My FTP dropped 40 points with the new AI rollout. I have emailed several times, and everyone says that seems to be accurate. Can’t be. I am barely working out at all at the new levels, and I signed up for the Hill Climbing training session. I have entered an FTP that I feel is more accurate (still lower than before the rollout), but then the AI program doesn’t predict where I will be in the future. So, if I have to manipulate the data, why use this program at all?
2 Likes
If you post screenshots of your calender we can give our opinion if you want?
If you have a plan in place it’s good to see the difficulty levels of the future workouts picked for you and their predicted RPE.
2 Likes
Also good to see the week’s prior to the AIFTP detection to see what level workouts you were doing.
I’m in a similar situation. My current method of managing it is to effectively toggle the AIFTP detection by accepting it and subsequently deleting the acceptance prior to a workout. That way I’m able to be entertained by the AI’s predictions but still complete appropriate workouts at a good level.
1 Like
But changing the FTP shouldn’t change the difficulty of the workouts at all? 
Any change in FTP should be completely cancelled out by an equal and opposite change in workout level.
2 Likes
It does if you don’t let the AI set the workout…
1 Like
I don’t understand this response. If your FTP is 200 and you are slated to work 5% over, you’d be at 210. If your FTP gets lower by 40 points, or 160, and you are slated to work 5% over, you’d be at 168. Is there something I am not getting here?
Assuming you let AI set your workouts… then the %FTP and durations will change to make the difficulty of the workouts consistent regardless of FTP…. that’s the idea anyway.
With an FTP of 200w you might be prescribed level 1 workouts but with 160w you might be prescribed level 4.
4 Likes
If you want folk to take a look, you need to post screenshots as already suggested, or post a link to it (set to public)
3 Likes
Hey @bmc_58, I know you have a support ticket open, but I can take a look as well.
As everyone mentioned, posting your username, link to calendar and screenshots of what you are referring to gives more context to be able to see what may be happening here.
1 Like
I was curious about this as well. I’ve not done any serious cycling in several months. Maybe a 60-90 minute fairly easy ride here or there on our local rails to trails path, but no consistency whatsoever. I was in Hawaii last week and decided to rent a bike to enjoy the warm weather. To my surprise, I was able to do 2-3 hour rides with an NP of 221-237. Holding 240s felt like an “all day effort”. I figured all the running I had been doing helped me to maintain some performance for the bike as well.
Out of curiosity, since I had not been on TR for a while, I did the AI FTP detection and it said my FTP was 244. Considering how I felt great riding with an NP of 237 for two hours, I thought to myself “that can’t be right”. Maybe it is, but in the past, when my FTP was measured at 323, I could hold 330s-340s on 30+ minute climbs but definitely did not feel like I could ride at 320+ as an all day effort.
Is it possible the AI FTP is underestimating FTP because its based on my lack of rides for the past several months, and the fact that the ones I did were all easy (usually NP 170s-180s, or less)??? Running must be having somewhat of a positive impact but running is not recognized as a contributor to my bike fitness by AI FTP (I assume)?
I think this likely the case
1 Like
I used this odd window to increase my power levels. I basically went up a full integer across the board every week for 3 weeks.
No way I could do that normally.
I looked into your account and here is what I’m seeing:
Takeaways
TrainerRoad AI’s model seems to be giving you a lower FTP so it can give you workouts that will do a better job of growing your fitness. It’s crucial to remember that if it drops your FTP, it won’t give you the same workouts it would have given you at a higher FTP. It would give you harder workouts (read: higher intensity and longer duration with less rest) that would still feel appropriately difficult.
At your current FTP of 113, it will have the greatest ability to give you the workouts you need to get faster. No need to manipulate survey responses or manually tweak things–it’s got you taken care of!
- There were no indications from your power data over the last 90 days that your FTP was at 154w in early January. When you had your FTP set to 154, the model was trying to give you the easiest workouts it could find, but they weren’t easy enough at that high of an FTP, so you failed 4/4 hard workouts in the space of 2.5 weeks.
- The new AI FTP Detection suggested your FTP change to 113 on Jan 21. During the only hard workout you did in the week you had that AI FTP Detection-derived , you turned up the intensity to the point of hitting 164bpm (had recently failed workouts at HR as low as 160bpm) then ranked it as “Easy”, so the model might perceive this as is a bit of a mixed signal, but regardless, you changed your FTP back up to 150 the next day.
- Since then you’ve done one hard workout (yesterday) that you crushed(!). The workout was predicted to be “Moderate” in difficulty. You exceeded your power targets and hit HR around 159, but still rated it Moderate, so that’s a bit perplexing.
- This shows the model is trying to do it’s best to give you the right workout. It was struggling to do so at 154, could definitely do so at 113, and might be able to do so now at 154, but tough to tell from one workout where your difficulty rating doesn’t seem to align with the effort indicated by your HR.
- As of now, it is predicting your FTP will increase 12.4% from 113 up to 127 by Feb 18. 12.4% in two weeks is a huge increase, but your workouts look about right. Even though your FTP is lower, the actual wattage you would be doing at this lower FTP seems to align with what you could do. You have hard Sweet Spot workouts where you have intervals at 100-106w with minimal rest, Threshold workouts with 4x9 Over-Unders where you’d average around 125w, etc.
