GPS Accuracy

Yes, cycling is way better in fact much better. My theory is that the GPS reception is better because watch is facing up and when running is facing left.

Yes, cycling is way better in fact much better. My theory is that the GPS reception is better because watch is facing up and when running is facing left.

Cycling is better on any gps device than running because the arm isn't moving back and forth and the samples are usually clean. When you are running the arm swing causes lots of problems for gps calculations. If you were to run exactly 60 steps per minute (1 step per second) then 1 second recording would be great, but if you run at 90 steps per minute, then the arm swing is actually opposite each recorded sample. You also don't move nearly as far running as biking each sample so the watch has to do all kinds of calculations to figure out speed and distance.

Years ago on the Garmin 305, the pace was recorded without much smoothing and pace went all over the place. It was an ugly mess on Garmin Connect, but over distance it didn't matter. Over time, Garmin started smoothing out the data. It 'looks' better on the plots, but still instant pace is almost always wrong.

When in bike mode, there isn't nearly the smoothing algorithm used. Try running in bike mode and you will see a completely different plot.

I don't know if it is the arm movement or the face up/left. However went to the archives and found the same route with: 305, dakota 20, Fenix 5 bike and Fenix 5 run. if indeed the 305 didn't had smooth algorithms then there is a problem in Fenix 5 run mode (there is even a direction reversion).

Quick update from my last posts: there are a number of issues reported with the radios on the fenix 5. My latest experience is:
GPS: minor improvement since fw 7.10 - closer to real distance, but depending on where I'm running, still often very "fuzzy" as regards staying "on track". I agree with luximinator above on the inconsistency (I also reported on that before): some days it will be "close enough", some days way off. A lot seems to depend on when the watch recalibrates/reconnects GPS after having been away (i.e. running somewhere else).
I'm now out of the country and will report findings after returning...
Bluetooth: no problems worth mentioning
Ant+ (with stryd): seamless connnectivity
Pace: appears to be significantly improved since 7.10, but since I do distance and pace via footpod now, there's no real comparison possible anymore.
Re comparison between bike and run - not sure whether this was done, but the ideal way to compare would be to run with "bike" enabled, and to bike with "run" enabled to allow for real comparison...
On the run vs. bike and smoothing: this may be a very academic point and at any rate is pure speculation on my part, but it could be that the software designers were overambitious in terms of precision linked to the 1 second recording, and they now have to face the fact that the watch is not able to obtain enough precise data to allow for accurate once per second updating of the track when running.

Did a test and these are the results.
A run going up the street with normal Fenix position in the wrist;
A run going down the street with a 90º rotation in the wrist (Fenix facing up)
Also both run in smart mode without glossnass. I used Basecamp to see the GPS points and noticed that in the first run when there are deviations there are noticeable jumps between GPS points.

Anyone seeing stronger GPS results now?

Can't see any mention of GPS fixes in the F5 update changelog but does seem better than when I first bought it.

Here's a recent test between F5 and a FR935 I've borrowed from a mate.

Ran 4 miles earlier to compare GPS on each watch. To my surprise the 935 and F5 measured 4.01 with exact same average pace and 1 mile laps were 1-3 within each other. Strava on my android recorded 4.1m. There was some tree cover throughout and pretty overcast. Current pace was a little worse on the F5 but not massive plus that can be fixed with the footpod. Each device used normal GPS and had 1 second recording set.

Not sure what to say. I was under the impression after 7.1 that GPS precision was better, but then it went haywire again. I really have the impression (as seems to be confirmed from other users' posts) that the GPS resets after restart, change of location, or FW update, and that this affects precision - i.e. if you happen to catch a "good day" for a first run in a particular location, you will have "good" results until one of the above happens. After that, it's back to square one.
I have another interesting thing to note now as well. I'm now running with a Stryd footpod. The first 3 runs in my "usual" location were very coherent in terms of outcome (within 10m difference overall distance which can easily be accounted for by switching road side, crossing at different points etc.) Then, after FW 7.6 update, the same consistency but based on a reduction of the total distance by 80 m (on +/- 5.6 km).
NB: This is with auto calibration "on", meaning the footpod calibrates using GPS data. The GPS tracks themselves were a disaster - cutting corners by 10 meters, weaving from the middle of a road to well into buildings, etc, etc. So for me, this is, better accuracy notwithstanding, another demonstration of lack of reliability of GPS tracking. I'm sure there's an explanation for it - but I'd love to know what it is, because it is galling, quite frankly, to have a device that records one result one day, and the next day, a completely different result under arguably the exact same conditions...
At least, with the footpod (autocalibration is now off), I now have more consistency across runs (I used to have variations of between 30 and 100m from one run to the next), a total distance that is more or less OK, and instant pace that actually shows instant pace.

Anyone seeing stronger GPS results now?

Can't see any mention of GPS fixes in the F5 update changelog but does seem better than when I first bought it.

Here's a recent test between F5 and a FR935 I've borrowed from a mate.

Ran 4 miles earlier to compare GPS on each watch. To my surprise the 935 and F5 measured 4.01 with exact same average pace and 1 mile laps were 1-3 within each other. Strava on my android recorded 4.1m. There was some tree cover throughout and pretty overcast. Current pace was a little worse on the F5 but not massive plus that can be fixed with the footpod. Each device used normal GPS and had 1 second recording set.

GPS still terrible. Actually, worst than 7.10.

Just to add one more piece of info: I soft reset the watch to deal with another issue (since update to 7.6 there is no sound on laps in activity anymore) and the GPS promptly did not know where it was anymore - as per usual. Except that this time even after 5 minutes it had not found enough satellites, so I started the run without (thank you Stryd for providing distance and pace data), and it took one km to lock on... track is complete though - dunno how it did that - and just as bad as the day before.
Boy.

Is it just me or is this "so I started the run without (thank you Stryd for providing distance and pace data), and it took one km to lock on" really not a good way for letting the GPS lock on? I mean, shouldn't it be best to stand still until it does lock on?

In my case it is a Fenix 5S, and the GPS locks on in less than a minute, and, so far the accuracy seems good.
Last time I did a small hike (work to home, same route every time), I had the GPS set for GPS only, and then it had some inaccuracies, which I have never seen using GPS + GLONASS.