I can imagine it does exactly that - of course in the nebulous, indirect way that AI/ML does anything. Just as survey responses feed the AI of Adaptive Training, AI FTP Detection would be informed by that as well.
See Amber’s quick reply to my initial question, “super interesting case”.
I’m not an outlier in normal distribution model. The model predicted the perfect accuracy. Given how I feel last 6-8 weeks with my FTP on workouts.
Good job TR team!
Word to the wise, don’t keep fiddling with your FTP!
I deleted an ftp test from a few days ago. Used FTP detection and got a new FTP. Did a productive sweet spot workout and felt great. Hard but great. I was playing around with it again last night and deleted the ftp test to recalculate again. This time it froze with the “processing data” error. No biggie. I went in and manually entered the FTP that was given to me by the AI. This morning, I was confused as to why all of my workouts were suddenly “not recommended”. Did I train too much this week?
Nope! I’m a dork and by fiddling with the FTP and manually entering one. I nuked my progression levels. Doh!
Nate himself posted his decrease. I am hesitant to call it massive, to avoid offending him and get a personalized AT on revenge mode, but substantial at least
Yes, same thing happened to my wife. Was going to post a similar PSA, but you beat me to it. Her PLevels are way down now. Not a big deal though, just do higher PL workouts to get back to appropriate levels. Just wish I had taken A screen shot of her levels before running the estimate.
In terms of the estimate for her, it seemed decent. It was four watts higher than her Ramp done on Monday. I felt it should be higher as the Ramp seems to underestimate her capabilities…
Both have unique cases tho.
One have a concussion during a race and took time off (plus now its hanging out with the orange crew, or so i hear)
The other just started out after a maternity break.
I was wondering of active people getting a lower FTP number.
I wouldn’t call them unique cases as long as you where talking about massive decrease. Injury, maternity break and other similar reasons for being off the bike for months are what I would say is required for a massive decrease to happen.
There has been some examples of people self adjusting their FTP down and AI FTP bumping them back up, but it’s hard for us to say if that is wrong of AI FTP or if it’s a matter of such a decrease better being handled by a drop in PL values.
Yeah I know. He knew his power was down, and he declined. Other will see it and go like, awesome…more power!
The model seems to be pretty good, but the question is, how will it impact people when they see a 5% increase and suddenly they cant complete their wo even at productive…
Another positive user experience here!
Previous FTP at 263 and ramp tested last week up to 274, but that was following a few stressful days and I did not feel it reflected what I have been able to do on the bike recently. Deleted that update and tried AI - bumped it up to 284 and reset most PLs to 1.0. Completed McGregor -6 and it seemed to be spot on.
Signed up for my first ever TrainerRoad month as interested to test this. I’ve opted in on the Web and imported all of my ride history using Strava. I have a ramp test scheduled but I’m not seeing the option. Do I have to have completed a ramp test in trainer road once before I can see the AI FTP option?