Every time I finish a planned workout my projected FTP goes down. I’m following a Trainerroad generated custom training plan to increase FTP, its set at balanced. I do most of ny workouts outdoors, I do inside those harder to do correctly outdoors like 30-30s or recovery/endurance. All threshold, sweetspot and longer VO2 are done outdoors. I think I do a decent job at hitting the intervals (todays workout here as an example). Before doing the workout the projected FTP was 241 with 6 workouts to go, after it is 237 with 5 to go! This has been going on for weeks now, but this has been the biggest move down. My theory has been that my commute back from the workout was impacting the plans ramp rate. I do all my workouts in a climb that basically starts from my house, so I do intervals uphill and then pedal back down. That way down is not 100% down so I do pedal some adding between 5-19 TSS, today I was extra careful and added only 3 TSS (shown below). No idea whats going on. Last month the plan started with a nice projected FTP bump and ended up loosing a couple of watts having only missed one workout, and even after missing that one was still showing a good bump.
Hello @dsacasa 
Can I have permission to look at your Calendar and post images per the new forum rule here? This way I can help figure out what’s happening with more context.
If not, I’ll have to recommend reaching out to support@trainerroad.com
Sure, have a look. Thanks
This also happens to me. Nail a workout, feel great, projected FTP goes down and (importantly) future workouts get easier. Clockwork.
I’m not really complaining, because my intense workouts tend to be the perfect difficulty. But I kind of wish the 30-day projection wasn’t so optimistic? And if I’m doing something wrong that’s preventing me from progressing at the projected rate I’d like to know.
Im having a similar experience. Finishing the prescribed workouts, either indoor or outdoor, and the predicted FTP is always decreasing.
Mine has done something similar this week. Originally 4 weeks ago it said I was going to go from 313w to 330w which seemed very optimistic. I haven’t really missed any workouts and rated a couple of them very hard so I expected that to drop. It went down a little bit over the course of the next 3 weeks. At the start of this week It was predicted to go to 324, then I completed the workout on Tuesday and it dropped a couple of watts. Then I did the vo2 workout this afternoon outside and it dropped another 2 watts. I rated both of these workouts hard. I also did a track run workout this evening (I cycled too and from track but a nice easy pace) and it dropped a further 2 watts!
I’m also wondering why it aggresively dropped me this week when it had been fairly steady in its prediction for the proceeding 3 weeks. Hopefully I hang on to a few watts of gain by Saturday!
@dsacasa This is my pessimistic view of projected ftp and it stings me as I wish it weren’t true - it is a marketing tool not a fitness tool imo. It reads as “follow our plan and get huge gains” like many other platforms tout. TBH Trainerroad has amazing plans and will make you faster, but get into training in this sport long enough and you realize that progress is slow and non-linear. The FTP projection for me has done the same - execute and see it fall. I turned it off, remember you are not your ftp, and you are certainly not your projection. Hit your workouts and focus on doing the little things that make you faster - sleep, good nutrition, consistency. FTP is just a number - you’re crushing it no matter what the projection says
Not really that simple though. The new AI FTP lowered nearly 70% of users FTP according to TR. What was the marketing advantage of that? There are also many users who are seeing bumps of 1, 2, 3, watts over the next 28 days. Users that load plans and show a reduction in FTP.
Some people saw bumps, some people are seeing higher than usual suggested gains in a short window of time. Not the majority just some people. Others took off for a few months around the holidays and are coming back so the gains are just regains from what they lost, they look huge but are just from getting back into it.
It is a marketing tool, every feature is, but it isn’t a marketing tool how you are saying it is.
Conspiracy theory: TR is slowly inching us back down to the old FTP numbers after all you were raging. 
Not talking about the FTP number at all. I’m talking about the projected ftp it is forecasting. Imagine you’re new to TR you join on and what you’re greeted with is “follow this plan and you’ll increase X watts by the end of this cycle.” That projection is a useful marketing tool to say that their plans will get you gains. The “what is ftp” debate is a different topic altogether. Note that the best way to see a massive jump in projection is to utilize their recommended plans (which I still think are super solid). It’s a confidence point so the buy in at the start of the cycle feels nicer since you expect the big jump.
My theory is that the FTP number will be just right in a few months, just a little variable in the first 2 months. Workouts are already good, right?
Joe
PS if theory A isn’t right then theory B is that even a little off plan riding makes a big difference
Add me to the list of ever plummeting FTP prediction. I’m 1 workout away from completing the work portion of the second block of this plan (next week is the recovery week) and I’m sufficiently unimpressed with the FTP “prediction” that I turned it off.
At the start of the first block the robot adjusted my FTP down from 309 to 302, but gave me a prediction of … 309. A month to get back the 7W “lost” looked a bit overly optimistic. And it was. The prediction dropped every week and wound up with a gain of 1 W - to 303. Or maybe 304. It gave me two predictions on the same day. One with the app on my phone, one with the web UI about 4 hours later. I was racing in the Zwift WTRL during that block, so that may have contributed. I did, however, have those races in the calendar instead of hard days, so the I expected prediction to have taken them into account.
At the start of this block the robot predicted an FTP of 311 (I laughed). It’s been declining virtually every workout since. I did miss one workout with a cold (which cost 5W all by itself), but have completed all the rest, with all but one being just hard. And the last hard day of the block is supposed to be a work day, but it’s giving me a threshold with a level of 0.9, while the last threshold I did was a 2.7 which I rated as hard. The “threshold” levels of that workout are actually tempo. The predicted FTP (before I turned it off) was a flat 304.
