Strava Power
 

When I first started using Strava, I would compare my "estimated power" to my friends power meter. Usually my numbers were crazy high compared to my friend who is almost the same height and weight as me.

Yesterday we rode a very tough 50 miles with a strong wind.

At the end of the ride, my estimated power was within 5 watts of his.

Is Strava getting better at guessing power, or did the wind just kind of equalize things?

Strava power is not accurate for me at all, even on steady, steep gradients. Occasionally it will get close on a steady endurance ride, but that is a relatively rare occurrence.

Strava power is totally wrong, unless you already own a PM. I have a PM on one of my bikes, and when I ride a different bike on similar routes, I find the estimated power number to be decent, not accurate enough to use, but reflective in the range of what I am doing. It used to be total crap, but now it seems to be better at estimating my power. However, if I ride a new route on a non-PM bike, the numbers seem to be less accurate (just based on what I know about my power when I'm riding). So to me, it seems like Strava estimates it more accurately when it has comparative real numbers to go by, and when I'm on a new route, it doesn't have anything to go by, and is less accurate. Also, my GF on the other hand does not have a PM, and her power estimates are WAY off. I wouldn't trust those estimates at all. PM's are relatively cheap now, it's worth it if you really want to know your numbers.

I've compared my power meter readings from two power meters to estimated power from Strava. As a general rule, the longer the ride, the closer the estimate is to actual measured power. For short efforts or segments, estimated power is useless.

Crux of the matter, though, is that if you care about power, you need a power meter.

If you and your friend are about the same size and weight, riding similar types of bikes with similar tires and you ride the ride together, your power will be +/- 10% or so. If Strava is telling you twice as much energy/power compared to the metered ride, then go with the meter.

I can almost guess from here... Let me try.

Hard ride in the wind, 50 miles.
So about 3-3.5 hours at 500 Kj/hour which is about 1500 Kj or 150-160 so avg watts or so...

139 watts = 500 kJ/hour.

Was that 50 miles into a headwind, or with a tailwind?

Strava would have no information about either drafting or wind direction. So, if you're drafting, it should estimate the power high, and riding into a headwind, it should estimate it low.

If it routinely is off, perhaps the conditions of your ride just brought it closer to reality.

I ignore Strava's power estimates. It doesn't use wind speed or direction, and using the app on my phone means the amount of climbing is usually off. On a calm day on a known route; you might be able to compare your own estimated power numbers. I think it is coincidence if it matches any real power meter.

A broken clock is still right twice a day. Strava probably just got lucky. Then again, did you and your buddy do equal amounts of pulling? If he spent 70% of the ride on the front, his average power should be a good bit higher and vice versa.

My pre-PM "Strava Power" numbers were generally very conservative. I will agree with posters above that the longer the ride, the more useful the estimated numbers. Also, the more rides you have logged into Strava, the more accurate/consistent the power numbers will be.

But not their calorie estimates. That is like a random number generator.

I spent more time pulling on the hills and into the wind. That's what has me wondering if the "algorithm" factors in wind.

With no wind it seems way too high.

185w. We must be a little heavy for your calcs. Lol

It felt like the whole ride was into the wind... In reality, unless the wind shifted I'm guessing about half was.

And to update my findings, we went on a hammerfest last night.


I pulled a lot more than the buddy I mentioned earlier.... but he averaged 180w and Strava said I averaged 220w.


Disregard any notions that lead me to believe the Strava power was getting better. It still sucks.

Strava is either high or low. It's rarely close and if it is it's by accident. :crash:

Seems legit, actually. You'd have to define 'a lot more' more specifically, though.

You realize their calorie estimates are just estimated kJ (from their estimated power) x 1.1? So you can't really say that their power numbers are good but their calorie estimates are bad!

Sure I can. Because I look at calories as kJ x 1.0, just like everybody except Strava.


I stand by the potential for accurate Strava power guesstimates. From a ride earlier this year, pre-PM, Strava guesstimates:

21.2mi, 1:10:22, 1,165ft^
262W Estimated Avg Power
1,108kJ Energy Output
Speed 18.1mi/h
Heart Rate 143bpm avg.
Cadence 83 avg.

From yesterday, PM measured:

22.6mi, 1:11:58, 384ft^
263W Weighted Avg Power
1,042kJ Total Work
Speed 18.9mi/h
Heart Rate 141bpm avg.
Cadence 82 avg.

So the Guesstimator, given enough rides, gets fairly accurate.

OK, but that makes Strava 10% high, compared with your approximation. That's different from being a "random number".

Sure, and the calorie estimates will be consistent as well, this is my point.

There can be a reasonable variation in rider power outputs due to exact bike set up and how aero they are on the bike, even if they are the same size and weight. This combined with wind, drafting etc etc makes the Strava number fairly useless.

on a related note: ever look at the actual data being written by the garmin?

On my 705 w Quarq, I noticed that the garmin will display (for example) ~400kj for my 45:00 minute moving time commute (48 min total time)

If I upload that to Garmin Connect, it displays as 210w average power (no kj on garmin connect).

When it flows over to Strava, it becomes 215w average and some additional normalized power is also calculated. Strava then shows 550+ kj.

HOWEVER, if I divide the original 400kj by the 45 minutes (moving) time, I get ~150w average power.

IT APPEARS THAT BOTH GARMIN AND STRAVA ARE DROPPING ZERO SAMPLES WHEN AVERAGING POWER. Which to me, is broken. They should count zero power samples as long as moving. Coasting should reduce your average power. Otherwise, I can ride at 400w for 30 seconds and coast for 60 seconds (repeat repeat etc) and end up with 400w average power rather than 166w average power.

It appears that Strava backs into KJ based on their overstated average power numbers, rather than using the sumproduct of (watts * seconds) from the original power file.

When I open/parse/average out the raw data from the TCX (xml) file using C#, I get very close to the ~150w that correlates to the actual kj displayed on the head unit itself.

WTF???@!?!??@!??@@??@?@?@??

That's why you don't use Strava (or Garmin Connect, ugh) if you are actually using your data for something.

I don't see how Strava could accurately estimate power consistently over time, like many previously wrote here, if without a strain gauge there are just to many variables that are not part of the data collection in Strava.

Because they have access to data from riders that do have power meters-- not to hard to employ a formula for a rider of XX pounds riding at XX speed on XX grade when they undoubtedly have matching recorded PM data to work from.

I typed that before realizing the post I was responding to is from a shill. Shame on me! :roflmao2:

"then i pounced and i pounced again"