Strava removes zeroes - (actually it doesn’t)

*** my hypothesis was wrong about removing zeroes, the thread went another direction ***

Just an FYI if not known already, particularly if you’re using Strava for analytics, they remove zeroes from your power data which effectively over states your power. This came as a bit of a surprise to me, thought I’d mention it.

Remove zeros where? In the average power? Weighted average? I’ve never seemed to notice zeros missing…

Do you mean if you pair a power meter via the app?

It doesn’t strip zeros when I upload a file from a headunits.

2 Likes

Not a strictly TR related post.

So I do a ride, record the data on my Wahoo, the Wahoo is including zeroes. I upload to Strava, if I look at the lap powers they are higher than what the Wahoo app data shows. This is because Strava removes zeroes.

Maybe if you tell it to? I know you can select that on a Garmin head unit, not sure about Strava.

FWIW I just checked 2 different rides and avg power is close enough to be considered equal between Garmin Connect, Strava, Training Peaks, and TrainerRoad. I don’t think it’s been removing zeroes because I’ve seen what long descent courses do to my Avg Power in all of those apps.

Strava definitely isn’t stripping zeroes on my files FWIW:

https://www.strava.com/activities/2630799349

They do have a different weighted average power calculation, and they calculate average power based on moving time, so if you have your headunit set to run constantly and not auto-pause, that might be where the discrepancy comes from.

3 Likes

Hmmmm I could be wrong :man_facepalming::man_facepalming:. But then why do my lap powers differ on Strava to my Wahoo!?

Average Power Calculations:

Average power gives your average power amount during the ride, expressed in Watts (a measure of how much energy you are placing into the pedals.) This is inclusive of the entire ride, and Strava includes zeros (coasting). Average power is most accurately measured by a power meter, though if you don’t have a power we give a rough approximation through our power estimator.

Strava processes the file exactly as it gets it. All of my head units agree perfectly. Check your device settings. You can set power to ignore zeroes or not on the Wahoo. It does not change the file.

1 Like

Ok, something is going wrong then.

There’s a massive thread on the strava forum of people losing their minds over this. In short, Strava has an algorithm that reinterprets the data brought in from your head unit and the conclusion is that none of the data strava is showing you is actually accurate. People siting different power numbers on head unit and strava.
But how do we know our head unit is recording it correctly in the first place? Or that the data is being delivered to the head unit accurately?
“Oh. My power numbers are falling after an hour into the ride… must be a faulty meter.”
Or you’re tired. There’s no way of knowing. But that’s another issue.
I’ll try to find the link. Please hold.

Here it is

1 Like

Any software doing analytics has to do some cleaning of the power data or the results will be iffy. Null or missing data points, suspicious zeros, spikes and so on are common, especially from trainers. My app Intervals.icu uses a number if different heuristics to fix things e.g. 210w, 212w, 0w, 213w, 202w while pedalling is very likely bogus data and it will replace the 0 with 212w. For power spike detection it compares the power curve for the ride against the curve for your FTP. Power-duration efforts that are 20% above the curve indicate a spike (or your FTP is set way too low) and are replaced with nearby data that “looks good”. Strava probably does some of this as well.

4 Likes

To give you an idea of the differences I am seeing, this was whilst doing 60s intervals last night. The first interval I was a bit slow off the mark, the rest I put the power down 1-3 seconds before the interval started on my Wahoo. Note that I am using a workout so not having to hit lap button.

Strava / Wahoo

611 / 563

634 / 621

587 / 559

581/ 562

586 / 575

536 / 524

Strava includes zeros correctly while moving, but if you stop and your device autopauses, it will remove those zeros if you select an interval that spans the stoppage, as if you hadn’t stopped.

However the data that goes to the Power Duration curve does include zeros from stoppages correctly.

1 Like

Maybe try and understand your problem first before spreading misinformation online as if it is fact. Subjects like this are complicated enough as it is!

Strava includes zeros for me too.

If those power numbers are for your 70s intervals, then there shouldn’t be any zeros in that data since you should be pedalling for the entire 70s, so that doesn’t explain your discrepancies.

Well to be honest, I had an issue, I checked online, found plenty of threads stating that Strava removes zeroes and how this issue had gone on for a number of years. This then resonated with what I was seeing in my own data. I then did more digging after posting and it seemed to indicate that they now include them so the situation is confusing. I was hoping to get other peoples input to back up my theory.

So this thread has helped by people pitching in with their thoughts but yet still I have an issue I cannot explain. Can you offer any help with the discrepancies?

1 Like

I understand what you’re saying, but your original post didn’t present it as a theory, it was presented as a fact. I’m not trying to be rude, it’s just good if users visiting this site aren’t given too much misleading information.

I’ll try and help if I can. Are you able to share your strava file?

If Strava removed zeros then there would be a really significant difference in the average power from an outdoor ride recorded on my Garmin (which doesn’t exclude zeros) on Garmin connect and Strava. The file is uploaded to Garmin Connect which is then synced to Strava. There really is no significant difference.

I race on Zwift a lot and dual record some of my races on both my Powertap pedals and my Neo. The files from the pedals go to my Garmin > Garmin Connect > Strava, The files from the Neo goes Zwift > Garmin Connect and Strava. The files from the Neo and the Powertap pedals are identical on Garmin Connect, Zwift and Strava. Strava doesn’t remove zeros.

Dual recording data

Here you go: https://www.strava.com/activities/2642231137

There are parts of your ride where the power on strava is reported at zero watts ( looks like you were coasting downhill as your cadence is zero).

Indeed, I’m not concerned with those laps. I’m focused on the intervals themselves not the recoveries.