Really, where did they say this?
Probably the best way to look at each phase of training rather than the predicted FTP changes.
You didn’t do any intervals that should increase FTP. (IE, sweetspot, sustained threshold, or overunders) Its weird to me that you were predicted much FTP gains at all. This was a specialty block, you only did VO2, anaerobic, and endurance. You should expect to have increased performance around your short power, and basically nowhere else in your power curve.
This is one of the big downsides of the current system. Not every phase is intended to increase FTP, but we’re being presented with FTP changes as the primary method for comparison.
This was also a low volume block. Volume is one of the biggest determiners of overall aerobic performance, and will have a big impact on FTP. It takes time though, and isn’t necessarily represented well in the 28 day prediction windows. Its just a bit of an artifact of how the model is tuned currently.
The good news is your next block looks more FTP focused. Traditional coaching wisdom is that you need to do VO2 work to raise your ceiling, then more FTP work to bring the FTP up to your new ceiling. You’re lined up nicely for that.
Well I don’t know if they say it explicitly. Maybe more clearly when you have a response that is out of the ‘money’ on the prediction scale you will get an update to the downstream predictions.
Now I assume it’s not just your RPE response but that response plays a major factor. So assuming the prediction is 55 Hard, and 45 Very Hard, and you select Very Hard here’s the likely outcome. It shifts future workouts, those shifts in future workouts shift your ‘predicted ftp’.
I assume this goes either direction. If it is 55 hard and 5 moderate, and you rate it moderate (again assuming your body reacts in a moderate fashion as well) then your training will change, and likely…your ftp prediction will go up.
The thing about all this for me is that I have a tendency to be a go along guy, so even if independently I might have rated something slightly higher if I see it thinks it’s hard then I’ll probably mark it as hard unless it genuinely was much harder.
This is also why every panel that allows you to select the RPE should have a literal definition of what it ‘means’ by each RPE level. i.e. If it’s hard you could do one more interval, very hard you would not be able to, etc.
This also gamifies the RPE, you can decide how you want to improve your chances, some of these things (like proper rest, proper recovery, proper hydration, focus) are great, some (saying it’s hard when you are dying) are probably bad. 
Now that is all ‘opinion’ so here is the anecdata 
I rated a workout very hard and it dropped my predicted ftp 9w. It was right to do so because the prediction was waaaay off, and the new prediction turned out to be closer to possible.
I like to think of it like ‘storm prediction trackers’. Any future outcome is predicated on the thing that has transpired.
Your FTP on trainer road.
SST/threshold and O/U can increase FTP (but not always). Vo2 work can also increase ftp. I would expect a block of Vo2 work to push someone’s ftp up if done correctly.
Agreed, but you might not see the gains right away. Sometimes it takes a few weeks to materialize.
Yes. Everybody is conspiring against me to make my life harder. One day I’ll prove it I swear. Lol.
And don’t get me started about Area 51. I know we have aliens captured there.
This is the big thing OP. FTP as a number is weighed too heavily. You are not your FTP. As long as the workouts are doable and you’re progressing through them without struggling, you’re likely getting stronger even if your “FTP” doesn’t go up.
Explain how a machine learning model is going to show the same behavior that an LLM has to please you?
Let’s try to stay on topic. I don’t think anybody wants a tangent AI discussion that won’t help anybody.
I agree, but I don’t think you ever answered the question, did you?
I mean, rate a workout Very Hard and see what happens. Every time it’ll decrease your upcoming workouts, which shouldn’t be the case in my opinion. Especially during a VO2 block which is going to be pretty much every workout if you’re doing them correctly. With TR AI, if you rate a workout Very Hard it will give you an easier workout next time so you don’t rate it Very Hard. So put another way, they don’t want you doing Very Hard workouts.
Heck, even if you’re looking at subbing workouts and they predict a Hard rating most times they won’t recommend it. So in that case it actually is telling you not to do hard workouts; literally not recommending hard or very hard workouts.
Interesting. That hasn’t been my personal experience. If a workout is supposed to be heavily skewed toward Hard and I rate it VH, yes, I would expect it to make the next one easier. If it didn’t, I’d be disappointed and think it wasn’t working. But if it predicts VH and I rank it VH, I haven’t seen that have a downward influence.
I think you are right on the money, and I am ok with not seeing improvements in FTP. But it is somewhat demotivating to see the one metric you are presented with decline for every week of effort. But hey I am happy with the workouts I am being prescribed.
strong text
I’ve seen quite a few comments like this in various threads - but it doesn’t make sense to me. I would have thought it was very rare that personal bests are set during Workouts - as intervals are usually sub-maximal because they have to be repeated. All my PB’s come from single efforts or races.
Am I missing something?
I guess it makes sense that a rating of “very hard” should lead to a decrease in intensity of upcoming workouts, and a downgrade of the expected FTP - it seems reaonable that the model interprets this as it has overshot its expectations of my level.
I just came across another thread on much the same, and there is a very nice post from Jonathan adressing these issues:
I made the same experience. First prediction was +20W. With a nearly perfect 28 days of training (only one incomplete workout) I ended up with +2W.
Second prediction was again +20W. Again with a perfect consistent block I have only +6W left with 8 workouts to go. I assume I will not see more than 2-4W in the end. I rate the workouts honest with some „very hard“ ratings which downgraded the predictions.
I would never ever expect to see +20w in one block. That‘s why I try to shift the focus away from the prediction.
However the FTP prediction helps to make decisions when you think about modifications in the plan or finding alternatives. You can directly see the consequences.
I can recommend more to concentrate on consistency and collect the single watts over each block which will sum up in the end.
I‘m super happy with the workout selection and think that the training is close to perfect in terms of balance between intensity and rest.
Good to hear your experience!
Depends. This is my opinion btw. But if you do a sweet spot workout and rate it very hard, yes maybe the next one should decrease a bit. Depends on why it was very hard and/or the timing within a block. But Sweet Spot shouldn’t be Very Hard. If I had terrible sleep and didn’t eat all day and had a stressful day at work, the came home and a sweet spot workout was very hard, I wouldn’t change the upcoming workout because I know why it was hard. If I’m fresh and fuel right and it was still very hard, then sure it should downgrade. It all depends.
On the other side, VO2 max workouts should be Hard at the minimum, and probably should be Very Hard if you’re doing them correctly, even All Out. That’s the point. So if you do a solid VO2 max workout and rate it Very Hard, and it downgrades the following workout, that’s a problem. If I do a VO2 max workout and rank it All Out, that’s a good workout and it shouldn’t downgrade the next workout. VO2 should be very hard.
So in my opinion, to downgrade workouts following a Very Hard rating is counterproductive.