I wonder if they use other members power data for this.


Did you read this or is it a theory?

I really doubt Strava uses data from rider's with a PM to estimate other riders power. I think they use a relatively simple formula like can one can find on-line where the variables are rider's weight, speed and gradient - similar to www.bikecalculator.com Just too much variability even with folks who have PM's - mainly the difference between riding solo or in a group.


In my experience when comparing my pre-PM data on Strava to post-PM data, Strava was relatively accurate when I rode a steady state, solo ride with very little wind. On a steady climb, Strava was very accurate. When riding in a group Strava estimates are completely useless.

Gosh, if only there was an entire mathematical discipline devoted to making sense out of variable data or a company that had access to tens (hundreds?) of thousands of users with power data...

BB

Power data is useless for comparison purposes if you don't know the circumstances - a heck of a lot easier to go 24 in a group than solo. Same with non-power data (useless if you don't know the circumstances). That's why I think they take the simplistic approach.


If you look objectively look at Strava's power estimates and know what you are looking at, it seems pretty clear that they are not using their data to adjust for group rides. Just pull up a fast group ride and look at the crazy high estimated power numbers.

All I was arguing was that using fairly simple statistics, it would be possible to create an algorithm that would capture a reasonable estimate of power. That would in no way, shape, or form replace having a PM for actual measurement, but it would (and does) give a rider a general notion of what kind of power he's generating and allow him to spot trends over time.

Once again, that doesn't replace a power meter for anyone serious about training.

But for a casual rider curious as to how his power numbers stack up, it'll give an idea.

BB

I emailed their support about a year ago; their calcs don't include obvious things like cadence. They're a software company, you'd think they'd spend a little effort on the calcs. Unless the PM industry secretly pays them not to, in order to sell more PMs. I've seen stranger.