Why does my fenix 6 pace jump about so much? The pace it says is no where near the pace that clicks on every mile

Not sure how the comment is moot…that review shows a number of watches, Garmin included, that were unable to achieve accurate pace figures. Your data also backs up that assertion. It doesn’t matter where that data comes from when you are using calculated averages and smoothing. 1) There are very real limitations in using GPS on a small receiver, with small antenna and in a low power setting. 2) Trying to capture “accurate” movement data to augment the GPS signal is going to be suboptimal when the watch is strapped to your arm - arm movement doesn’t necessarily correlate with distance or speed. When the data coming in is inaccurate, the data going out is going to be inaccurate. 

A foot pod directly measures speed and distance through 3D space. In the case of a Stryd, it is one of only a handful of devices using 9 axis…. 3 axis accelerometer, 3 axis gyroscope and 3 axis magnetometer AND has a barometer/ altimeter. Most foot pods use only a 3 axis accelerometer.

I don’t see how you can possibly make a case that a GPS watch can A) accurately and precisely measure pace and B) be anywhere near as accurate or precise as a Stryd foot pod measuring in 9 axis directly at the foot.

Happy to be proven wrong but physics suggests it isn’t possible right now…..and every review of every GPS watch concurs.

I feel like we've already had this discussion. I used Suunto watches for 6+ years. I don't remember thinking even once - "the pace of this watch is absolutely horrible!". Sure, it wasn't super accurate, but it was good enough for me to not notice it because I thought it was reasonably accurate.

In fact I invite you to go to Suunto Forum and see how many treads you'll find where people complain about instant pace. I can tell you - there is one thread which was actually started by me, where I complained about Suunto 9 pace being 15 seconds per mile too fast, and that is when running a trail race in densely forested park. That kind of accuracy is unthinkable for Fenix 6. When I run on similar forested trails it regularly off by 0:45-1:30/mile. Suunto 9 uses the same Sony GPS chipset, and it also uses built-in accelerometer to implement what Suunto calls FusedSpeed. So yes, it can be done, but Garmin obviously doesn't care because majority of Fenix users use their watches as glorified Fitbits, and it doesn't matter how accurate pace is.

Ambit 3 Peak pace was even more accurate. I remember running a road marathon with it and I had no problem maintaining the pace within 5-10 seconds/mile, and it was consistent with mile splits too. 

Now regarding Stryd. I don't want yet another device strapped to me that can run out of battery charge or get lost. I sometimes run for 6-12 hours, or longer, and getting worried whether Stryd would have enough charge to finish the run is the last thing I want. I don't even mention running through mud, puddles, deep snow, crossing creeks, kicking branches with my feet - plenty of opportunities to lose it.

If I have a chest HR monitor recording heart rate and Stryd recording pace and distance, what does this expensive $1000 watch do then? It is reduced to a simple recording device, and any cheap watch could do that. Why do I have to pay for all the R&D and the latest technology, if I am not going to use that?

Thanks for the reply, I really do appreciate the time that you took to respond to me. I too have had a number of GPS based watches from a range of manufacturers including Suunto and Polar.

I don’t doubt that under ideal conditions, you may get a pace reading that you are happy with. What I am saying is that it’s not possible to replicate that, every single run under varying or difficult conditions. Often times, you don’t know when those conditions will hit and your training is then left with inaccurate data and you’re likely over/ under training.

Sure, you may not want to buy a foot pod - I didn’t either and really wanted my watch to magically be as accurate and precise as a foot pod but it just isn’t. Same with optical HR…it just isn’t a comparison against a HR strap that measures electrical activity directly and not a pulse that may or may not get mistaken for movement and obscured by cold extremities/ wrong skin colour or band that is too loose/ tight. There’s too many assumptions and smoothing and algorithms rather than direct measurement.

The fact of the matter is, if you want things “to just work” every single time, you will need to pay for the Stryd and HR Strap. I just go for a run….it just works, data is accurate and precise consistently and I don’t feel left frustrated.

If you can live with data that may or may not be accurate/ precise depending on cloud cover, satellite coverage, buildings, trees, rain, cold, humidity, strap too loose/ tight, dehydration or what wrist you wear it on, then just the watch will be fine. Knowing what is and isn’t accurate becomes a frustration when you have nothing accurate to compare to.

I have directly linked to two articles and reviews of some very in depth testing across multiple GPS watch manufacturers.

This quote was from that testing and isn’t mine, even though I do agree with it….

“However, none of the watches have GPS accuracy that is good enough to be used for displaying your current pace.”

This included Suunto Ambit with FusedSpeed (I still have one of those lying about), which has a positively massive antenna in comparison to the Fenix series. 

TheOtherPhil said:

Great meme, full of verifiable data.

Let me make things easy for you….

