MattTheHat 

Fair warning...I like data.

I installed a Stages power meter last night and took my first ride with it this morning. I'd often wondered how accurate Strava estimates were. Turns out, probably not very accurate. The ride this morning was only 20 miles. I recorded the ride in Strava on my iPhone (without the power meter) and on my Wahoo Elemnt Bolt with the power meter so I could compare. The data was depressing, encouraging, enlightening and interesting all at the same time.

To start with, Strava's estimates aren't too accurate. I'd read folks estimate their results were within 7-10%. That was nowhere near the case with me. My ride had 15 Strava segments. Only on one segment did Strava and the power meter agree and this was the longest segment of them all at 3.2 miles.

Strava estimated more power than the power meter on three segments. Interestingly two of the segments were back to back. One a max effort on a short climb (481 Watts vs. 527 watts) and the next was a short downhill segment where I was so wiped out that I didn't even peddle much (59 watts vs. 89 watts). So Strava read 9% high on the up hill segment, and 50% high on the down hill segment. On the third segment, 162 watts vs. 169 watts, or 104%.

On all the other segments, the power meter read higher than Strava. Between 111% and 195% high. That seems like an awfully wide margin. Interestingly, if I throw out the first four segments of the ride, the range is between 111% and 128%.

The depressing: when the legs start to hurt and I back off a bit, the power number is, um...unimpressive. Also unimpressive are long down hills, even when moving pretty quickly.

The encouraging: if the meter is correct, my weighted average power for this ride was higher than what I was expecting, at 190 watts, over 1 hour and 24 minutes.

The enlightening: it's interesting to put a label on various pain levels. Going up hill vs down hill, regardless of speed, you quickly learn what 250, 300, 400 and 500 watts feels like. I now know that 400 watts requires some serious attention and 500 watts starts to hurt pretty quickly. I also learned that I put out 749 watts at some point this morning...for a couple of seconds.

I need to do an FTP test now. Maybe I can get that done Sunday morning.


-Matt

TrojanHorse

Strava power estimates are worse than a joke! Very sensitive to small inaccuraces in elevation / grade and completely ignorant of the wind situation.

MattTheHat 

Yep, wind would be an obvious source for error, and it was fairly windy this morning. I had not thought of the elevation/grade inaccuracies, but there would definitely be some. Strava, and the Elemnt Bolt have always had significant differences in elevation gain since I've been using them. Sometimes by as much as 10% for a given ride.


-Matt

ericy

My intro to riding with a power meter was when I got a smart trainer and started with Zwift. Aside from this, my experiences are similar to yours. But I never took the Strava estimates very seriously before.

jpescatore

If you really like power-related data, there is a site MyWindsock.com that takes your Strava data and matches your route against wind data, and gives you all kinds of data - like power adjustment and (I think only on the pay version), essentially an estimate of your drag coefficient - if you try different positions or configurations on the same segment at the same power under the same conditions you can see if you decreased your CdA.

jadocs

Yep any calculated power is going to be off based on some of the reasons outlined above. The equations are based off estimates or generalizations and do not include actual factors such as wind, rolling resistance, aerodynamics, drafting, temperature, etc., etc...

I have a power meter and was in a 40 mile fast group ride a few days ago. One of the guys I ride with never pulls and sits in the middle of the pace line. His watts was estimated at 198 straight average with normalized power above 200. No way in hell he put out that much power. Riding solo that number would be closer, but he probably only averaged 130-135 drafting the whole way.

Here's another factor, he sets his Garmin to pause when he hits a certain speed....not sure if he has it set at 3mph or what, but that severely skews your actual average speed which also throws off power estimate. In summary he's not as fast as Strava thinks he is.

DrIsotope

Strava ignores custom auto-pause settings. I have my auto-pause set to 0mph, and Strava will always manage to add around 30 seconds of moving time to a 90 minute ride. I've had Strava tack on as much as four minutes to rides where my total stopped time as recorded by the device was less than 7 minutes. I call it "Strava Math." So that guy isn't helping himself with auto-pause settings. Now running a wheel speed sensor and setting it to a larger circumference, that will artificially inflate average speed.

On topic, I found that Strava's "estimated power" was fairly close to my measured power, after recording 200+ rides-- and only when viewing power for an entire ride, not segments or individual efforts. Most of the routes I rode with and without power were in that 5% area for variance, unless other variables came into play. For most people, Strava will always guesstimate high, because I think low power numbers make people sad. Just like how they always inflate calorie burn numbers. People like those big. Strava always counts zeros when doing power measurement, then gives you a weighted number, which is pretty close to average power when not counting zeros. They just like to throw on an added layer of math.

My current favorite Strava + power meter game is highest possible average speed with lowest measured power. Once Strava Math is done with it, I can go pretty far, pretty fast, for very little power output. Two weeks in a row, ~75 miles @ ~17.5mph from ~155W. Weighted in both cases right around 200W. I guess I'm a very effective coaster. Pedaling time in the 80-85% of activity area both times. I've learned to just take it for what it is. So long as the load/stress scores are consistent, I can use the power meter for what I bought it for in the first place-- not overdoing it.