I understand that the predicted FTP is not used for determining workouts, or measuring plan progress, etc. It’s calculated based on … stuff. The problem for me is that the prediction is either pure fantasy, or I’m doing something that’s making me miss the prediction. But there’s no feedback as to why the prediction changes, so there’s no action I can take, and no understanding of what’s happening. An initially aspirational number turns to regret quickly, with no apparent cause. Life’s too short for that, so I turned it off.
In the spirit of the new guidelines, here’s a couple of screenshots of the last block, and this one:
And I turned FTP detection back on for this one
Maybe I am misunderstanding but why would the ftp prediction be more accurate in 3 months but less accurate in the first 2?
At the start of the block, the AI puts in (made up numbers alert) 16 total workouts. 10 endurance rides. 6 interval sessions, and decides what you’ll be doing on each of those days, and how it thinks you’ll rate each workout.
The total amount of stimulus (combination of total hours and intensity therein) will drive its AIFTP Prediction.
So, basically, if you do the workouts and rate them how it expected, you should arrive at the end of the block in exactly the shape it thought you would and get the number it predicted. At least, that’s how I understand it - in theory.
So, to flip that around, if you don’t execute all 16 workouts perfectly, and don’t rate them as expected, you shouldn’t expect to arrive at your destination.
The first question we all need to ask ourselves is “Did I actually do my workouts?” If not, there’s no reason to expect we’ll arrive at our destinations. No skipped workouts. No workouts performed at reduced intensity. No failed workouts. Then we can look at RPE ratings. “Have I given a rating that would reduce my prediction?”
The tricky part, of course, is that comparing our path through the block to the path we were supposed to take is almost certainly going to be impossible, because as soon as we deviate even a tiny bit, the remaining path through the block is recalculated and the workouts change.
As I said previously, going forward taking screenshots of my upcoming block on the day I get an FTP Detection is going to be A Thing for me. That way when the block ends I can compare what I actually did with what I was originally scheduled to do and see where the watts went. 
We’ll have to see your training for the proceeding weeks as well to give an educated guess. But my main question would be, If you did a WL 2.4 Threshold rated hard the week of your FTP update, how much harder do you really want your workouts to be?
Here is the previous four weeks. One of the rides I had to move from sunday to monday so the third week is longer but I haven’t missed anything. Oh I’m not really complaining because I thought my previous FTP was too high anyway and the predicted FTP bump seemed incredibly unrealistic. I will take the 5w increase and run. It’s just quite interesting that the prediction really started to drop after the last couple of days. The prevoius weeks it was somewhat stable in what in predicted.
Your training looks good, and I really don’t see a good reason for the 2.4 threshold workout you got this week. You seemed to have gotten on top of the 3.9 WL workouts last block and also doing well in the VO2Max workouts. For some reason the Ai wanted to hold you back a bit when starting this block which has resulted in the prediction dropping, but as long as you feel that is a good number than all is good.
My last tip would be that if you in the future feel that you’re being held too much back you can try switching to a bit harder workout and see how that goes.
If you’re going to claim it’s marketing you can’t ignore that many people are upset that the new system lowered their ftp. I understand that is an the specific part you mentioned but it’s all related.
You also can’t ignore that people have put plans on their calendar and the projected ftp is lower than where they are. How exactly is that a marketing tool? It’s doing the opposite of what you claim they are doing for marketing purposes. This happens to be a thread about people getting numbers that drop, but there are many not seeing this. No gains or losses have been noted.
multiple things can be true. we are saying two different things and that’s ok.
It’s a great tool to market their plans and their mission/claim to give you the right plan at the right time where you can feel more confident that a plan from them at the training approach the AI sees as best for you will trend you upwards toward a higher FTP. (Regardless if the number for the FTP you train at drops when the transition happened)
the way I think projected FTP could be used well is to view it directionally not objectively. Plan A at training approach aggressive is trending me downward but plan A with training approach balanced is trending me up or neutral, I probably then should expect a better response from balanced (replace any plan or approach in this metaphor and focus on the directional progress or decline it projects).
The use is actually pretty cool if you detach from the number itself. If I designed it I would’ve put it as a range or a visual up/neutral/down instead of a number to be honest. We’ve seen similar tools in lifting, weight loss, even learning instruments that help encourage users to pick a program best suited for them. I think most irl coaches would agree that an athlete or students belief that the training they are doing is going to “work” is an essential component of getting the consistency needed to get efficacy.
Isso também acontece no meu plano de treino do TR. Acho muito estranho várias questões nesta projeção de FTP pela IA.
1º Acho difícil em um período de 28 dias um atleta que não seja iniciante elevar a FTP em tão pouco tempo. (acredito que o período projetado deveria ser maior)
2º Essa semana executei um treino de VO2 de nível 5,5, automaticamente a projeção do FTP aumentou. No dia posterior um treino planejado pelo TR de Endurance de nível 1,9 diminuiu a projeção. (Treinos de recuperação prejudicam a manutenção ou evolução do FTP ?)
3º Outra grande dúvida que tenho é se não existe o conflito entre evolução do FTP e treinos planejados com foco em provas registradas no calendário. Pq existe o período de “polimento” antes de cada prova para te deixar “fresco” para o dia da corrida.
Acho a iniciativa do TR muito legal e ousada, mas acredito que deva ter muitos ajustes ainda para colocar em produção para todos os usuários.