Avg GAP higher then avg pace

In my case the avg GAP is indeed mostly faster than the avg pace, and it is hard to guess why it is not so in your case. If you post a link to a sample activity, perhaps we can see some clues. For example, I could imagine if there are some steep descents, they can actually slow you down. I guess that the formula Garmin uses is not necessarily linear.

Extremes said:

Can someone explain my why Avg GAP for my runs is almost always higher or the same then avg pace when elevation difference is zero
or even when total ascent is higher then descent ?  

Extremes said:

My understanding is that when route is not flat avg GAP should be almost always lower then avg pace as running up and down is more difficult then on flat and
Strava seems to be correct but I'm not able to figure out what is Garmin logic here. 

Personally I see this kind of thing all the time with Garmin GAP, and as a result, I don't trust it all.

GAP from strava or runalyze seems more realistic.

I'm not able to share from connect as is not public, but my point here is to understand what is logic here and what it means from body point off view that my gap pace is slower or the same then average when I run loops and elevation difference is zero but route is not flat. As FlowState mention here I also see this situation all the time and Strava looks like much more realistic. You mentioned step descent slow me down, when gap pace is slower only for this descent that's completely fine, but still why slower then average if later I need to go then same meters up. 

Extremes said:

As FlowState mention here I also see this situation all the time

I cannot confirm that. Among many hundreds of my runs, I have almost none where the GAP would be slower than the average pace. The only exceptions are some weird runs with a lot of pauses and walks, so my guess is that you are comparing the Avg GAP (probably based on the Moving Time) with the overall Avg Pace, and not with the Avg Moving Pace. Strava indeed shows the pace based on the Moving Time by default, while Connect's default metrics is the plain Time (neither Elapsed Time, nor the Moving Time), which includes the non-moving (while non-paused) periods.

So if your runs use to have a lot of breaks and pauses (without using the Pause or Auto-Pause functions), then that would explain the discrepancy. In such case, open the activity, and rather compare the Moving Pace with the GAP.

I almost never stop, walk, stats from two runs:

Extremes said:

I almost never stop, walk, stats from two runs:

Yes, and the Avg Moving Pace and Avg GAP are identical - 4:46 min/km, so there is apparently no problem here, since the Total Ascent and Total Descent are also almost the same.

Extremes said:

that it should be not the same as I understand running flat is fastest.

The descents are usually the fastest. Except of really steep dangerous descents, where you have to brake. I've set most of my PRs (up to 5k) on long descents done at paces that I could not hold so long on a flat terrain. Though, I do not know whether the GAP calculation uses any threshold for a dangerous slope where you have to slow down, since that is very individual - I know many trail runners who can descend even the most steep slopes practically in freefall, and definitely still going much faster than on a flat terrain. I have my limits much lower.

GAP is normally faster than plain pace on ascents, and oppositely slower than the plain pace on descents. So if the total elevation gain is close to zero, it is not surprising having the Avg GAP and the Avg Pace close to each other or even identical.

I completely agree but GAP should show you what is your pace when you run the same route flat and my understanding was that for example running 1km when you need to go 10m up and 10m down, should be always slower then running 1km with no elevation change. You changing your stride, cadence, pace so it's more taxing for your body then running just flat with the same pace. Ok let say that it's just simply meters up and down so in that case this still mystery why for intervals when I start and stop in the same place my GAP is slower. 

AFAIK, GAP only uses a purely physical coefficient to normalize your ascending/descending pace to the equivalent pace on a flat terrain. It does not take in account the extra fatigue you may accumulate during difficult ascents. So for example if you ascend 10 meters on 1km (1% slope) you go X-times slower, but when you descend again 10 meters in 1 km, you go X-times faster, and the coefficient X is the same in both cases, so the resulting average GAP is expected to be identical to the flat pace.

GAP should not be slower though, and in my case it indeed never is, except if I compare the wrong Avg Pace instead of Avg Moving Pace with Avg GAP.