Question for Strava Experts

[Vamoots58]

For several years I've noticed a consistent discrepancy between different piece of training software, the outlier is always Strava. When I come in from a ride, Strava always shows a longer moving time for the ride than Garmin Connect or Training Peaks. The latter two always match each other to the second. When I do a Zwift ride, all three match within a second (rounding notwithstanding)?????

[madsciencenow]

I'll add that strava always screws up my power number relative to garmin. It routinely pisses me off because I'm retentive and it's always lower.

In all seriousness, I believe that strava uses a different algorithm to process the data. I'm not sure why this is but I feel like it's been discussed here before but I'm not certain if this is me mis-remembering the conversation here or with riding friends off of the forum.

With regard to Zwift there's really no way for strava to process the data differently. I can't really explain this but think it has to do with the gps data which is virtual for Zwift and actual gps data points from a real ride.

Hopefully someone else can provide a more coherent explanation.

[CDollarsign]

I thought strava has 'cafe time' or time not spent moving. Maybe your other apps are just using moving time vs elapsed time? This would also support all three matching with a zwift ride.

[MattTuck]

My hunch is that Strava is pulling the raw GPS data and doing a calculation. Garmin is probably doing the same, and then Training peaks is probably pulling from the processed Garmin data.

Are we talking off by seconds or minutes? It seems fairly straightforward to determine when velocity is greater than some threshold value. So the only difference I can imagine is on things like stop lights, etc. And then I'd imagine it is just an arbitrary difference between "wait until he is stopped for 2 seconds" to stop the "moving time" vs. "wait until he is stopped for 6 seconds". May also factor in cadence in someway.

[Vamoots58]

the difference is a minute or so per hour of riding. NOT the end of the world, but periodically, it gets me thinking (always a dangerous thing)...Thank you for the thoughts. With respect to power numbers, I trust my Vectors more than the Zwift power that I get from the KICKR, which generally seems a little lower.

[rnhood]

The reason that the power numbers, normalized in particular, do not match is because Garmin licenses the specific sum of squares (RMS) formula that Training Peaks copyrighted (or patented), while Strava merely applies a simple weighting correction to the average power number. Strava is not meant to be an accurate training tool so they are not interested in paying a license fee. Their weighting formula is consistent among all Strava reporting, so it still can be used for relative efforts and training, but it's not in line with industry standard root mean square formula.

Strava does use GPS in reporting moving time, while Garmin takes the moving time figure from their device, assuming the auto pause feature is not active. So there will be a few seconds or so difference in these readings. Not a big deal.

There is also a "Time" field on Garmin, and this is the total ride time if the auto pause feature is not activated. If you run auto-pause, and I think most do, then this number has no significance that I can tell. Its usually only a few seconds different than "Moving Time", and I'm not sure what accounts for this small deviation. But it's really not important. If you do not use the "auto-pause" feature, then its reporting will be accurate per the definition. You can find these definitions on the Garmin or GC website.

Verdict: If you want accurate info, use Garmin Connect. Strava is first and foremost a social app. While it does have performance info, and some of this is actually quite accurate, it's not going to be your best option for serious training.

[Vamoots58]

Great explanation - Appreciate the information!!

Originally Posted by rnhood

The reason that the power numbers, normalized in particular, do not match is because Garmin licenses the specific sum of squares (RMS) formula that Training Peaks copyrighted (or patented), while Strava merely applies a simple weighting correction to the average power number. Strava is not meant to be an accurate training tool so they are not interested in paying a license fee. Their weighting formula is consistent among all Strava reporting, so it still can be used for relative efforts and training, but it's not in line with industry standard root mean square formula.

Strava does use GPS in reporting moving time, while Garmin takes the moving time figure from their device, assuming the auto pause feature is not active. So there will be a few seconds or so difference in these readings. Not a big deal.

There is also a "Time" field on Garmin, and this is the total ride time if the auto pause feature is not activated. If you run auto-pause, and I think most do, then this number has no significance that I can tell. Its usually only a few seconds different than "Moving Time", and I'm not sure what accounts for this small deviation. But it's really not important. If you do not use the "auto-pause" feature, then its reporting will be accurate per the definition. You can find these definitions on the Garmin or GC website.

Verdict: If you want accurate info, use Garmin Connect. Strava is first and foremost a social app. While it does have performance info, and some of this is actually quite accurate, it's not going to be your best option for serious training.