Bike Forums

Bike Forums (
-   Training & Nutrition (
-   -   'splain this to me, Strava (

Myosmith 01-30-14 11:21 PM

'splain this to me, Strava
I rode a short loop outside of town four times back to back today so virtually the same conditions, same bike, etc. I have one segment on that loop which today was into the wind riding on a snow covered shoulder. I averaged 10.1, 10.5, 10.9 and 10.2 mph on that segment during the four times around the loop. Strava said my average power on that segment varied from 58 to 109 watts, but get this, my slowest time (10.1 mph) showed the highest wattage. Looking at the analysis I averaged 81 watts (which I can believe as this was a slow, recreational winter ride) but my peak was over 1,000 watts, three different times at three different points in the loop :twitchy: Strava also said that I had a max speed of 19 mph which I also know to be inaccurate as with the snowy road surface and my on bike computer I know that I didn't pass 15 mph even with the wind at my back.

I've noticed this before on other rides, but this is the first time I had four virtually identical passes on the exact same segment in such short order. I don't expect 100% accuracy but this makes me question any or all of the wattage estimates that I have gotten from Strava.

bmontgomery87 01-31-14 06:06 AM

strava drops the ball w/ gps connections from time to time or something.
I went out and did some sprints one day and strava said I hit 38 mph. While I'm thinking it was much closer to 28, and it appeared that way on the graph I looked at as well.

chaadster 01-31-14 06:30 AM

Strava can only estimate power, not measure it. It is, of necessity, inaccurate. To measure power you need a power meter.

Also, Strava is great.

Myosmith 01-31-14 08:32 AM

I like Strava and use it to document the vast majority of my rides. Don't know if I'd go so far as to call it "great" but it is probably as good as anything out there. Variances are expected when power is estimated but a nearly 100% variance (58 > 109w) on the same short course under the same conditions at virtually the same speed (10.2 and 10.1 mph) isn't an estimate it's a wild arse guess. What algorithm could they be using that inputs the same bike/rider weight, the exact same elevation/location, and virtually the same speed and comes up with the power "estimate" for the lower speed being nearly double that of the other pass? I could understand it if there was some glitch where it misregistered the speed, elevation, etc. but all the other data is there and matches so you would expect the results to be at least similar.

Strava's "estimates" on another part of the same ride had me cranking out 1,227w at 19 mph on a flat, straight section of road. I know the speed at that point was incorrect, but even if I had actually been doing 19 mph at that point, there is no way the calculations should come up with over 1,200w for my weight, on that road, at that speed. I have ridden that road at that speed and estimate my power output to be more in the range of 200w and I'm willing to bet my off the cuff guess is far closer than the results I got on Strava yesterday. When power is estimated sure there is an acceptable variance, but on this occassion and quite a few others, I have noted that Strava hasn't even been close.

GeorgeBMac 01-31-14 09:47 AM


Originally Posted by bmontgomery87 (Post 16455796)
strava drops the ball w/ gps connections from time to time or something.
I went out and did some sprints one day and strava said I hit 38 mph. While I'm thinking it was much closer to 28, and it appeared that way on the graph I looked at as well.

... I don't think it' the algorithm, it's erratic GPS input. That problem may be due to the phone or it might be due to Strava's interface with the phone... Interestingly, the IPhone 5S that was introduced last September added a second processor to aid apps like Strava. i have no idea if it would fix (or even affect) this problem, but it seems likely to me...

While it's true that the algorithm is an estimate of power, that does not mean it's estimate should be erratic -- and the same with it's erratic speed readings. Actually, I would suspect that both of those problems related back to a problem with the GPS or its interpretation by Strava.

Darth Lefty 01-31-14 10:02 AM

I'd be interested to know if Strava is only using the GPS or if it's using the accelerometers too. If it's just the GPS, it doesn't really know how fast it's going, only where it was each time it solved the equations for the satellite signals. That's never been perfect (ask a geocacher) and deriving the speed, incline, power etc. is only as good as the input data. A couple of wonky points, due to natural error, a swerve, a shift, riding under a tree, etc can really throw it off. Averages should be pretty good, but the peak should probably be ignored.

RPK79 01-31-14 10:10 AM

Wattage estimates on Strava are for entertainment purposes only.

Carbonfiberboy 01-31-14 11:20 AM

If you are a Strava subscriber or whatever they call it - pay money - there's an analysis button. If I go there and look at calculated power - I don't have a power meter - I see that the software has created power surges, I suppose one for every time the Garmin updates position. The exact location of these surges vis a vis the ends of the segment will cause different power estimates, especially on a short segment.

Those high wattages are the peaks of those surges. If you really want to track power, you have to have a meter. IDK about the 19 mph. Haven't had that happen.

For long segments, Strava watts estimates track very closely with those provided by online bike calculators, without the hassle of inputting data into said calculator. They should, because they and Strava use the same formulae.

OneGoodLeg 01-31-14 11:27 AM

Map My Ride also suffers from a similiar lapse in GPS connectivity. I commute the same route pretty much every day, yet the mileage on every ride is always slightly different, usually by 10ths or 100ths of a of mile, but still.

Myosmith 01-31-14 06:02 PM

Strava does a good job of recording ride routes and elevations. The average speed is pretty close to my cycling computer the vast majority of the time. Wattage has always been more of a crap shoot. I can ride the same bike, on the same route, at very close average speeds and often have a 25-30% variance in power estimates. On a few occasions, like yesterday, the readings are all over the place. The segment in question from yesterday is pretty short, just a couple of miles long, so that may be part of the reason. I'm not too concerned as if I wanted accuracy I'd get a power meter, but it still strikes me as funny that Strava can input and record the same bike, rider weight, location, elevation, and speed but still come up with such a wide variance (almost double) in power estimate. Maybe I'm just not tech savy enough, but it would make more sense to me if one of the input factors was significantly off, but when you start with the same numbers, why wouldn't you get the same answer?

nkfrench 02-02-14 03:32 PM

My garmin seems to overstate elevation changes so I end up with double the "climbing" each ride that others see. That definitely skews power estimates.
I've been advised that my garmin mounting should be changed to be more level (parallel to road surface). It seems to affect the barometric pressure readings used to calculate elevation.
Barometric pressure also will change for a given location as weather changes. I've had many rides where the elevation of my parked car changed 50 feet between ride start and the finish an hour later.

I upload my rides to strava to get the analysis. I gave up on the phone app so I could get the ant+ cadence and HRM data and save the phone battery for phonecalls.

All times are GMT -6. The time now is 06:11 AM.