I must admit that it was so close just on the day I wrote my reply. Today we are already further apart again. Tonight is a FTP test on the program, then TR and Intervals.icu show the same again anyway.
I don’t think TR is going to be estimating FTP based purely on max efforts in any case.
What I was getting from Nate’s comments was that it is going to be analysis-based, using your personal training history and the FTP gains their ML model expects to see from that. I.e., predictive rather than analytical.
It is basically power output @ MLSS (maximal lactate steady state) and more convenient for people to test / estimate than to determine MLSS via ramp test with blood lactate measuring devices and then estimate / calculate FTP from the blood lactate / power curve recorded during the test.
Assume they haven’t announced a timeline for this?
The question is, would an estimate be more accurate than a Ramp Test?
I don’t mind doing the Ramp Test but I’d like to be getting an accurate result so my training levels/workouts are set as best they can be.
Not really. They rarely give hard timelines for anything. This is what I pulled from the related cast:
Yes, no and maybe. Testing is what testing is, and that means potential variability for all the reasons we know.
FTP estimates based on max power efforts are applied in a number of apps (intervals.icu is the one I use, but many others work similarly) and those have a been seen to be relatively “accurate” by a number of users. But these typically need “max” efforts and depending on recent training, the estimates may be misleading to as good as dead on.
TR says they plan to show an estimate at the time the FTP test is offered, and leave it to each user to choose what to do. For those willing to test for any reason, they will get an instant opportunity for a comparison. I predict we will see the same juxtaposition we get with my “Ramp says this…, but I got that result… what gives?” comments wide and far once this hits the user level. The same analysis and more will be appropriate, but overlooked or oversimplified as we seem to do too often
This would be great in a world where the only cycling people do and the only place they use that FTP number is in TR. But when a large number of people are using TR to train for outdoor races where they are using a power meter that outputs a real wattage number they need to be training to that same wattage reading.
See but I don’t think that’s actually what most people want. Of course if you asked them, they’d say they do but I think a ton of people would stop training as much if they didn’t have that rising FTP number to hang their hat on. For a lot of people that is the only measure they know to use to indicate to themselves that they are getting fitter and without it they’d be lost and quickly lose motivation.
This I 100% agree with. Sure pacing is an issue on those longer tests but after 1 or 2 of them you kinda get the hang of it. And I hate the BS line of “in order to pace well you need to know the target ahead of time. And if you know that then why test at all?” It’s just a silly excuse. If that was the case then why does anyone do TTs. You could just estimate your CdA and use best bike split to get your time with your target pacing. And personally, I would rather do a 20+minute test where it is mildly painful for most of the time then do another ramp test where I finish with shaking legs, metallic taste in my mouth, pin-hole tunnel vision, etc. They’re awful.
N=1 and largely worthless, but I found this interesting.
Did my scheduled Ramp Test last night. Ended with 240w FTP per TR. This is based upon the 0.75 times best 1-minute power (and a cross check against the 5-min power for potential issues).
Then I got notified via email of an increase of my eFTP from Intervals.icu… of my new 240w eFTP based upon a 6min 30sec effort.
Generally speaking, I have watched the ICU eFTP casually and found it to pretty closely follow what I think my FTP was (specifically after I bumped the minimum time up from the 3 minute default to around 5-6 mins for me).
No grand conclusions, but it parallels what seems to be a potential aim for TR. It’s interesting this happened in the same test despite different times referenced. That and the general trend of the past give me some faith in the potential for future estimates replacing testing in at least some instances for me.
100% this. It is a functional (ie not lab based or using lactate measurement) description of power@MLSS.
My understanding was that they wouldn’t be using max efforts to estimate FTP increases, I think that is where the skepticism lies with most of us. I think there is generally agreement that you can get FTP from max efforts of various times using different models, whether you can extrapolate out from submaximal efforts is the question.
Good point. I haven’t dissected the discussion, and from the little I have, it sounds like we really don’t have much depth of info at this time.
No idea if it’s possible, but if they could provide examples (or direct access to each user) of personal estimates related to their recent past training, that could go a long way to seeing “how it works” at least from a surface level.
Not to be too pedantic, but given that there is no agreed upon definition of what FTP means, or how to calculate it, what does “accuracy” even mean?
Look at just TR: There are 3 different ways to get a FTP value - 8 minute test, 20 minute test, ramp test. And it is widely known that all three will given different values.
In the ~18 months I have been using intervals.icu, I have had 8 new estimated FTPs, and every single one of them came from a ramp test (and correlated very closely to the ramp test result), so I don’t think you are n=1.
Both techniques rely on the same assumption that a synthetic power-duration curve, anchored by one maximal power/duration point, gives you a “pretty good” estimate of a rider’s abilities at any other duration.
- it is still based off a maximal effort. As previously discussed, ramp tests are the only maximal efforts in a TR program.
- “Pretty good” is not good enough, hence progression levels etc.
The definition is clear, but some people are uncomfortable with it. Spend some time training for 10 mile and 40km TTs, something few do anymore, and you’ll get comfortable with the definition.
I’m being serious when I say that there is NO definition agreed upon for FTP. The power you can hold for 40 - 70 minutes? A certain number of mmol? Show me where everyone agrees / uses the same definition for FTP? Forget different methods of estimating it - we don’t have an agreed upon meaning for what we are even trying to estimate.
It’s still useful to have a reasonable idea of what your vaguely threshold zone is, even if it’s only a ballpark. See all the posts where people struggle with 20 min “threshold” intervals.
I’m not arguing that having some type of number to set training goals isn’t useful. Just that the idea of trying to determine which method is the most “accurate” is a fools errand, as there isn’t a “gold standard” that you can compare the different estimation methods against to determine which is most accurate.
Coggan’s definition is clear, he came up with the term and its his definition. Coggan gave examples of different ways to estimate it. And that’s where the problem begins - you either a) read about FTP secondhand and come to believe it requires some specific test protocol that must be followed exactly, or b) you accept the definition is inherently fuzzy, the estimate is subject to multiple sources of error, and there are many ways to estimate and correlate one result to results from other estimation methods.
Those who’ve done ‘more exact’ tests will tell you the human body exhibits time varying behavior. I’m of the ‘just go with it’ camp, accept its an estimate with a margin of error, and treat an FTP estimate as a range not a single number.
We are in complete agreement on this. This was my point: talking about more “accurate” FTP estimation methods doesn’t make sense, as there is no gold standard for getting a “true” FTP for anyone. So you cannot compare different estimation methods and say which is inherently more “accurate”.