Have you reported this to TR Support? The best way to improve AI FTP is to report instances where its recommendations diverge from “reality”, so that TR engineering team can take a closer look and try and understand why things in your case didn’t work like they should have
It would be good to have a way to lookup / learn why there has not been a prompt for an FTP change.
always thought every 14 days was too quick. Even 28 days is a bit much. 42 days is probably best, every 6 weeks. It allows someone to complete a base phase.
On one hand extremely excited.
If one can use a reliable FTP detection without going through FTP tests on a regular basis one should be quite content .
But…and there’s a big but…dependability.
I got notified that AI detection recommends increasing my FTP to a number X, which is 6W higher than my previous FTP AI detection.
So…I did an outside FTP 20 minutes test with the said goal of X x 1.06.
Well…I held it for 5 minutes before starting to bleed watts till I was trashed by 15 minutes with no juice and a result that it neither consistent (a 20 minutes steady slope decline) nor close to my previous FTP.
It’s as if AI FTP detection is 1.06 higher than my actual FTP,
Sorry, are you saying you tried to hold FTP + 6 watts, or FTP * 1.06 watts (i.e. 106% FTP)?
- This is my bet. Presumably he did that to account for the 0.95 multiplier used to de-rate a 20-minute FTP test block to estimate the FTP. Essentially, he reversed the math to set a target and added 1% for extra measure.
As we’ve covered in many topics here, that theory is fine but flawed and should not be taken as a blanket validation or disproof of the accuracy of AIFTPD. Many factors are present and it’s not necessarily worth diving into that in this specific topic. It’s better in the main AIFTPD topic or the others where people have attempted test comparisons like this.
As the Other Chad wrote… *
You take the average power from the 20minutes interval and multiply it by 0.95 to adjust for an approximation of a one hour effort.
So, if the AI detection was correct, with everything else taken care of and constant, then adding the 5% from the top (which is approx 106% from under) gives you your target power for the interval.
If AI FTP detection gives me a benchmark figure that means that the workouts that I get for the first bit of the upcoming block of training are in the right ballpark …
… and then Adaptive Training brings me nicely through the progression levels …
… and I’m nailing my workouts …
… and my fitness builds …
… and I’m getting faster …
… why would I care so much about the absolute value of that benchmark figure?
Some people leverage FTP in ways other than the mostly training focused aspects TR intends with the current and future implementation. Things like pacing plans and such for various events, nutrition planning and other reasons are all considerations that leverage FTP as part of the calculations. As such, it can be beneficial to have “an accurate FTP” on hand for those that use them these ways.
Nate admits to the confusion overall (above and elsewhere in our discussions), and has floated the idea of TR renaming the “magic number” they give us, in order decouple it from the FTP confusion that we see frequently here. We have to remember that there are usually more than one way to look at and use values in this area, and FTP is one of those with plenty of uses beyond setting workout power targets.
I do agree with Helvellyn’s assertion that the absolute number does not matter if strictly looking at it for TR training purposes. However, I am one of the folks that Chad references as having ulterior motives for having a historically consistent FTP number to benchmark. Apart from all of the good examples Chad sited, having a historical measure of FTP that I find consistent with itself is of use for me as I track my general fitness. Thus, I do find useful to intermittently do a 20 minute or ramp test just to see where my TR Detected number sits vs how I used to benchmark it in the past.
Fair points, Chad.
Of course, even the 20 minute/95% protocol is only an approximation, rather than something that Moses carried down from Mt Sinai.
We’re all just feeling in the dark towards some sort of fundamental truth, I guess. I do wonder how much variability “being human” introduces to the equation though. I imagine all our FTPs (in practical, what can I do today terms) are “plus or minus X% depending on where our mind and body are today”
I hear ya. I suspect I’d be the same if I had a longer training history.
- Sure, just like the other half-dozen methods for estimating it as well. None are perfect or absolute in any way. I put no personal stock in one over another. I only mentioned the 20-minute aspect in direct response to the one other person commenting on it with respect to their results.
Yup, and despite that, it does not necessarily invalidate if from use or consideration. It’s just one of the many variables that people should keep in mind overall when looking at their training and progression.
It’s true that comparing method X to method Y and/or Z is a nearly futile effort. We’ve seen it countless times in the topics here over the years. It’s worthwhile to point out the potential for misalignment between these things, but we can still consider using comparisons like this to better inform our wider picture of our fitness and progression. It just requires some asterisks along the way
My apologies if this was mentioned but I missed it. Will auto ai ftp detection allow you to accept or reject the “prompts” or “smart recommendations”?
I have not seen it stated directly, but I am 99.9% sure that you will have the same option to “accept or decline” the AIFTPD offering each time it is given.
Essentially identical to what we have now, but it will be triggered by other methods than the simple Ramp Test association at present.
Cool. That’s what I figured since Nate used the terms “prompt“ and “recommendation“ in his post
I think this is the key risk to FTP changes during Specialty/race season. Obviously, as you mentioned in another thread, sometimes you do genuinely have an increase in fitness and need to pace/set expectations off a higher benchmark, but it requires a bit of experience to know the difference.
I also vote for a name change to “Functional Training Power” (yes, I know it wouldn’t work but I like it)
It’s a bit of a tangent, but I would STRONGLY advise any new name for this number to be something with initials very different from “FTP”. I get the idea of swapping a word and keeping the digits, but it will likely lead to endless confusion.
We’ve touched on this in other topics, but something truly different like “TrainerRoad Power” (TRP) or anything else along those lines might make the most sense. They should “own” the value and make it clear it is not necessarily meant to be the same as FTP in the other uses we see.
I know . I agree, it would cause endless confusion as it wouldn’t be able to be used as an acronym. It was tongue in cheek, and honestly I think it’s an accurate way to describe it.
You’re right a new name is needed. Back to the drawing board.
Two nanoseconds after TRP comes out there will be an equation published to convert TRP to FTP.
Just like 4mMol lactate, we be stuck with FTP 4-eva