Course ETE / ETA - what's it based on?

I am quite sure the Edge uses the set pace in the Route and that you set when you create the route, in the Connect web or the mobile app. It is not based on average speed or anything dynamic. So if you set a realistic pace when you create the route you get useful values, else they can be way off because then it uses some default value. In other words these fields are accurate for the virtual partner since it uses the set pace as well. The ETA will update when you pause since the virtual partner also pause.

Hmm, from someone that has never used Garmin's course creator but instead all imported from third-party platforms / stolen GPX files from randoms.. I seem to have fairly good results so it can't be exclusively this.

Also, the same data fields are available when you do on-the-fly route navigation to a place/POI/address etc so there must be some averaging used.

I do however know this is the case if you set a target pace on the watch - I did this for a flat marathon PB attempt recently, forgot about it, then went for a long gentle trail run (using the same activity profile by accident) = I was somewhat confused why my watch was being rather optimistic with its ETE, before realising it was trying to keep me to marathon PB pace!  :)

This is not an answer to your questions, but I also wondered about the predicted finish time the watch shows before starting a course. I could not find a good explanation about this yet.

I believe it is a formula like the following:

ETE = (distance travelled so far / elapsed time) * distance remaining

I don't think the elevation profile of the course is factored in, but I might be wrong about that.

This is a really interesting question. I've been thinking about this myself many times and also screen shooting many times to analyse it further. Never done it though.

I think it's most based on the time data you have in the file. If you would like it to compensate for up and downhill, you can use for instance plotaroute to configure percentage for compensation. Also pacepro has some sort of simple elevation compensation.

But it's not only based on file data, then it should be very stable from the beginning, I think and it never is.

That is my observation as well - it used the average speed so far. Specifically, I noticed a case where I had a fairly slow trail run with a much faster final few miles on a dirt road. And even though I was moving fairly fast at the end, ETE was still twice as long as it would actually take me based on my pace at the end. ETA was similarly wrong.

 Other watch brands like Suunto calculate ETE/ETA based on a recent speed, which works better for more dynamic sports such as trail running.

I too think it is very much skewed towards the average speed of the current activity. I say this as whenever I look at it, it seems to be reasonably accurate, albeit I know to adjust it in my head for what I know about the remainder of the route (easier/harder etc) - thing is I rarely look at these data fields until past half way - ie. several hours in and by which point I've already established a steady speed for the activity so far.

All in all I think it gives a reasonable guess, though agree it could be improved by fusing historical activity-type specific data and factoring in elevation profile (especially the latter, assuming it doesn't already). That said, it's always going to be a guess, so perhaps a case "good enough".

I always asked myself the same thing.

For sure the estimated time for a Hike, when show while choosing the course to follow, is far from a reasonable value, at least for me. 20-40% longer than my average pace.

During the activity I usually read the ETA during the last 1-3 kilometres left to walk, and if the speed is pretty much constant, the estimate is very good indeed. When reviewing the ETE/ETA during the first ascents is usually waaay of, as if it would take between 30 and 70% longer than expected: that's leads me to think it using the average speed while climbing to estimate the whole course.

Yesterday I was following an externally downloaded course with a few waypoints, and ETE to each waypoint was consistent wrong. This was an ultra-marathon on a very steep mountain terrain. In the begging we were moving at about 20-22 minutes/mile, yet ETE seemed to be based on pace that is slightly faster than 10 min/mile.

So for sure it wasn't based on either average or recent activity pace, but some pre-defined fixed pace that I never configured.

That makes ETE pretty much useless for me.