Yeah, I guess it’s just a psychological thing for some people. And TR doesn’t do themselves any favors in this area by making FTP such a core metric and marketing tool. But I’ll also give TR the benefit of the doubt that they know their users and a good portion of their user base is probably fine with it. Reading your initial post, you say that FTP isn’t that important to you, but it’s clearly important enough that the TR implementation is rubbing you the wrong way. From the numbers you’ve shared, that’s all within margin of error for a trained athlete in my opinion. Again, I get the motivation problem for some people, I’m just surprised anyone would read much into differences of ~5 watts in a value that is going to fluctuate by the day.
And for what it’s worth, I don’t even use TR’s FTP estimates or prediction (other than looking at it for curiosity). TR had my FTP overestimated by 20+ watts for a long time and then dropped it by over 40 watts the following month (right after my A race when I was coming off peak fitness). Neither of those extremes are anywhere close to my actual physiological FTP, but I just set my FTP manually and it’s close enough and the system works fine for how I use it.
After using TR for several years now, my current experience is very similar to the OP’s.
I find that the predicted FTP negativity impacts my RPE ratings because I don’t want to ‘disappoint’ the AI engine and get my predicted FTP reduced. So the current system incentives me to be aggressive rather than honest with my RPE ratings.
I also get the folks who are saying don’t worry about the numbers, and FTP is just one number, and all that. But that misses the mark IMO. Take one look at the Career page of TR. The FTP is in giant numbers, bigger and bolder than anything else. Clearly, the platform is suggesting that this is the metric to focus on. So when that metric is unreliable, it’s hard to just ignore it. At least it’s hard for me.
With respect, the part that misses the mark IMO is the part where people are making this issue TR’s problem. Why not just turn the FTP prediction off then? Lots of riders do it. I don’t really understand the urge to complain about a feature no one is forcing on you. If you find you are having trouble separating honestly rating RPE from the resulting adjustment in the prediction then this seems to be a “you problem” not a “TR problem”, no? Plenty of us are finding this feature useful. For those who struggle with incentives not to disappoint the AI engine, as you said, why don’t they just turn it off. I just don’t understand
Yeah, that seems like the obvious answer. Sure, FTP is an indicator, but honestly, I train so I can go faster up hills and over longer distances with lower RPE. I guess maybe it’s different for the people that don’t pedal outside.
This is why I have Prediction turned off at the moment.
I’m on Build Phase now and those workouts are hard enough.
I want to focus on the workout and how it honestly felt so TR can give me good workouts.
Now I have an FTP/AIFTP & Watt/Kg goal that I want to hit for this CX season, so in the back of my mind I know the Very Hard and Moderate rating may drop the AIFTP prediction but at the end it work out because I’ll keep getting good workouts and stay consistence rather than do a workout that may be too hard and have to quit the session.
This has worked for me so far. Note I do my TR workout on the trainer so will be different if doing the work outdoors with that bug that affect AIFTP/Prediction.
Another thing that I look at is heart rate as a trend.
TR gives me the same workouts sometimes, I compare the heart rate to power.
On a few time I’ve have seen higher power with same or lower heart rate, same RPE at the end.
I don’t let that dictate my training but it good to see that type of progress.
With all du respect, I feel this is the problem with posting in the TR forum, even thought they have a buggy system, that gets predictions wrong and ai detection wrong (I know aiFTP isn’t FTP and FTP fluctuates daily and it’s a range … if TR doesn’t get it right) . it’s always the users fault, and they shouldn’t be posting in the forum, next week, nobody is posting this issue in the forum … we have billions of happy users,
If users have issues with software features, they should be free to post it in the forum, not just suck it up as it might upset the TR fanboys
I agree with you, the latest much vaunted feature, is a pile of , and should be turned off, but if you don’t post in the forum, how will people come to the conclusion / realisation to turn it of, as to it not being TR problem, I don’t understand how a software company can produce a tool to help you improve and because it doesn’t work, it your problem not theirs … if Wahoo produced a update where ERG didn’t work, nobody would be saying … you don’t have to use it. stop complaining, you don’t have to use it, use resistance mode
um ok. I didn’t say anything about people not reporting bugs and I didn’t say the “latest much vaunted feature” is a pile of shit, which you claim to agree with me about(?); not sure how you arrived there but anyway…
Sorr if it was unclear but I was responding, in my post to niterider, to the fact that - the prediction caused them to behave differently and in a way they don’t like.This has nothing whatsoever to do with whether the feature works or not. (my initial response to the OP was regarding a simple misunderstanding I believe that person had about the AI “punishing” them for accurately rating a workout. Different situation but also nothing to do with bugs or glitches) By all means, if you find it buggy or its doing weird shit (plenty of threads on here about the outside ride bug) then obviously raise a flag. That’s how the platform gets better. But if you’re going to raise a flag and then loudly disparage the platform or the specific features then its only fair to have the community scrutinize what may be causing that for you since many many riders don’t have these problems. Its funny how many people come on here all full of rage and then don’t let anyone check their work or are unwilling to entertain and discuss plausible reasons for problems that aren’t just “TRAI is shit”.
