Try this, turn ALL those outdoor rides to indoor TR workouts and see what happens to your predicted FTP.
If it increases drastically after doing this, it means the outdoor workouts have a significant amount of variability which is being thrown into the model and will give you variations in predictions. What i think is happening is the model can’t really predict if you’re going to be fatigued or fresh after these rides due to your previous history
I’m outing myself here b/c I did the same thing on my calendar. When i looked back at my outdoor riding history it’s all over the place with volume (3hours - 5 hours) and intensity within those durations. So the model is going, “i don’t know how fatigued he will be after this workout”.
I suggested pretty much exactly this to someone else. I think TR might assume that a free ride, even if it’s the same TSS as an Endurance workout of the same duration, will be both less productive and more fatiguing by default.
I don’t think it’s TR assuming, though. I think TR is looking at what that athlete has done for those free rides and sees on average they spend X% of their time above threshold which is really fatiguing. The model the goes, ok, knowing that they will be fatigued coming out of this free ride.
I would say it’s very hard for most cyclists to do a free ride and not go above threshold at all.
How many testers done longer volume and duration work during beta? I’m wondering if there needs to be fine tuning for big volume riders. Might be worth a try is to stack short workouts one after another, like one day with a high intensity 1 hour with a 1 hour easy post ride.
this really doesn’t make sense. mainly just because you can ride higher watts for 3hrs doesn’t necessarily mean you should. the coaches I know wouldn’t tell their athletes to ride harder on endurance rides and in fact riding this kinda high endurance/low tempo state just creates more fatigue than helps with anything.
and take, for example, someone I know whose ftp is like 250-260 and was given a 290 ftp with this latest release. if that were me, being given “threshold” workouts at 112% of my real ftp would actually be quite hard vo2 workouts. I get the counterargument will be “well TR will learn and adjust,” but frankly that’s a crappy way to be where you’re constantly having to screw around doing bad workouts while the system learns you.
I’ve read the post. I think the connection to Stages is fine now. I still have an FTP of 100 in the app when I can ride for hours at 150 watts. Tge app won’t accept a manual entry to correct it.
Changing the plan to indoor rides increased the prediction from 284 (-4 watts) to 290 (+2 watts). Also, even though I had dynamic endurance up to 5 hours selected for all of them, it put all of them over the next month at 3 hours. Thank you!
it still doesn’t explain why the prediction randomly dropped 5 watts when nothing had changed, but now at least I feel I am on the road to improvement again.
I’ve been reading quite some messages about topics such as the system won’t allow you to override AI FTP, and if you do - then the AI will not try to adapt or suggest workouts from there? Is that true?
I’m off the bike a bit due to my back as I’ve mentioned elsewhere, another 4 weeks off the new AI puts me to starting off the couch at 115W → I know I’m not strong in general and I’ve had a few weeks off, but my 11 year daughter pushes close to that and I doubt I’ve tanked that much (I’ve got a new goal! ).
Thinking about planning in TR into my recovery, I feel that setting a known easy FTP and stick at that for some time is going to be important. In the past if I’ve had an injury, I’ve manually set an artificially low FTP level to work with. It doesn’t look this this is really possible with the new implementation. The AI might want to ramp me too fast, even if I set conservative. Of course I could add rides and pin them, also an approach.
How can the system be used in such a scenario to keep things ‘as easy as possible’, anyone any advice on that front? Go with what AI sets and then set a 50% level in the workout? How does this then screw up the AI?
If you want AI to take over again, what do you do? Delete the last manual FTP entry?
Not sure if this advice is good or not so hopefully others will chime in but if you feel that the AI has given you a far too low FTP then I would manually change it to something closer to what you think is accurate or do a ramp test. Then after a month of workouts under your belt I would trigger an AI FTP recommendation to see if it comes more into line with reality by looking at the work you have completed over the past month.
The new FTP prediction is So Dumb. I follow the plan as laid out. Follow the workouts staying very close to the power outputs prescribed. Finish the workout to find that my new predicted FTP is now lower than what it was before the workout. This has happened every time after multiple workouts. I still have 22 days till the next FTP. I knew that a 30% gain on day one was not realistic, but why did it even think that was possible. What would I have to do perfectly to achieve 30% in that time frame. And if Ai thinks I am working to much, why did it prescribe this workout and this schedule. Why can’t it look at my many years of training and see what I am capable of and prescribe a workout schedule and workouts that are actually benefitting me instead of me guessing in the plan builder which is best. I was so looking forward to this….. Now its just more questions with no answers or direction.
Hi Caro, I posted in the Beta thread as it’s not been locked but not seen any TR activity there - details of my last week are posted there, I got to my prediction date and ftp detection was 14W lower.
I’m kinda worried people are going to be let down if they follow the plan and fatigue mgmt and get a worse result than predicted.
It does seem that you need to ignore fatigue mgmt to retain a higher prediction.
Surely if the system detects fatigue and you follow its recommendations than the predicted FTP should factor that in? Maybe the pFTP needs to have a setting - with or within fatigue management included?
I feel as though pausing during a workout is impacting FTP way more than it authentically should and that the AI should be able to detect the difference between a pause for fatigue, and a pause for technical issues.
Today I did a 2.5 hour endurance session. During the opening 5 minute warm up, I realized that my TVs Bluetooth was having issues. In preparation for a long training session, I jumped off the bike and spent 15 minutes fixing it. Then continued and completed the full 2.5 hours without any other pauses. Literally the only pause was within the first 2 minutes, during the warm up. This docked my performance to 4.4/4.7 and dropped my predicted FTP by 4 watts. If I would have known that, I would have just started the session over again.
Personally I do not want to add ramp tests into my schedule just to correct the FTP. There are many reasons an FTP can be incorrect, tech issues as in my case, injury, health. There really should be some way to adjust the FTP and let AI re-adjust as training progresses. The manual FTP input is not accepted by AI. I changed my FTP but the app won’t generate any workouts unless I accept the (wrong) AI FTP or do a ramp test.
I have years of data from indoor and outdoor rides with FTP around 170 - 200 but AI set it at 100. It’s really not acceptable to do a month’s training at the wrong intensity just to get a correct FTP value.
).