In-Depth Review
Going back to the week of January 5, your FTP was set to 154 and you had 4 workouts above endurance between then and January 21 (when you accepted the 113 FTP from the new version of AI FTP Detection).
Here is the report on those high intensity workouts:
-
January 5: Rejected new AI FTP Detection number
-
Sunday, Jan 11: VO2 Max Attacks with 2.5min sets at ~154w. You rated this Max Effort and weren’t able to hold the power targets.
-
Thursday, Jan 15: Threshold Hard Starts with 8min sets at ~147w. You had to turn it down and weren’t able to finish, noting the reason as workout intensity.
-
Saturday, Jan 17: VO2 Max Attacks with 6x1min Hard Starts at ~200w. You weren’t able to hit those numbers and were 30-50w low on the final 3 intervals, but you just rated this one as “Hard”.
-
Tuesday, Jan 20: Threshold Hard Starts with 5x4min at ~152w. You weren’t able to hit those numbers for the first two so you turned down the intensity by as much as 30% on the last set and just rated it as “Hard”
At this point the model was trying to do it’s best with your training, but for every hard workout you had, which were all toward the bottom range of difficulty in those specific workout categories, you weren’t able to hit your targets and/or needed to turn down intensity. These are all signals that your FTP is too high.
Further, looking at your recent power records, in the last 3 months your highest 20min power has been 139w, and your highest 60min power has been 127w.
On Jan 21 you accepted the new AI FTP Detection of 113w. Here are the workouts you did with that:
- Fri, Jan 23: A 60min Endurance workout that was intentionally “Easy”, but you turned up the intensity for the first half to make it harder. It makes sense that you marked this “Moderate”, but it was supposed to be an Easy Endurance ride.
-
Monday, Jan 26: A Threshold Hard Starts workout that had 8 sets between 3-4min at an average of 134-147w. This was predicted to be Moderate, but you turned up the intensity significantly and reached an HR of +160 on 4 of the intervals. For context, you had failed workouts at an HR in the 150s before, but you rated this workout as “Easy”.
You then manually updated your FTP to 150 again and have had 3 endurance workouts and one VO2 Max workout.
Your out-performed yesterday’s VO2 Max workout which had 5 sets of 5x35 second low VO2 Max efforts at 159w. You surpassed your power targets and hit 160bpm again in this one, but you rated it Moderate.
20 Likes
This is bad advice in the case of this athlete. This was the initial approach that resulted in 4/4 workouts being failed.
6 Likes
Hey Kjdhawkhill, I get what you’re saying, and you’re not the only person on the forum who’s thinking this way. “If my FTP is 10-20 watts, higher, TrainerRoad AI must be giving me workouts that are 10-20 watts higher”.
I’ve said this a few times, but the AI doesn’t know your FTP and isn’t basing its recommendation on it. In our system, a lower FTP will give you longer workouts with fewer watts and more TIZ compared to a higher FTP.
I went into your account and looked at what Saturday’s threshold workout would be at 302 vs 328, which is what we wanted to set you at.
Best 10 Minute vs Planned Workout
At a 302 FTP, you’re doing 309 AP.
At 328 FTP, you’re doing 310 AP! Literally one watt difference, even though you’ve got a 20-watt FTP difference.
This is what I’m talking about when I tell athletes to look at the actual watts you’re doing rather than the FTP.
For NP, at 302 FTP, you’re doing 312 NP.
And 328 FTP, you’re doing 321 AP. This is mostly because we’d have you on an hour workout at 328 vs a 90-minute workout at 302. That’s still 10 watts below your best 6 week 10 min NP of 331, but 11 watts above what you would do at 302.
If you look at the images , you can see that the 328 threshold is keeping you right at recently power that you’ve done, but in a VO2 max workout and only predicting it hard vs the very hard you put.
The 90 minute workout does have you do a little bit more power from 60-90 minutes, but it’s not the sort of power that’s going to push your FTP up like the other workout.
You’re at 3.1 watt/kg. Because of that, we recommend you work on raising your FTP through higher watts rather than longer TIZ.
Let me know if you have any questions.
Note: This tool is underdevelopment to help athletes better understand the wattage of their workouts and how it compares to what they’ve done recently. We’re trying to release it ASAP.
19 Likes
Cool, where do we sign up for this beta?
2 Likes
No beta for it, we have a few display bugs to iron out then we will launch it. Hopefully very soon.
12 Likes
Hello Nate, I’ve recently made a post just trying to understand what you just posted about how the system just looks at watts and how workout progression.
I find it very interesting but I just have trouble understanding how this works.
Nothing against the new update just looking to learn.
Is there a blog post or forum post or even this post just explaining this super simple?
Sorry I just have a hard time understanding how anything training related works, I just see calendar and do the workouts 
No, but we’re going to do a podcast when it launches and we talked today about having a blog post too. Let me check your account and I can show you, one sec.
3 Likes