I would attribute things feeling better more to AT setting PLs correctly more than the intensity setting. Have you been using AT with ramp results for a while now?
Thanks Chad. Found this out literally a few seconds after posting my query. I always underestimate my ability to fail miserably when it comes to operating technology 
1 Like
So it tells me I need to wait 14 days since last testing. Fair enough. 
No, that’s not the case. My FTP was too high from Ramp Test results in the past.
My estimated FTP was 3.75% higher than my previous FTP test 8 weeks earlier.
Felt about right, but I wanted to verify this because I’ve had bad experiences with ramp tests in the past, they tend to overestimate my FTP and TR AI detection predicts a ramp test FTP result.
I Did a 20min FTP test today while targeting my TR FTP estimation * (1/0.95), I completely burned up half way through and was unable to finish
I am done with FTP tests for now, I assume overestimation won’t be much of an issue with adaptive training. Only my ego is a bit bruised 
1 Like
I was just curious at to how eFTP tracked with work during a block. I had no intention of accepting the result no matter what it showed until start of next block. I was not looking for “How much better am I after todays workout?”
Less frequently would serve that goal, yes. So the frequency is just due to the new toy.
3 Likes
Just to clarify, “past” being pre-AT? If so, I had the same problem. At least I couldn’t keep up with the progression of workouts outside of SSB unless I used a different method to set my intensity.
AT detecting PLs has largely addressed that issue. To the point where testing didn’t reveal any useful information. So PL Detection >>>> FTPD, but the FTPD is really nice for some significant refinement of AT for sure.
1 Like
For me it’s a bit of both.
I could score on the Ramp Test an FTP that I couldn’t hold, the AI-FTP I can, but I need to manually drop my Workout Levels even a tiny bit from my PL drop.
I don’t have any interest personally in trying to ride at 102 or some such percent for really short threshold intervals.
All that is to say. AI-FTP is more accurate for me than doing a Ramp.
2 Likes
I tried out the automatic ftp estimation. It gave me +3W.
However I was just about to do a workout so I thought I’d do that first before accepting (just in case it gave me a boost). I completed the workout, a good one too, and tried again. This time it said +1W.
I thought that was a bit weird, so I waited a day just in case it hadn’t yet accounted for the new workout yet. Now it says +0W 
1 Like
In practice it’s going to be based on the same tech though isn’t it - what the detection does is look at what you’ve been doing and make an estimate based on that, what prediction will do is the same but with the planned future work.
I guess a key difference with the prediction is trying to factor in an amount of non-TR work which may not show up in people’s forward TR calendar (unless they’ve scheduled recurring club rides etc but even then that’s just a TSS number). And will have less data to work with since apparently the FTP detection looks at “a range of data” likely including HR which you obviously won’t have for planned work (only the planned power trace).
In any case I have been checking mine pretty often too… and the variability has been in the noise, have seen figures from 279 to 282 which all feel about right.
1 Like
I was and still am really excited by this news. I have progressive resistance rollers and a “dumb”, wheel-on trainer. I use the rollers for Sweetspot, Tempo and Endurance workouts and the dumb trainer for Threshold, VO2 Max, etc. workouts. Regarding the Ramp Test, I’m not great at doing sustained big watt efforts on the rollers since I have trouble controlling the bike especially as fatigue sets in. That leaves me with the dumb trainer option. However, I can’t get enough friction to make it suitable hard enough to hit the upper end of the ramp test, so I must up the cadence (110+) in order to hit the wattage targets. So…I skip more ramp tests than I actually perform and attempt to estimate the FTP based upon RPE. This morning I had a ramp test scheduled and was really looking forward to see what AI FTP detection was going to suggest. It added +7 watts which I feel is exactly right. Thanks Trainerroad!
1 Like
Well, I can’t complain about the accuracy… I wasn’t convinced it would work, given I’m returning from serious injury so I was sceptical that the model would hold for me… Now I feel a little bit silly for subjecting myself to an unpleasant ride 
I gave it absolutely everything I had, possibly more so than ever before, so there was no “giving up at a preplanned point” or similar…
10 Likes
Dear TR: Please don’t take this as a criticism but rather just for referenced. FTP estimator came in at 257W … 6 days later I was scheduled for a ramp test and came in at 284W.
It should be clear that I am a coached athlete so all my information would have been imported from my Training Peaks prescribed workouts. My base training was sweet-spot in nature with Torque Intervals and Z2 rides. NO Vo2 max yet. By contrast the Strava FTP estimator has me at 241W which was even below my FTP of 245 going into the test. It had been a couple of months between ramp tests due to illness in early January when I was scheduled for a Ramp Test. I hope this information is useful to improve your FTP estimator.
1 Like
I moved your post under the official AI FTP Detection topic, for better visibility to TR reps.
In addition to that, you may consider sending it do them directly via the “Early Access” portal where you signed up for the beta tool.
1 Like
That’s a bit of an exception more than a general thing regarding an FTP you can hold from the FTPD. At least based on what TR has said.
They could detect that for everyone if they wanted. But the stated goal is setting workout better. Which is not necessarily the same thing.
I don’t understand your point.
I’m providing an affirmative anecdote regarding AI-FTP.
My point is your affirmation doesn’t seem completely aligned with what the detection is trying to do. So it’s great it works out that way in your case, but It’s by chance more than intention. At least based on what I interpret from your posts.
Yeah, sure. It’s not the new norm, and until I do another detection and potentially another Ramp I won’t know whether it’s a one off, or if it’s a repeatable improvement.
I’m explicitly talking about it being better for me though, so I think I’m ok with what you’re saying. Perhaps I’ve implied more than that by accident.
2 Likes
Sorry, you didn’t imply anything. Just me feeling the need to add commentary cause I enjoy talking about this stuff.
1 Like