And your analogy to Wahoo producing an update where ERG mode doesn’t work doesn’t really hold up IMO. For one thing, AI prediction does work, for lots and lots of athletes. But for another and more important thing, turning off AI prediction doesn’t cost you anything at all. If you don’t like it, turn it the hell off. The workout selector still works the same, the platform still works the same. In fact, behind the scenes, its working better overall; the AI has workout selection dialed in much better than it was before. For everybody? No. But it wasn’t working better for everybody before the update either. There are always edge cases and those people can and should raise alarms so the issues can be addressed.
But I’d appreciate you not conflating me taking issue with someone complaining that they are unable to discipline themselves to use a tool honestly and blaming it on the tool with someone having a legitimate software problem. They are not the same thing and I have never said so.
Two things can be true at the same time. We can like the predicted FTP as a feature but not like the way it is behaving.
If a significant number of TR’s long time customers feel that the platform is not living up to expectations, that sounds like more of a TR problem to me.
If this were one or two people complaining, I would dismiss them as outliers. But it seems that there are a substantial number of people who do not like the way the predicted FTP is behaving.
With all due respect, this is a more complex issue than just “Why don’t you turn it off?”.
I very much take your point in the sense that “just turn it off” is an absolute bullshit “solution” to people having technical problems. I hate it when people pull that move. We agree 100% on that. But I have reread your only other post in this thread (twice to be sure) and you only ever commented on the psychology of using the prediction. It changed your behavior. You were trying not to disappoint it, so to speak, which is quite a different thing and not really tied to whether the tool works or not. You made one claim in your last sentence that the “metric is unreliable” but then failed to elaborate any further.
You now appear to be updating your complaint to include whether its giving you the correct feedback at all and is therefore technically flawed in some way. In other words, you are crushing workouts and your prediction is plummeting, that sort of thing. Well that’s cool, let’s talk about that
What seems to be the problem? Can you make your calendar public? People are very helpful here about that sort of thing and TR staff are happy to get involved.
As far as liking the feature and not liking how its behaving, well that was the thrust of my response to the OP’s initial post. There was a claim that OP rated a workout harder than the AI guessed he would and he “got punished”. (I’m using “he”, could be “she”, I don’t know) This is insane logic to me. Your coach says “you should be able to do A easily, go do it”. You go and do it and it was hard. Your coach has no choice but to reassess your fitness. How is this punishment? And therefore, what don’t you like about this behavior? Granted, that was the OP’s experience and not yours - you haven’t given us any information about why you believe the AI is giving you erroneous data - but it’s an example that hopefully clarifies why I don’t see this (yet) as a technical issue.
I have posted in other threads about my general experience with the AIFTP and it is similar to what many other people have posted. Basically, the problem I have with how the tool behaves is that it promotes an unrealistically ambitious predicted FTP 28 days out, and then, even if you follow the program fairly religiously, the predicted FTP slowly (or not so slowly) goes down and down and down to what ends up being probably a realistic number but a far lower number than what was initially predicted.
I have made my calendar public and people have been very helpful, including the TR folks, and that is all much appreciated. The general feedback I received was that if I am happy with the workouts that are being scheduled for me, which I am, then the program is working as it should be and I should be fine. That is all well and good, and I agree that is the most important function of the program, and that is why I am still using TR exclusively and religiously. However, that does not mean that every aspect of the platform is performing to expectations and I would still submit that the AIFTP feature is not performing to expectations. I still maintain that any AI engine that predicts a number which (consistently) ends up being wildly wrong is not a good AI engine.
