hamster

I think that Strava does something weird when averaging power for the chart. My chart above was from GoldenCheetah with 3 minute averaging. The same ride looks spikier in Strava (not as spiky as yours, I think). But yes, I was actively trying to stay at FTP or below. When you're climbing a 3000 ft hill, there's no good reason to exceed FTP. With FTP of 190 W, my peak 5-minute power was 200 W and my peak 1-minute power was 214 W. (If you don't have premium Strava membership, you can calculate this for your rides in GC.)

65% FTP for 6 hours continuously is very different from 65% FTP for 6 hours moving time + 3 hours of rest. Strictly speaking, that's 65%*(6/9) = 43% of FTP for 9 hours. (But, if you can do 65% for 6 hours of moving time + 3 hours of rest, you can do more than 43% for 9 hours if you take shorter breaks, eat on the bike, and recover by coasting or pedaling at 50% of FTP.)

I see the point about performing above FTP in racing.

mr_pedro

Yes, the 3 minute average would remove most spikes. I think Strava is also doing some type of averaging, where the period depends on the length of the ride being plotted in the current zoom level as probably a fixed number of point are plotted each time.

Just checked what the max numbers were for that ride: 5 min: 306W, 1 min: 434W.
I also use GC and WKO+, but find myself nowadays mostly using Strava after each ride and only using the other tools every now and then for more in-depth analysis.

jayoyo

Lots of good advice in this thread (go slow!). Here's a different kind of strategy I use on hills, when I have a long climb and I'm in a nice low gear but I'm starting to get frustrated/tired because I feel like I've been climbing forever without getting anywhere (because I'm slow). I sing songs to myself. I have a mental collection of pop songs that are around three minutes, have peppy tempos to go with my spinning, and distract me from how slow I'm going up the hill. I might say to myself, "oh, I will keep going through one more song and then maybe take a break". <la la la la la etc>. Then, "well, maybe I can hang on for another song". Works for me.

mr_pedro

Ha, on that tour, when I really need a boost there is the rubber band. You see what you do is imagine a very long rubber band attached to the bike and stretched all the way around a tree or post further ahead. Just don't forget that as soon as you pass the tree or post you need to fix it to another tree or post further ahead, otherwise it starts pulling you back...

chasm54

Strava's power numbers are junk.

mr_pedro

Would you care to explain why?

chasm54

I may be overstating my case (wouldn't be the first time) in that I believe Strava now offers figures based on measured power, rather than just the crude estimates it used to make. However, all the PM users I know still find it to be inaccurate, and often (but not always) low. Why, I can't pretend to know. I have heard speculation that Strava averages the data in some way that may eliminate, or fail to take account of, very short-term variations in output.

mr_pedro

If you don't have a PM Strava shows its own estimate, which doesn't know about wind, aerodynamic position and depends on weight input and accuracy of grade and speed measurement.

But you are talking about how it draws the graph? I noticed that the peaks change when you zoom in and I suspect that it is drawing a fixed number of points. So it needs to average over each segment and the segment length depends on the length of time being plotted. So if you zoom in, the segments become sorter and the peaks become bigger. I wouldn't call the graph junk because of this, but I can imagine it is confusing if you don't understand what's going on and you see the peaks changing as you zoom in.

mr_pedro

Okay, I just had a look and was able to back engineer how strava is plotting the data from power meters. It is indeed not optimal. The problem is that strava is actually not averaging! If you are plotting e.g. a 3 hour ride, Strava will divide the 3 hours into about 300 segments, so lets say 36 seconds each. It will then just plot what the 1-second power was at seconds 0, 36, 72, 108, ... without averaging. So for the 9 hour ride I showed before, Strava was plotting whatever the 1 second power was every 108 seconds and ignoring everything in between.

chasm54

I'm impressed. You're the first person to come up with an explanation. That would certainly explain why it varies from the GC, WKO etc. data.

hamster

Aha. Good job! I've just double checked and yes, that's exactly what Strava is doing.