“Under good conditions most of the watches are remarkably good, but when things get a little tough the differences become more apparent. However, none of the watches have GPS accuracy that is good enough to be used for displaying your current pace.”

fellrnr.com/.../GPS_Accuracy

That's very strange conclusion considering that 3 devices on the list provide data comparable to foot pods (Polar V800, Suunto Ambit3 Run, Suunto Spartan Trainer). All 3 provide data which is in line with 3 out of 4 top results (top 4 being foot pods).

Fenix 6 is not on the list, but Fenix 5X is - with pace and distance errors 2-3X compared to Polar/Suunto watches and Adidas/Garmin/Milestone pods. If you compare Fenix 5X to Stryd error goes up to 5-6X (maybe that's how Fenix 5X got it's name...).

FFS Apple Watch 3 is doing much better than Fenix 5X...

But let's slap "inherent limitations" and call it a day...

iLikePotatoes said:

But let's slap "inherent limitations" and call it a day...

I've posted about V800 being very good for instance pace, but I got slapped down by what I'll call the usual suspects (they hunt in packs I think!), when I linked to Fellrnr - to be fair the 'slappers' I didn't link to that chart. But at the time of posting I knew I'd seen something there that put the V800 really close to footpods.

I have actually done a run using a Polar V800 with F6 Pro / Stryd together on separate wrists - the results are interesting. But spoilers: the V800 is very close and very usable for instant pace.

I should caveat this by saying that I essentially run in clear open sky areas, my results could and would be different in built-up urban areas.

But the 'inherent limitations' thing is BS and it's disingenuous to say that because Garmin doesn't do instant pace well, no one else does IMHO, although GPS pacing with a watch on a wrist moving around is never going to be perfect - however, techniques to improve GPS tracking via compass and accelerator inputs have been known for years. My previous employer was doing this around the time GPS started to become mainstream in mobile phones, so it's not a new technique. 

Because the distance measure is nonsense.  It does not work.

iLikePotatoes said:

That's very strange conclusion considering that 3 devices on the list provide data comparable to foot pods (Polar V800, Suunto Ambit3 Run, Suunto Spartan Trainer). All 3 provide data which is in line with 3 out of 4 top results (top 4 being foot pods).

It’s not a very strange conclusion at all when the very table you linked showed the TOP performing GPS watch is the Polar V800 but it still had an average pace error of 28s at a 9min/mile pace and an error of 17s at a 5:30/km pace (95% CI)….and that’s best case scenario, not difficult conditions. Although it’s distance reading is good, it’s precision and repeatability is poor in comparison to a Stryd. The Stryd is 4x better than the top watch for current pace and many times more than the bottom half of the table.

That’s likely where the conclusion: “None of the GPS watches tested had accuracy good enough to show CURRENT PACE” came from.

And even though Polar had the top performing watch, it also had a number in the bottom half of the table along with a number of other manufacturers such as Apple, Garmin and Suunto (despite using FusedSpeed).

I run the same course every day, I leave my device to GPS soak for at least 30 minutes before I run.  As I've done the course 100+ times I have a fairly good idea when I hit km marks.  I'd say at least 1 - 2 days a week it will "lose" about 500m - 700m over a 8.25m course, I know by the end of the first k if I'm going to get nonsense readings.  If you cannot rely on the distance measure then the pace measure cannot be accurate.  

As an experiment I did a half marathon over with other people using a variety of devices or even just free phone apps.  My Fenix lost almost 2k with every other device being the same to within 0.5%.  

The pace data field on the watch is not directly connected to the measured distance. I think that the lap information is connected to the measured distance but I'm not 100% sure how that works. 

If we look in the .FIT files for an activity we have 4 data types (or 5 if we count step_length combined with cadence) we can use to calculate the pace for the activity.

1. gps_metadata.enhanced_speed [m/s] (Some kind of "raw" data from the GPS but its not accurate if compared to GPS position data. Smothered data. Don't know what its used for.)

2. record.enhanced_speed [m/s] (This data is shown in the pace data field of the watch. Really smothered data. This data is also used for the pace graph in Connect.)

3. record.distance [m] (The measured distance. I think this data is also used to calculate the lap pace.)

4. record.position_lat/position_long [semicircles] (The GPS position for each data record. Can be used to calculate distance delta between the positions. And that can be used to calculate the pace. The altitude data is also there if you want to calculate "3D speed".)

When I look at my latest slow run on a 6.189 km gravel track (distance controlled with odometer) and 38:25 minutes I can sum up like this:

1. gps_metadata.enhanced_speed [m/s] sum up to 5.295 km (14.44% too short).

2. record.enhanced_speed [m/s] sum up to 5.921 km (4.33% too short). Remember, this data is used to show the pace on the watch.

3. record.distance [m] sum up to 6.046 km (2.31% too short).

4. record.position_lat/position_long [semicircles] (own calculation without using altitude data for 3D speed) sum up to 6.102 (1.40% too short).

But, the above data does not explain why I have these strange pace dropouts on some of my runs. They seem to happen on the same places on my runs. Next run I will wear both my old Forerunner 310XT and my new Fenix 6X Pro Solar so I can compare them afterwards.