Pretty cool! Sort of akin to Wahoo’s 4DP without the testing! Just looked and mine was pretty accurate relative to what Wahoo has told me in the past. Will be curious if they extend this out to your “weakness” to see if that also correlates to what Wahoo has told me before. I don’t have SYSTM anymore so won’t be able to compare real time.
Interesting new feature I discover just now in the forum. The overall ‘qualitative’ outcome seems right (It says I’m good in long climbs), but I wonder who is considered in the percentiles it gives : probably only persons my age, as it says I’m 92.6% percentile this year in 1 hour ‘box’, knowing I’m 59 years old, 2.95W/kg on the 1 hour chart . The percentile would be much lower if they consider everybody.
It also recommends climbs like Alpe d’Huez, close to where I live. I can tell you I’m far from the top 8% of the Strava segment at Alpe d’Huez 
I really like this visualization for this type of data, also give you a sense at first glance of what you’ve been prioritizing or not over the last 8-weeks
5 Likes
Isn’t this just about the fact that not everyone was at the peak of their career in 2024 but when you set it to ‘all time’ you’re comparing your career peak to everyone else’s career peak regardless of which year that was for each of them?
Imagine that the pool was ‘Tour de France Riders’ - all of them that have ever raced the tour (with a power meter) ever. If you’re comparing your all time PRs to the numbers they could do in the year(s) they rode the TdF then you’d expect to be in last place (or close to it). But lots of them are now retired and quite a few probably don’t even train anymore if they ride bikes at all, so if you compare your 2024 to all of their 2024s then you’d expect to be appreciably higher than last place even if it wasn’t your best year ever.
TR CEO Nate is a good real life non-pro example of this - longtime podcast listeners will know that he’s a tall guy whose all time peak FTP was something like 365w (I could be way off but definitely 350+). After some concussions and other life events he stopped racing and wasn’t even really riding for a long time, with FTP dropping as low as ~200w (ish? definitely less than 250 though). So in ‘all time’ he likely moves most of us down one spot in the rankings, but in his trough years he’d likely move a lot of us up one in the annual rankings, even if it was only our 3rd best season or whatever.
As long as we assume that the dataset includes more riders in their trough years than at/near their peak in any given year, then you’d expect to be higher in the rankings for any single near-peak season than you would for ‘all time’. I guess if you were trying to quantify the effect what you’d want to know is: what percentage of the average TR career do most riders stay near their all time peak? If it’s 50/50 then you’d expect very little difference between ‘all time’ and any one year because every trough year is cancelled out by a peak year. But we don’t even need to look at any data to know that’s not realistic - if the average TR user logs 15 years of data and we assume that a max of 5 of those years are at or near their peak, then in any one year 2/3 of all riders are in one of their 10 non-peak years. (In reality I’d guess it’s even more extreme than that - 3/4? 7/8?)
Oh and let’s not forget that lots of riders change their power profile over time, so their all time 1 minute PR might not necessarily have been logged in the same year as their all time 20 minute PR, meaning that in 2024 you’re more or less comparing to individual athletes, but in all time you’re comparing to Frankensteined PRs that might not be physically possible for a lot of those athletes to achieve within a single season. This one will be especially relevant when switching the comparison to w/kg from just raw watts, since the same rider is likely to benefit from a fairly significant change in body composition in order to optimise for 20min w/kg vs 5s w/kg, but it takes at least a season or more to make big changes like that.
2 Likes
Nice feature
I agree with the phenotype and do prefer hilly or undulating courses
Although im not sure about the accuracy of the percentiles
1 Like
Can I assume that “Your Power Ranking” is age AND sex adjusted? I’m not finding that anywhere. It’s gotta be right?
@ZenTurtle that looks really nice - what app is that?
1 Like
My understanding is yes, and the page does say “compared to other athletes like yourself”. I have to assume that people are put into age and gender buckets and compared within those buckets. The specifics of those buckets don’t seem to be clearly described anywhere.
I would not be so quick to assume that. What they actually say is you are being compared to “real TrainerRoad athletes striving to get faster”, which could be anyone with a TrainerRoad account. The problem I have with their percentiles, as others on this forum have pointed out, is that it does not match the numbers from the age/sex stratified distributions from intervals.icu. For example, my 20 min power (W/kg) for males under 40 is 15th percentile, while TrainerRoad puts me at 76th percentile for the same power. Clearly the two populations being used to make these percentiles are different but at least intervals.icu tells me what the age/sex for the population are.
2 Likes
A little late to the party – but I got through this whole thread and there is still no answer for what “Trainerroad users like yourself” actually means. People who own 12 bikes? People who live in my metro area? People of my median income? # of pets??
2 Likes