Depending on how long you have before your next FTP Detection, changing one workout can have a butterfly effect of sorts on the rest of your training.
Pinning workouts isn’t something I recommend often becuase it forces the selected workout into your plan even if it’s not ideal and will derail upcoming training session.
It’s next Thursday. Two hard workouts before. Okay I’ll avoid pinning workouts.
Quick update…yesterday AI detection, again lower than last one (except it even though I think I will spiral down instead of spiral up). New detection for the next 28 days was 27w+ today first workout, immediately lost 3w.
I was able to do some digging, and I can see that drop in FTP from 357 to 346 was not due to you changing your weight. The drop came first, then you changed your weight. That’s what the back end shows me. 
Eddie, Can you see what triggered the drop? I don’t recall doing anything in the app other than clicking around (Calendar, Account, etc) and then updating the weight. I changed my workout today to an outdoor group ride. When I checked the estimate after the ride, it was back to 357.
This is not quite correct. The prediction may be inflated as you suggest, and slowly decreases with workouts. However, at the end of the block the AI DETECTED FTP matches the final PREDICTED value. Effectively, 4 weeks of BUILD, followed perfectly, results in no gain.
So if the predicted FTP is inflated and this inflated value is fixed, what happens? Do you get a correct prediction that matches what the detected FTP is after 4 weeks? (i.e. no gain)
Looking for blindspots in my view of my FTP drop. Projection was 271 with four workouts to go before next detection. Dropped to 265, then 263, 262 and today after last workout it dropped from 262 to 253…One point lower than what it was on April 11th. Four of five w/o are outside. All executed very well and marked accordingly. Halfway through second week of the block. Swapped out one endurance ride for a solo ride and no missed w/o’s. I know they are working on the bug, just seeing if there is something else I might be needing to do or not do. Thanks!
Today I had a new AI FTP detection of 187W and it put my FTP prediction for the next update at 202W. Which, admittedly does seem really ambitious. But, after my outdoor workout today (Mount Baldy -2) which I feel like I did pretty good at (got a 91 workout adherence score on my Garmin). My average power for each interval was a few watts higher than what I would have been locked into if I was indoors on ERG mode and I was a little out of control in the first couple minutes of the first interval.
All of that considered, I ranked the workout as Hard (which the AI predicted I would) and then after the workout my entire calendar changed and I have a new FTP prediction of 193 for the next update.
Good point…Yes, build and base the ftp is not changing that much. So this could be anouther bug as the AI prediction is showing a growth..
It really is odd and my pattern loving brain is perseverating on this a wee bit now… I rode my 1hr 45 min endurance inside today in ERG partially just to see what would happen. My prediction was 251 before the ride. After the perfectly executed ERG ride 250. I rated the ride easy. Only factor I can think of is that it is hot inside so my heart rate a bit higher than normal. Prediction seems to get more sensitive as it gets closer to an update. Before rest week endurance rides it was predicting 254-256 which I think is a bit high for me.
I don’t want to speak for others but my main frustration is that for 3.8 weeks FTP prediction can show a modest or even significant bump in your projection and then in the last 2-3 workouts it can drop to basically a flat month or even a few W down. It’s just a bummer to work really hard for 4 weeks and then at the end of it feel like somehow you’re actually regressing in your training.
I think this will ultimately drive people away from the product because the entire reason we spend $20/month is to hedge against the risk of wasting time on the bike training. I’m happy to pay the subscription fee if it affords me peace of mind that my time spent on the bike is actually productive
exactly my issue. it is very demoralizing to have a good solid month of training and then have the AI tell me that all the work resulted in my getting less fit
especially when i felt i made solid gains
This happen to me today, I’ve had my predicted FTP at 317 for almost 30 days now, and just 4 days before my new FTP prediction it dropped down to 311 for no particular reason… super annoying 
I did a workout Avalanche Spire today and tried to keep the power on track, despite the traffic on the street. Comparing the Workout graph (blue graph) with the done activity (orange graph), it looks quite equal. But AIFTP has dropped the prediction once more.
The peaks at 45min, were the warmup surges, which was not displayed on the workout of the Garmin head unit. So I have not done them. Just followed the recommendations.
Maybe I have overlooked something, why the AIFTP-Prediction is dropping?
I’m guessing this is a by product of the AI FTP detection feature that was released and ballooned many people’s FTPs higher than they should have been. Now, lots of people are seeing drops because their FTP probably wasn’t that high in the first place.
That would make sense if people were struggling with workouts, rating them harder that expected or failing them. But there are DOZENS of posts with riders sharing their data showing excellent adherence to the program and no gains or even losses after a block. So it would seem that there is fitness gain from the training. Maybe it’s less than the initial prediction but a LOSS in FTP?
A conspiracy theorist might argue that many riders are at their genetic max already. The true prediction before a block will be flat or a loss due to fatigue. But the AI doesn’t want to admit that because you’d cancel. So it gives you the incentive to keep spinning the hamster wheel and subbing for 3.5 weeks and then slaps you in the face with reality in the last few days.
-edit: to summarize the conspiracy point: The AI isn’t focusing on your FTP (it may improve if you have room to improve). It’s focusing on keeping you subscribing to TR
How do you create that graph inside trainerroad? I don’t seem to be able to do so 
Search the workout on TR you have done. Click on details and you see the both graph, blue the workout and orange per default on last six weeks. Choose the setting of the orange graph for the date you have done this workout and you are there and can see this comparison.
My FTP keeps rising and last I checked I’m a paid member and not part of TR staff
I have detection in 2 days and on Tuesday it’s ftp detection time where I also have pioneer -3 - 1 hour endurance ride. Currently prediction says 254w FTP. If I move the 1 hour endurance today to Monday, Prediction increases to a woppimg 258w!! 4w ftp for 1h endurance ride? Can anyone explain this and could this also be related to this bug?
I think 254w fits good, so I would rather skip a workout tomorrow just because of an additional increase if prediction is right?
