during timed activities, or when excessive movement is detected (which was apparently not this case). That told, there are some exceptions - for example Stress is being logged during Yoga (and possibly Meditation) activities.

Venutianiii said:

what was previously displayed in the Android app get rewritten minutes later and slightly altered, presumably after some additional processing.

You mentioned in your other thread that you use two different watches. Hence it can easily happen that the data that you have seen previously, were adjusted, when the data from both watches is merged together, which may indeed happen with a significant delay (sometimes both watches have to sync twice, in order to exchange all data completely).