I did some 2 min intervals today, and the distance shown differs from the calculation of average cadence × average stride length. What should I trust more? Interval 1: 0.47km 191spm × 1.21m = 0.462km Interval 2: 0.47km 177spm × 1.20m = 0.425km Interval 3:

What was your GPS setting? Or was it on treadmill? Do you wear anything besides the watch (stryd, foot pod, RD pod, some chest strap with RD)?

It was set to the default of Auto Select. The session was done on a flat course with minimal trees. I did not wear any other tech.

The 1st interval has 8m difference, which is not bad. The 2nd is worse though. I think that it can be caused by different rounding and the fact that the distance is calculated by the GPS and not based on steps. It would be interesting to test it indoors with GPS disabled.

I would always trust distance from GPS (minus any obvious GPS errors) over avg cadence x avg stride length.

I think average cadence is reliable enough, but average stride length appears to be an average of all the instant stride length values. Since instant stride length is based on the accelerometer and not GPS distance, then I would actually expect there to be a discrepancy between GPS distance and cadence x stride length in general. (Since it's basically 2 different ways of measuring distance.)

And if the accel was reliable enough that you could measure distance by cadence and stride length with results just as accurate or better than GPS, then you'd think that the accel would be the preferred source for distance (and instant pace). But on the contrary, the watch prefers GPS for distance and speed [*] outdoors.

Btw, a long time ago, when there was no instant stride length in Garmin devices (neither from wrist nor chest strap), average stride length was actually calculated by dividing total distance (usually from GPS) by average cadence. So in that case, you would never see a discrepancy since average stride length was simply directly calculated from GPS distance.

To underscore how this stuff isn't necessarily an exact science:

- note that average cadence on Connect, Strava and runalyze can all be slightly different for each of the 3 platforms, for the same activity

- average stride length is also slightly different on Connect vs runalyze, for the same activity

TL;DR if 2 related metrics (distance and stride length) are measured by different instruments (GPS vs accel), it's almost expected that they won't match up.

--

EDIT: although I think outdoor instant speed/pace is a fusion of GPS/accel data.

This can be seen a couple of ways:

- for very short laps, average lap pace can be faster than the max lap pace (because the former is based on distance while the latter is based on instant pace)

- if you sync your Connect activities with runalyze, you'll see that the max speed runalyze calculates for an activity is sometimes much faster than Connect max activity speed (because the former is probably based on GPS data while the latter is based on instant pace)

I would trust that you ran for the interval length.... 2 minutes 

;-D

Neither GPS or Cadence sensor are 100% accurate, so hoping that they match is a fruitless endeavor.  Just stick with the GPS, these watches have been proven very accurate.  If you were in the city/jungle... and the GPS looked wonky, i suppose you could use the steps calc, but.... in that case why?  you ran hard for 2minutes... that's the interval that matters for physiological / training benefits.  If its a benchmark / TT test... pick a good location! :-)