Strava power estimations: how accurate are they?

baribari

Maximum Pace
May 28, 2010
492
113
63
Fukushima
#1
So several weeks after starting to use Strava I finally look online to find that there are some nice advanced analysis tools... Long story short, how accurate are the power estimations? I assume the spikes are miscalculations, since I some as high as 1,300 watts on a hill climb. Do the calculations actually use your weight and the slope of the hill?
 

Sibreen

Maximum Pace
Jul 23, 2010
563
241
63
Hanno, Saitama
#5
As @theBlob says, not very accurate. I compared Strava's guesstimate to an actual power meter (Stages) a few weeks ago here: The Stages pages
I thought the comparison of Stages with the longer, endurance segment showed the Strava guesstimates in a good light. Apart from one anomaly, they were all within 5% of your power meter result.

Would be interesting to see how results compare on a long climb (say Shiraishi).
 

joewein

Maximum Pace
Oct 25, 2011
2,435
882
133
Setagaya, Tokyo
joewein.net
#7
As @theBlob says, not very accurate. I compared Strava's guesstimate to an actual power meter (Stages) a few weeks ago here: The Stages pages
The GIGO principle of data processing applies: Garbage In, Garbage Out. Without a power meter, Strava's power ratings can be no higher quality than their elevation data.

On the 20 second blast up the hill comparison, I'd be curious how accurate Strava's altitude gain data is. Often their segments contain questionable elevation data, e.g. if the initial ride was by someone without a barometric altimeter. I found it not unusual to see segments listed as supposedly 17% climbs when they are essentially flat segments that people routinely do 35+ km/h averages on.
 

Musashi13

Maximum Pace
Aug 27, 2012
1,774
1,105
143
41
Ichikawa, Chiba
#8
On the 20 second blast up the hill comparison, I'd be curious how accurate Strava's altitude gain data is. Often their segments contain questionable elevation data, e.g. if the initial ride was by someone without a barometric altimeter. I found it not unusual to see segments listed as supposedly 17% climbs when they are essentially flat segments that people routinely do 35+ km/h averages on.
That would explain some of the CAT 4 climbs near me, interesting.
 
Dec 16, 2012
605
824
113
Tokyo
#9
@joewein I can't remember if I set the segment or if someone else did, but the elev. profile and grade match the actual hill quite well. Saying that, the Strava app shows me climbing a sheer 50m-high wall on my commute as I pass over the imperial palace moat.
 

theBlob

Bokeh master
Sep 28, 2011
2,864
1,450
129
...
#10
Anyway does it really matter? The real advantage of having a power meter happens when you are riding not after. Knowing an estimate of what you put out after the fact is kind of useless. And really pales in comparison to what strava offers in terms of definitive data. i.e. Speed, distance, ranking.

The things that effect a power reading are largely outside starvas control. Because it is a guestimate Strava can have no idea as to the following factors that will substantially power vs speed
1) Atmospheric conditions (Temp/ Air pressure, wind assistance/ drafting)
2) Accurate elevation data. (All devices read differently, some substantially so)
3) Accurate weight. It relies on the user accurately calculating their weight on that particular ride at that particular time.