Stages (in my year of experience, anyway) isn't far from Strava in terms of padding the numbers. I know every meter is a little different, but when I switched from Stages to Power2Max, my FTP immediately dropped 40 watts. Turns out it doesn't take 250W average to do 19mph for an hour. I do miss those Stages numbers though. Three plus hours at 240W average? Looks good. Now, 70 miles in 3h37 @ an avg. speed of 19.4mph? Like 180W. Somehow less impressive. Maybe I've just learned to be super aerodynamic.

MattTheHat 

That should provides hours of entertainment! I'll go check it out.


-Matt

jadocs

Very interesting. I wonder if you didn't have a defective unit....or those numbers were based off a windy day (headwind)? I just did a short solo 20.4 mile ride yesterday and averaged 20.1mph and my stages had me at a 192 average, normalized at 200 with a slight headwind. Sounds like your output is very similar to mine.

Edit: 93kg here

DrIsotope

Mine was erratic, and ate batteries like candy-- great for Zwift, though. Sold to a guy in Spain, hope he has had better fortune with it since.

jadocs

It wasn't a GenIII was it? It sounds like some of the issues I have read about in the earlier GenI, II models.

An interesting thing happened last week during a ride. My PM just died at mile 22...just dropped to zeros. The battery registered as "good" on my Bolt (connected via ANT+ not BLE). This was after 5 months of using it. I replaced the battery and all is well again. I can only surmise that battery level checks are not very accurate and the battery was at end of life.

DrIsotope

I'm going to guess it was a GenII-- I can't recall, as I sold it almost 2 years ago. Stages guesstimates ~200 hour battery life for a single-sided meter, but I never got anywhere near that. If I could get a battery to last a full month (50-60 hours,) that was something. I had cells die in just 10-12 rides. I think I had the Stages about a year, and went through probably a dozen batteries. By contrast, I put a new 2450 in one of my Power2Max meters on August 31st of last year-- and that replaced the battery that came with the meter in the late-fall of 2017.

I'm not bad-mouthing Stages-- plenty of people have very good luck with them, and they're probably the most popular PM out there-- but my experience was not great (sporadic calibration issues to go along with the unpredictable battery drain,) and I couldn't pass up the deals P2M was throwing down when I bought my pair of Type S meters.

jsigone

my stages battery last about 6 months

jadocs

How accurate is checking the battery level? Do you see a difference between the head unit and the companion app....or are they equally bad at doing it?

What keys you off that a battery change is needed besides time?

jsigone

my Wahoo Bolt kept telling me the battery was low, procrastination last about 1-2 weeks before I logged a ride w/o power number.

CraigMBA

My Garmin prompts me when the battery is low. I have it set up now to calibrate the arm every time I ride, thanks TrainerRoad guys!

jadocs

Interesting, mine said it was still good, then it just stopped.

Riveting

The calibrate feature of the Garmins simply tares the unit with zero load, and adjusts for the temperature of the strain gauge at the moment of calibration. Which may change significantly over the course of the ride. To better calibrate, you'd need to hang a known load/weight from the pedal at 3 o'clock, and at the ambient temperature of the strain gauge at the time of use, not in your warm garage before going out on a cold ride (or vice versa).

jadocs

What is the known load hung from the pedal and how is this calibrated? I do not see any option for accounting for known load in the calibration process.

Riveting

Hanging the weight (aka Calibration) is typically not neccessary, though I believe the Stages instructions called for it when you mounted the PM yourself (but I read that several years ago when I was researching which PM to buy, I went with the PowerTap G3 hub), and that may no longer be an option. Here's a quick article that discusses calibration vs zeroing: https://powermetercity.com/2016/03/1...s-zero-offset/

jadocs

Ok I see what you are saying now. Yeah, zero offset is still done through the head unit and/or Stages app before every ride. The actual calibration is done at the factory before a unit is sent out.

ericy

For what it is worth, I bought a PowerPod for one of my outdoor bikes. It doesn't measure strain in the usual sense - it measures acceleration, climbing/descent, and it has a wind sensor in the front of it, so if you are riding into a headwind it will account for that. When I first got the thing, I needed to do a calibration ride.

I went with this thing mainly because none of the other options would really work very well for me on this bike, and the price was pretty good.

Glenn6870 

I had to have my Stages PM replaced back in October. i was going through batteries every 30-50 hours. I did all of the known checks, and then got on the phone with Stages support and they had me do a hard reset, etc. Still drained batteries too quick. They finally had me send it back, checked it over, and then just replaced it. Been on the same battery ever since getting the new unit back in November. FYI, my head unit and the app would say i had a full battery and then i would lose full power during the ride. So if you’re having a battery problem, call Stages and complain politely. They will hook you up. Luckily mine was under warranty so it was a no charge replacement.
And no, i don’t recall which version i had. All i know is it was the FC6800 model.

jsigone

I have the same FC6800 when they went EOL and on sale. I got it for $360 I think.

Glenn6870 

That was when i got it too! The Ultegra version was about $100 less than the 105 version. I though it was a no brained.