Should people adjust their plan if they aren’t progressing? I think the answer is yes.
I think TR could do itself some favours around this subject by giving a dedicated technical article explaining (without giving away IP) the models decision making process.
I found AiFTP great throughout the winter when following the plan to the letter. But now the seasons started and I am constantly changing scheduled workouts to fit in group rides, sportives, short TT’s, 100 mile gravel races or whatever, most of which are at a higher difficulty than the workout planned for that day, the FTP prediction is all over the place to the point I am tempted to turn it off as its a distraction.
What are you giving the algorithm that you deserve a higher FTP? Can you do more? Do you want to?
Maybe you are just at a physiological ceiling? Maybe it’s age? Maybe a plateau that needs to be broken through?
This would be huge.
Then people will just complain that the model doesn’t decide in the way they want it to decide, try new ways to game the system and criticize the outcomes, etc. Then they’ll have to announce and speculate over every change in the decisioning to get “approval” from the forum, which will never come.
Call me a cynic, but that’s the way I see it.
I think that as self coached athletes focus on outcomes; we need to evaluate whether we are getting the workouts we want from it - regardless of how they do it.
Isn’t the point of TR to give the most effective training to reach an athlete’s goals? Right now it’s “follow this plan and you’ll gain 5W in 4 weeks” gradually becoming “sorry, you lost 1W but I can’t tell you what went wrong.”
It was always going to be a bold move to predict future fitness. Everyone else including coaches get away with it by not putting a number on it. Nor are they able to, which is handy for accountability.
As I said a few months ago, I think we need to (re)start measuring by measuring performance itself. You want to know whether you are fitter than last year? Race the same race you raced last year. Do a time trial. Check your other performance markers in races. Segment PBs.
After the first few months, I’m no longer chasing an FTP score, I’m using the predictor as a guide to my calendar changes. If I miss this workout, what will happen? If I move this there, or extend that, does it increase or decrease my fitness?
The problem here isn’t the prediction alone but that people aren’t getting FTP improvements. FTP isn’t the only measure of improvement but unless you are already at your potential which many are not then why are they not seeing improvements or worse decreases. On the old ramp test system and even earlier AI FTP I would see steady increases. They may have not been big increases but I always had faith that in 6 months time those increases would add up to something measureable. That doesn’t seem to be the case now for many and they don’t understand why. The prediction just adds salt to the wound giving a number that those people have no confidence in and then watch it slowly deteriorate to another month with no gain or a loss.
I pay for TR… TR has a FTP builder plan… 6 months in… I have followed the plan… 1w gained. Either its a broken tool, or I have plateaued. I just want to know!
It may be 1W gain in aiftp but try testing after every block and track it. Depending on the day, folks say aiftp is ftp, is not ftp but a training score, ftp doesn’t matter at all. Just turn off prediction and aiftp and track how your fitness improves for a few months.
I think TR should turn off prediction for everyone and fix it before releasing it again.
But TR says use FTP detection not testing…
Also I have no way to test my fitness other than TR FTP (or I used to use Zwift FTP ramp test or auto measurement)
Aiftp is a feature. Don’t need to use it if you don’t want to. I’ve started doing the TR ramp test and manually updating ftp. It’s been working out and I agree with TR that the workouts still adapt accordingly.
I did a 20 min test in the past and it was a few watts lower than my ramp test and that is about 10 watts lower than aiftp.
With all that said I still really like the TR app. The workouts and calendar are the best compared to other apps.
Why not pick a few Zwift KOMs and use those to track your performance?
Pre TR that is what I would do (and may still need to), but I was really sold on the TR promise that FTP tests were superseded by the superior AI FTP detection. I did a crit race (and won!) on zwift 2 days ago… my AI FTP prediction immediately went up… which begs the question … why is the TR plan not stretching me in that way?)
Just reading this thread on the forum (and others) I find it rather frustrating.. if one has issues with TR not progressing their ftp upwards then the user is told they are their own coach and obviously need to be adjusting the plan themselves, adding work, changing the type of work and doing their own testing.
But when users deviate from the plan then they are told that is why it’s not working and they need to follow the plan TR gave them if they want to improve because TR designed it that way for a reason and addding extra rides was obviously too much for them so it made them slower 
Because tracking fitness without testing is literally what TR is selling.
Listen to Jon’s recent podcasts with Keegan. Two different takes there.
Keegan doesn’t sweat FTP. Clearly his FTP is enough to finish on the top step consistently. So it makes sense that his goal isn’t to grow it but fine tune specific efforts. In reality I’d also be fine with TRs predictor telling me, “we’ve looked at your data, you ARE at your physiological limit. Don’t bother trying to increase FTP. Nothing you do will work.” I’ll take that over “if you do this you’ll see an improvement over the next 4 weeks” and at the end of the 4 weeks “yeah, about that improvement, it didn’t happen. Thanks for playing.”
Jon then claims people are obsessing with FTP. Moments he later reports a typical week with 3 interval sessions and 3 endurance sessions (plus he coaches NICA) to reach a target W/Kg to get a specific result at Leadville, using TrainerRoad.
so which is it?
I don’t disagree with your sentiments and frustration.
I also think my suggestion is sound (and practical).
Both can be true.
I actually think I will do as you say … not only will it answer the question of whether I have improved/plateaued, but would also demonstrate whether or not the TR plan/detection/prediction is wonky.
Edit: I have added 3 x zwift “events” (1 x crit race, 2 x rides with a personal kom attempt) to my future calendar … for each of them my FTP prediction jumped by 1w, without having even completed the event!
Well the prediction assumes that you complete everything on your calendar, so that’s no surprise.
I’m with you though; there’s very little transparency on the predictions and they do seem to change the prediction too greatly for each difference from exactly perfect execution. And the training plans aren’t necessarily handling common situations well to prescribe the right plan.
Sorry my point (which i didn’t make very well) is that just following the FTP builder plan has led to a static FTP for several months … but as soon as I went off piste and started adding in my own “events” the FTP prediction has immediately started rising. It’s almost like I’m having to challenge myself because the plan isn’t.