So I have been running along the lake in downtown Toronto, where the CN tower is located. The CN tower, along with a wind turbine and other strucutres (railways, etc) may contribute to GPS signal being lost at times. However, I have consistently noticed when I run the exact same 8 mile (4 miles out and back) route day in and day out, my garmin (Forerunner 735XT) shows that I have run 8 miles, whereas once the run is uploaded online, strava shows this route to be 8.5 miles. EVERY SINGLE TIME.
How does strava “compute” its distances? Does it determine which segments I have run and add them? Every single meter of the course that I run is included in some sort of a strava segment because the lakeshore is a hot spot for runners and cyclists.
Strava seems to, with certain files, do some strange post-processing. I know people who, for instance, have files that show different average power values on Strava vs. Garmin or TrainingPeaks.
From what I’ve heard, Strava seems to always be the outlier. I.e., TrainingPeaks will always show the same as Garmin, whether for distance or power or whatever. Strava, on the other hand, seems to give different values.
Why exactly this happens I do not know. But just know that you’re not alone in experiencing this.
I cannot imagine Strava is simply stitching together segments, because you’d rarely have 100% of every ride covered with segments. But if certain segment distances have value overrides, I can see that affecting things. I.e., even if Strava isn’t adding Segment_A+Segment_B+Segment_C, if your GPS track has a distance of Xm for a segment that Strava has registered as Ym, then I can see it maybe overriding that. But it’s pure speculation.
I echo trusting TP or Garmin vs Strava. I use Strava for “fun” and TP for my actual training log. The one thing that is WAY off with Strava are power values if you’re using a power meter. I have no idea how Strava calculates power from the data you upload. TP and Garmin always seem to match regarding power data.
It might make sense to discuss how different devices and software work to calculate distance.
Here are my beliefs about how stuff works:
Garmin uses the distance between the watch/device and the Garmin satellites to calculate the distance traveled by the watch.
Problem- if the satellites reception is poor and/or the satellites are not well triangulated- the values will be off.
Map-my-ride/run- has maps that were originally created with accurate satellite data. Specific routes can then be transposed onto the map (using a device) or you can draw your own route on the map. Distance are then calculated with the accurate map information.
Problems- the routes that you draw on the map assume that you took the most logical route. Routes that are drawn by poorly triangulated devices (Garmin for instance) will show that you ran off course.
Strava - I assume Strava works by averaging the data from different individual devices.
The problem with this is that the most popular devices will dominate the results. And the most popular devices are NOT necessarily the most accurate.
Training peaks - I think it accepts the data from ones device and imposes the data on accurate maps. Unlike map-my-run, however, it does not use the map to calculate distances. (Unless you hit the recalculate for elevation button and then only elevation data is used to recalculate).
Bad data caused by poorly triangulated devices are taken as gospel.
But the source of the error is visible as the map.
The short answer to this puzzle is … believe nothing.
Look at the map of the route you supposedly ran on the map…
Does the map have you running over the rooftops of houses, running in the middle of the river, and swinging very wide (or cutting) corners?
Does the elevation data seem plausible?
Does the route look like the one you actually ran?
How to estimate real distance when your device is producing a bad route map:
Ignore the rooftops, fences, rivers or forests (that you effortlessly passed over).
How much slower (or faster) would it have been, if you ran the route the watch says that you ran?
Draw the route that you actually ran on map-my-run. Adjust for the side of the road you ran on and/or other irregularities of your real route.
Per Garmin watch, I ran 13km in 1:08:08 (easy run). My total moving time is 1:08:08, but my total elapsed time is 1:11:01 due to the lengthy red traffic lights during the first and last km of the run. This data is consistent with what shows up on my Garmin Connect log online.
When the activity automatically transferred to Strava, it showed that I ran 13.7km in 1:08:08, with last km split being 3:20min/km, which is not true because I know I didn’t push myself that hard.
So I deleted the run from Strava. Then I exported the run from Garmin connect in a gpx format and manually uploaded it to Strava online. The result on Strava this time showed that I ran 13.7km but in 1:11:01 (my elapsed time). The average pace of this version of the run is closer to what my watch shows, but I am still really annoyed at the extra .7km that shows up that I didn’t run!
I tried splitting the run on Strava to cut out the .7km from the run, but it produces weird results.
Anyway, not a big deal for a recovery jog and life goes on, but still…lol
Never trust strava with data. Ever. It’s all massaged and their data science is horrible.
I agree completely.
Just wondering…why in the hell would power change? If I have a power meter, why does Strava somehow end up with different power recordings? Why would Strava make things difficult for itself? Just duplicate the data from my power meter. Jeez!!
Strava is trying to implement some sort of solve for all of the various files from all of the various devices that get uploaded to its service. Not every device/manufacturer passes the GPS data the same way or in the same file format. Add in the complications of some using auto stop/start and other various settings and I can see how it quickly becomes a shit show of ingesting and interpreting the data correctly/accurately.
I’ve had irregularities with Garmin files across various devices, using the Strava app on an Apple Watch, using a few different Samsung products that include GPS (and a few different apps on the device itself). And they all export the data in different ways. But strava has to ingest that data, interpret it and then display it accordingly. Seems thats not so easy. Especially when factoring the complications of validating segment data within a given workout. Across all of those various devices and apps within devices that can export a file to strava.
Never trust strava with data. Ever. It’s all massaged and their data science is horrible.
I agree completely.
Just wondering…why in the hell would power change? If I have a power meter, why does Strava somehow end up with different power recordings? Why would Strava make things difficult for itself? Just duplicate the data from my power meter. Jeez!!
That’s the annoying thing. I can bike on zwift while capturing data on my bike computer and all three will come up with different numbers, the calories on Strava also seem way higher than Garmin.
Look at the map of the route you ran on Strava. Enlarge it and see if it shows you running a straight-line or zig zagging. According to Strava I like to run from one side of the road to the other side a lot. But it makes me feel good about my pace and run times. Just can’t beat those PB’s in a race for some reason.
Strava is trying to implement some sort of solve for all of the various files from all of the various devices that get uploaded to its service. Not every device/manufacturer passes the GPS data the same way or in the same file format. Add in the complications of some using auto stop/start and other various settings and I can see how it quickly becomes a shit show of ingesting and interpreting the data correctly/accurately.
I’ve had irregularities with Garmin files across various devices, using the Strava app on an Apple Watch, using a few different Samsung products that include GPS (and a few different apps on the device itself). And they all export the data in different ways. But strava has to ingest that data, interpret it and then display it accordingly. Seems thats not so easy. Especially when factoring the complications of validating segment data within a given workout. Across all of those various devices and apps within devices that can export a file to strava.
It isn’t tough to write logic to match whatever device the data is coming from. Sporttracks and training peaks have no problem matching my Garmin or wahoo device perfectly. When I cycle Strava has never once matched my device. Instead of just taking the device data Strava is taking the raw and recalculating it for some reason. And it’s worthless.