That is what I mean when I say that the AIFTP is unreliable. If I check my AIFTP on day 1 of 28, it will say I can expect a 10% increase. Then, on day 14 of 28, if might predict a 5% increase. Then, on day 27, it will be a 1% increase. I am making up numbers here, but this is fairly typical of what I have experienced (and I am following the program strictly and not riding outside).
Again, two things can be true at the same time. i can love how TR structures workouts for me and I can believe in the workouts I am given, and I do. But I can also not like how the predicted FTP gives me wildly unrealistic numbers and then inexplicably retreats from those numbers to numbers that have nothing to do with the original forecast.
https://www.trainerroad.com/forum/t/how-it-started-how-its-going-aka-predictions-vs-reality
I don’t know if I link it correctly but check out these post.
The Prediction has been working, I have even beaten it by 1 or 2 watts higher.
Again this has been if I’ve done all my workouts inside and following the plan.
One cycle even had two workouts as Very Hard RPE and somehow didn’t get a Prediction drop.
It maybe like this for more people that why TR still has it.
From the looks of your calendar you been steadily getting stronger keep it up
I am with you on the frustration. My predicted FTP dropped 5W on the day it should be updated and the day before was a short and simple pre race workout that is really just a warm up. No challenge in completing it. So what happened overnight?
I agree the marketing is annoying, and is annoying as well that AIFTP will drop at the slightest deviation. For example if after a VO2max session i keep going for 20-30 mins in zone 2 to reach home I have seen my predicted AIFTP to drop even if the workout was completed correctly.
it’s says BETA, hence I believe it does not have the maturity of the product and in the forum there is enough feedback to TR that clearly indicates what needs improvement. I would prefer an FTP test to be inserted in the workout plan and just get a reading. This though will not avoid that sometime even working hard does not yield the hoped improvement.
I did read, and I just now re-read it. You’re right and I think you have a valid point, so your gripe with the way AI FTP is marketed and highlighted in the platform could turn off a lot of users like yourself considering how precisely you followed the plan.
I’m curious, do you think that the ramp test overestimated your FTP, and now that you are a couple of months in do you notice that you are indeed stronger? As a new user, do you think you are getting your money’s worth or is this part of TR a deal killer for you? I ask the last part because I’m the kind of user that has come to ignore that part of TR in way that FTP predicted number is something that I take with a big grain of salt.
Except that the predicted number essentially becomes the number from which all your workouts are derived. After a pretty successful block where I nailed my workouts, my predicted number continued to drop after each workout despite the workouts feeling and being rated as AI indicated they likely would be. My FTP is predicted to be exactly what it was when I started the block. Was it too high when predicted initially–maybe, but there’s just no way it’s where it was a month ago either. It’s hard to ignore it when it’s the basis of how your plan unfolds. I appreciate the TR gangs efforts, but it’s becoming frustrating to use the software (for the first time–and I’ve been here from the start). Really hoping they get it sorted soon.
Just completing my 4 week training block and had my prediction drop from +10% at the start to flat. Not too disheartened though, had some struggles with heat, injury, and didn’t 100% every workout. I’m sure I’ve gained a lot of fitness in this block even if it’s not registering in the FTP number yet, so I’m just writing it off as a block to consolidate my recent gains. Was able to increase endurance volume nicely during the block and I’m looking forward to a nice long rest week and hitting the next training block fresh. It’s nice that every four weeks is a chance at a fresh start.
Yup. I had a whole training block where I nailed all the workouts, rated them as it expected to, didn’t miss any except one sweet spot workout at the end of the block, and my AI FTP dropped by 5. Not the prediction, but the actual FTP. All that work and missing one workout means I lose 5 off my FTP? No way. If that’s true, then the underlying training is very, very bad.
If you’re on 4 week training blocks with sweet spot once a week, then skipping one sweet spot workout is essentially skipping a third of all sweet spot assigned for that block. Probably not bad enough to drop your FTP as it did but still a significant amount of work being skipped.
, and should be turned off, but if you don’t post in the forum, how will people come to the conclusion / realisation to turn it of, as to it not being TR problem, I don’t understand how a software company can produce a tool to help you improve and because it doesn’t work, it your problem not theirs … if Wahoo produced a update where ERG didn’t work, nobody would be saying … you don’t have to use it. stop complaining, you don’t have to use it, use resistance mode
