Can we get a dedicated thread for the intervals.icu power profile flex? 
2 Likes
Is this what was used to get the data above? I want in…
1 Like
Awesome. I look forward to the update.
Also, great job with the work on intervals.icu, it keeps getting better and better.
1 Like
Have you done some good 5min intervals or tests? That “V” shaped profile doesn’t jibe well with the physical abilities required to have good 60sec and 20min power.
Not really, I’d imagine my profile is more just a slightly sloping downwards all-rounder. My sustained power isnt as impressive as it looks when compared to people around my age even after a good base. I’m hoping to see it improve over build, but i’m not entirely confident it will.
I noticed the same. I’d guess that the sample sets for TR and intervals.icu are lacking in the amount of top performers in their data. Most elite level or even near lite athletes are coached and will be using TP (or their coach will be) to monitor training. I don’t personally know any (really) fast folks using TR unless they’re heavily adjusting/customizing. I know there are some, but more middle aged weekend warriors and time crunched folks using the platform.
Hey David,
Fantastic work.
Just a note, you may have to automatically throw out some of the data that makes the curve. Maybe removing data outside certain known parameters?
In my 40-49 age group the top power is listed as 5min 658W, 20min 594W.
Obviously, these numbers are only possible for Superman. Even then…
I will look into that but whatever parameters I set something is going to sneak in. I was hoping that the power spike detection would prevent this sort of thing. Might have to just live with the fact that the top 1% of the chart is out of reach 
I could just toss out the top 1% I suppose.
I’m sure there’s a mathematician on here who could chime in.
I suspect some sort of automatic binning of data outside certain W/kg parameters may be the solution.
Really it just means that 98-99% is probably 100%.
I was going to ask you about power spike detection. The default value seems a bit low. It removed all my sprints until I figured out you could change it.
Oh what a simple change of metric can do for one’s cycling confidence and self esteem… 
Thank you David, you basically became my psychiatrist, you made me go from ‘meh’ to 'actually pretty decent ’ in a span of single click
Keep up the hard work mate!
2 Likes
I’m pretty sure these outliers come from a calculated (virtual) power. The correct way would be to omit all data from users that don’t use a power meter from the model. I found this article about identifying outliers interesting but in this case I think just tossing the unrealistically high values for 5min and 20min bins would be enough, it’s not an exact science anyway.
Strava supplies a “device_watts” field and I only use power from rides with true. Supposedly that means real power but Strava is at the mercy of whatever creates the data in the first place.
2 Likes
I would assume all virtual/indoor rides would have “device_watts” as true?
Its interesting how few people seem to do 20min efforts given that the 20min column is usually lower than the eFTP number - unless there is something wrong with the eFTP calc?
Ah thank you… now I feel good because my 2h power is so close to 1 hour suddenly I’m up there.
It’s amazing how these silly things can make me feel better being all number obsessed.
Seriously though the descrepancy between 1 hour power and eFTP when you’re around the 3.8W/KG mark is interesting.
2 Likes
I suspect so.
The estimated FTP number is calculated from efforts of 3m or more and there will be very many more of these than 20m efforts + ramp tests and so on. So it makes sense eFTP is higher than 20m.
1 Like
You can now select “Coggan Male” or “Coggan Female” in the age group drop down to view the old Coggan charts:
6 Likes
This is sort of the unofficial TR data thread, so @Nate_Pearson I was wondering if there’s been a spike in TR use since many countries have implemented Covid19 isolation/lockdown measures?
1 Like
Yes, there has been a spike in sign ups. Nothing earth shattering but definitely more than normal.
14 Likes
@Nate_Pearson Could optionally overlay this on the power duration curve?