Yep, I have my age entered. I totally understand there are genetic outliers, but you’d think the predication would be somewhat grounded.
I don’t think the raw watts are anything crazy for my age, but keep in mind that watts and body weight correlate quite well. I’m only 64.8kg, and could easily be 63kg with some work. I checked the chart Nate posted for w/kg per age bracket, and over 5w/kg is way outside - we’re talking .1% or less.
Obviously I’d want numbers like that more than anyone, but just seems like the AI is glazing me hard lol.
ya I think we’re all getting a bit glazed with the new system.
I think about my new FTP detection as the FTP I am starting on rather than the FTP that I already have:
Mar 1: detection 312 (so on detection day I started a build towards 312)
April: detection 318 (complete 312 build, starting a build towards 318)
May: detection 330 (complete 318, starting a build towards 330 - or whatever the number comes down to)
That’s actually a really good way of looking at it and in line with my experience after 4 blocks. My new FTP is something I’ll be working toward in the current block. That feels a lot more grounded in reality.
Apparently the AI thinks I am on a thin line with fatigue and predicted FTP numbers.
It’s my rest week at the end of a base phase. I had a bigger ride yesterday planned and accounted for in the training plan as a C race (it’s not a race, just a longer annual ride I do, 3/4 of it was below sweet spot, and my time in zones is shown below if anyone cares to check). Fatigue detection gave me a red day today (that was expected). It still kept my 60 min endurance workout for Saturday per the plan. FTP prediction is on Sunday. Prediction stayed the same before yesterday’s ride and afterward.
I went for a walk today at lunch and had my watch record it like I usually do. I looked at my calendar afterward and it had Sunday’s prediction down by a point from where it was yesterday, and it was prompting me to take a rest day tomorrow and showed it as a yellow day. I accepted that suggestion and FTP prediction stayed the same (1 point down from yesterday’s prediction).
I changed back to a 60 min endurance workout for tomorrow, yellow day and prediction stayed the same.
I deleted the walk from the TR calendar and no more yellow day tomorrow, no more suggestion to take a rest day tomorrow, and the prediction went back up 1 point to the same it was yesterday.
I’m taking the e-cargobike to pick up my kid from school in about an hour. Wondering if that ride will have the same effect as the walk. Usually records 9-10 TSS for these school runs, but I consistently do them even though they’re not planned in my calendar. The walks are less consistent.
Editing to add that yes, my 8.5 mins of an e-bike ride for school pickup (10 TSS, avg HR 93) was enough to trigger it all again: yellow day tomorrow, rest day recommended, and Sunday’s FTP prediction dropped by 1 point.
Prediction yesterday: 244 (had been as high as 245 during the block).
Detection Today: 244.
This last block was much more consistent from me. I didn’t miss any of the regular workouts. I opted for an indoor workout on one day where I originally scheduled a solo outdoor ride (got an FTP predicted bump from that). I also added a race in place of another originally scheduled solo outdoor ride (got an FTP predicted bump from that).
Prediction for 28 days from now: 257 (it was 254 before I detected and accepted today’s FTP update).
I suspect that number will get smaller during the block. I’ve got about a week of travel scheduled beginning next week. It’s blocked out in my calendar, so the AI should be accounting for it, but we’ll see.
Long answer: I’ve been tracking this sort of thing longer than I’ve been on TR and so it makes sense for me to continue to do it. TR didn’t use to bring some of these other activities in. When they started doing it, I feel like they wanted that information to feed into their models for fatigue tracking, etc. I feel like the more information TR has to evaluate the better they should be able to make their predictions and models.
I don’t track everything. I’m not recording myself mowing the lawn or walking my dog. But if I’m going on a walk for more than 20-30 mins, I’ll usually record it. Sometimes my office has fitness challenges where tracking it gets me swag and bragging rights.
The school pickup runs are shorter, but I track them anyway. TR doesn’t have a separate activity type to recognize them as e-bike rides and I feel like they should. I have HR in the activities from my watch so if anything that would reflect how little effort I put into them on a regular basis.
Also, if anyone wonders if I fill up my Strava followers’ feeds with all these, no. I default to have my activities upload as private. I then make public only the rides I really care about sharing with friends. These walks or school pickup rides are rarely (if ever) shared.
Stared the month at 260W, predicted at 280W, which had crept up a bit. Skipped a workout at the end and ended up at 274W/3.3W/kg. Predicting a 291W FTP if I do all the work, although life starts to get a bit chaotic this month. 300W by race day in August would be nice.
Every power PR on Strava beyond 5 minutes is in the last two weeks. I’ll call that a win. I do need to get better at using it, pacing out of erg mode is not my strong suit
I do think the newbie gains are starting to plateau a bit, the last couple rides felt right on the line between hard and very hard, whereas they’ve typically been hard.
March 7: detected FTP 231 W, predicted next FTP 255 W
March 31: predicted FTP 261 W
April 6: detected FTP 256 W, predicted FTP 251 W (due to A Race and training plan ending half way through)
I’m happy with the results, especially considering my compliance has been mediocre. I’ll keep trying to improve compliance but work trips and family commitments are scheming against me.
I’ll figure out a new plan once I get through my A race and work trip right after that.
I did almost every single planned workout, all on the trainer. I logged every time my predicted FTP decreased and tried to explain why. The biggest changes seemed to occur after I rated VO2 max or threshold workouts as “very hard” when AI thought it would be “hard.” I don’t regret being honest. The workouts are perfect, and I am very pleased with the FTP increase.
Next prediction for 12th of May is 242. That’s what I started with.
I had two unstructured rides with PRs in both that helped with lowering the frp prediction. I am a low volume athlete. Can’t do more riding and I know this is stopping my progress, it is what it is.
Prediction for today 400 watts, one workout in dropped to 386 watts where it roughly stayed til one missed workout last week and dropped it to 380 watts.
Aiftp today of 379 watts, prediction of 392 in 4 weeks…
Not a good round, AIFTP bounced around the predicted 386 watts only for one missed threshold workout a few days to drop 7 watts off that to 379 watts where it remained. Now back to essentially where I started but with poorer repeatability and sustained efforts in my recent training past, now testing that fitness on a weeks riding with friends and yep, it’s not what I’d have expected from where I was in January.
Don’t get me wrong my compliance this time hasn’t been as good but the continuing diminishing difficulty/retrograding means I’m less fussed about listening to it anyway and I’m not really motivated to care to listen. For full disclosure I am a shift worker, I work irregular shifts and have to change my plan to fit in with life and that will mean I do and rate workouts according to fatigue that has nothing to do with cycling so it’s battling a lot of unknowns.
Same, I am shift worker and change schedule constantly. Do you train only outside, there is known bug to raiders training only outdoors, it inflates the ai ftp prediction. The actual detection should be accurate.
AIFTP update - 249 April 19, 2026 (one fail workout and one Very Hard rating)
AIFTP prediction - 237 May 17, 2026 ( this one hurts, the drop is due to missing six workouts due to family holiday, the following block I miss three more workout, same holiday)