Hey all. Have been on TR for many years, but at the moment just trying to build volume so aren’t really following but have the maintenance plan loaded in and do the workouts occasionally opting more for outside riding. The workouts seem easy so have been increasing intensity.
I just did a ride for 2.5hrs at 0.91 IF based on my AI FTP. It was also a recalculation day so did that post ride. FTP hasn’t moved which is no big deal but that is a new w/kg PB for me so seems strange that it still thinks my FTP is the same.
Hey sifi1975,
AI FTP detection seems to work quite well for me. Anyway this year, I’m trying to make periodicals checks by performing a 20 minutes FTP test at the end of the main training blocks, usually the Saturday of the recovery week. During the past winter, as the majority of the intervals days have been performed on the trainer I have completed all the tests indoor with the Trainerroad 20 minutes test protocol.
Now, as I perform outside the majority of the intervals, I’ve made a copy of the test in trainingpeaks and use it to perform the FTP protocol outside, on the bike I use to train. Usually I schedule the test before AI detection day for FTP, in order to avoid any kind of mental conditioning on my test day. After the test I look at AI FTP detection value and finally decide what my FTP is.
I’ve found that AI FTP and the tested FTP, in my case, are very close when the FTP test is performed indoor. I still need to evaluate how they match when the test is performed outside, because I’ve just scheduled the first outside season test for tomorrow. I’ve tried to verify the accuracy of AI ftp last season but I didn’t use a strict methodology like this year.
So, the long story short is: try to perform a FTP test and look at the results.
Thanks @Danidoc I hated doing FTP tests before and got really fatigued buy them so was thrilled when TR introduced the AI function. It’s good to know you found them really close.
Yeah. I find it surprising the degree of agreement I usually get between actual 20 min FTP tests, Garmin calculated FTP based on activities, and TR AI FTP. They usually only differ by few W, maybe 4 at the worst. Of course if using FTP to track progress, I like to stick to one method and use that, or only compare results between the same methods.
If you used Average Power (IF calculation uses normalized power), what would the IF have been? A really staccato power ride can “over inflate” IF.
Also, what does your power duration curve (go into any ride to see it - I don’t know why you can’t see it without going into a workout) for the last 90 days? That is really the best way to get a sense if AI FTP is over / under estimating your “actual” FTP
What was your RPE for this ride? Or where was your heart rate relative to LTHR? Are you suggesting that you felt like you were in zone 2, or do you think you were at the top of zone 3? You’re not really telling us how hard this ride was for you, only what it averaged out to relative to an estimated functional threshold power (i.e., 2.5 hrs at 91% is doable).
It was a very hard ride for me as RPE and a new PB and 5 all time power so i wasn’t sure if it was possible to ride for 2.5 hours at that intensity, which is what initially threw me.
So I understand now a one off such as this doesn’t have the effect as I thought it might
I am having a hard time understanding how FTP hasn’t moved. I see you got a 4-watt increase on that day on your AI FTP Detection.
I see, however, that you completed that workout on Zwift, so maybe you’re talking about the FTP not transferring over to Zwift?
If so, while workout targets are based on your TrainerRoad FTP, you’ll need to manually update your FTP in Zwift to match. This only affects the color-coding of workout blocks in Zwift.
Judging from how I’ve feel during the workouts, I think AI FTP has been good enough and have been confident in the numbers.
Where I have my doubts is when you miss a few days in a block and AIFTP brings your FTP way down.
Back in April I missed about five days due to medical and family holiday.
AIFTP took my num from 230 back down to 214, that was a setback to was I was in the month of January.
I accepted it and did most of my workouts at 105% intensity to see if I could get a bump up.
It did bring me back to 223 FTP which felt about right at the time.
Hey all,
I would like to give a little update to the results of AI FTP vs FTP test. Past Saturday I’ve completed the 20 minutes ftp test with the Hunter/Coggan protocol, the same of Trainerroad, and it resulted in an FTP of 256 W. 3 days later, waiting for the 28th day from the last detection, AIFTP gave to me an FTP of 260 W. Considering all the possible perturbations to the system and to the test and the different powermeters I use, the results seems very close to me.
It’s wildly off for me but I don’t mind it. It’s just a number to train with.
If you think it’s low, increase it manually a little bit and train that way.
If you think it’s high, lower it and keep training!
All that matters is that you keep following the training, then do the exact same protocol and see a little increase in your power next time.
FTP in itself IMO (and many world pros’ coaches according to interviews I’ve seen) is highly overrated. It’s just a ball park. People with exact same FTP or even watts / kg can have wildly different results in a TT or road race.
