We’ve had this future FTP prediction in x days for a few weeks now. It really hates if you miss anything or fail anything. I’ve used if for two cycles now (3 week load, 1 week rest). The first time it was estimating 318w in 30 days or whatever, it ended at 312. This last one, It estimated 324w in 30 days. I’m a week and a half in and it’s already down to 315w.
I struggled through one Threshold workout, and deleted two workouts next week, because I’ll be out of town. I initially liked this feature, but now I’m not so sure I want to keep it on. It’s demotivating. What is everyone else’s experience? Do you still have it on? Thoughts?
6 Likes
You are getting solid improvements, the previous lift was 2%. It looks like you had another workout that was pretty challenging as well recently. It might be beneficial to slow the ramp slightly.
3 Likes
I think of the feature more like a tool now.
If I had to move workouts around or miss a workout, I use this to help me plan the workout with the AI options that I got.
I try not to get too hung up on it now but when I do all my workouts on the trainer the prediction has been dead on, even when I rated two of my workouts Very Hard.
5 Likes
That’s some pretty good increases by only riding 3 or 4 days per week, and on not that many hours. The dropping prediction looks what I’d expect from my small view you showed.
I’ve gone through 4 cycles now, and I agree that the prediction really doesn’t like missing workouts and especially failing them. Deleting a key session is def going to give you a hit. This does make sense though.
I personally don’t put much stock in the number - you can only do what you can do with life commitments - it’s just giving you the straight goods!
My prediction is in exactly the place it started on my current cycle and only 3 days left - I didn’t miss any workouts and the effort was in line with all workouts predicted. As an aside, I even blew out my rest week ending yesterday, and it didn’t seem to care which I thought was strange. I did 650tss instead of 425tss.
5 Likes
That’s a good idea. I haven’t thought of it that way. It does help to see what happens when you move things around on the calendar.
2 Likes
With the old system you’d train all month, hit the button, and hope the number is bigger. With the new system you get that running estimate all month long. It starts off with the most optimistic result possible and then as you screw, up the number comes down. It’s like….this has always been the case but it’s been hidden and the effects of moving stuff, skipping stuff, and (the big one) adding stuff were all unseen. Now they are seen. I’ve gotta say, for me this has been super helpful. But my wife hates it so apparently there are different opinions.
Joe
10 Likes
Well…shouldn’t it? If I assign you homework and you don’t do it, shouldn’t that affect your grade in teh class?
If you find it demotivating, then I would definitely turn it off. As another user said, it can be useful if you change your schedule a lot, so you can try and optimize your workout structures by moving them around and changing them. So I would maybe suggest only turning it on if you have to make changes, otherwise hide it.
1 Like
This seems to be a common refrain among those frustrated with the new AI – it’s punishing me for this or that or it’s demotivating – and I feel as though there is a fundamental misunderstanding baked into this viewpoint:
The AI isn’t punishing you for missing workouts or changing things around or ignoring the prescribed workouts or rating something harder than expected or any of the other things that occur outside of the AI’s expectations; it is merely predicting what will happen in your body. It isn’t the AI’s fault that these behaviors reduce the efficacy of training, it is simple biology. The AI is getting sh*t for it because it’s the one lowering your number but it isn’t the algorithm’s fault, it is merely predicting–based on an enormous amount of data about you and many many many others–how your biology is likely to respond either to changes you are making to the plan or simply the plan as you have it configured. If the AI makes in incorrect prediction it will show in your subsequent performance and the prediction will correct. If it makes a correct prediction it will reflect in your subsequent performance and not correct. Your own physiology is the engine driving the FTP prediction not the AI. The AI is merely translating your body’s performance into a number you can see on a screen and extrapolating that performance into the future to give you something to work towards.
If you would rather not see how deviations from the plan (or even just the plan as you have it set up) affects your likely future performance then by all means turn it off. Maybe for you you’d rather just ride and enjoy yourself without worrying about optimizing your training. That’s totally allowed! 
4 Likes
You would think… but there’s a big push for homework to not count as part of the grade. Students will do it for the “love of learning” (so we are told by the consultants) and we should not punish those who have home lives that are not conducive to completing homework. Also on tap… as many retakes as you want, no late penalties and the elimination of zeros (lowest grade is 50%).
It’s the education equivalent of riding an e-bike. 
(Sorry to derail the thread, I am just an old school teacher that got triggered).
6 Likes
Today I did this workout, it was fine, but challenging because, well, duh. So I rated it very hard. AI took another 2w off my prediction. So now we’re reduced 324 to 314. So I redid the survey to hard. And now it’s 315. But it still took 1w off. Why? That’s what I can’t understand. Even when you do the workout and don’t fail, it reduces your future FTP. So what’s the deal?
1 Like
Some people just can not cope with it - I like it but maybe it should be off by default 
2 Likes
I have generally experienced the slow degradation of AIFTP, but on my latest block it basically hasn’t moved. I’m 3 days out, and I actually switched my V02max workout for today from 1.5 hours, down to one hour and it actually went up 1 watt! Unprecedented! My question: Why didn’t TRAI think of that??
5 Likes
yes, that’s happened to me before. I dropped a workout and it went up. It’s probably not a feature for it to tell you to drop a workout or downgrade it…
2 Likes
Keep in mind that the AI isnt just looking at how you rated the workout. It’s also taking HR into account (and maybe other data points?). So, even if you rated this is ‘Easy’, the AI would look at your HR, compared with other efforts, and add that to the equation.
3 Likes
My guess is that it is a rounding error. Maybe before you were 314.5 rounding up to 315 and then you rated it very hard. It dropped. Then you rated it hard and it recalculated and ended up at 314.4, so it rounded down.
4 Likes
This exact same thing happened to be today with a scheduled VO2 workout. I showed up, I crushed it, my FTP prediction for this Friday went down by 1 point. There is nothing inherently “wrong” with this. Yes, I did the workout as prescribed but my heart rate was a little higher than usual. Perhaps I was carrying a little more fatigue than is ideal. I do have a very physically active life perhaps I overdid it on my rest day. For one thing I had a disastrous night sleep last night. Lots of things can affect how your body reacts to a workout whether or not you “stayed in zone”.
Just because you “complete” a scheduled workout and rate it as the AI expects doesn’t mean there isn’t still something to interpret from your performance. How consistent was your cadence? How steady was your heart rate? Was it within normal ranges for this type of effort? Did your cadence betray higher fatigue than usual? I know I tend to be a little less precise with my technique as I get tired. A lot of possibilities.
3 Likes
I think the execution was impeccable. No cadence drop/increase, no standing, steady power. My heartrate stayed below threshold mostly. Never went above 170, my max is about 190.
Must be other variables. IDK Here is my strava file if interested.
1 Like