Yes, but that’s still not magically going to raise my threshold 10 watts where I could complete a workout that’s clearly an “over”
I’ve tested all of this for me and it feels like you’re arguing for the sake of it. Stop trying to be Coggan.
And the personal accusations have started so I’m also out - I’d say it was nice while it lasted but… 
1 Like
Those in the beta may have seen me whinge about this decoupling from FTP as a universal measuring stick - and I still am kind of baffled by the need to do this - but even as I type that I find it hard not to acknowledge that even in the best of testing environments a real world FTP test is perhaps only marginally better than AI if it’s even better at all. How did you sleep? What is the temperature of the room? If you were outside was there traffic on the road? Did that shitty conversation with your spouse ruin a few minutes of focus? And so forth. Every single testing technique has slack in the system and to pretend like it doesn’t is kind of silly. There is just no scenario where your FTP can be considered perfectly accurate to what your fitness actually IS. It’s a guide albeit an important one I suppose. Or to some at least.
Additionally - and this is the one place where I am 100% in agreement with turfing the whole concept of “FTP” entirely - is the difference between different kinds of fitness. Endurance, raw power, sustained threshhold, etc. These have different physiological effects on the body and different people have different strengths and weaknesses that can’t really be captured very well by 3 digits. My FTP has gone up quite a bit over the last 2 years - I think I started my TR relationship at about 230 and now I’m up around 300. I can certainly see that progression in my Sweet Spot efforts in particular - my legs seem to be able to handle a lot more sustained muscular load now - but I’m not as comfortable red-lining it as I used to be. So this is supposed to be measured by one single number? Its not possible.
All that said its nice to have a simple metric to gauge roughly where you are in the cycling fitness universe and I think - given the universality of “FTP” as a metric - I think it probably feels alienating to athletes for TR to dissociate from it. And I still can’t see why they would have to but I’m growing less and less attached to the old way of doing things.
Not only did I get sucked in here but I think I’m talking myself into the new world order of FTP-less training 
4 Likes
Just to chime in from a different perspective here, the previous AI FTP never worked well for me. I was kind of excited to try TR this fall after a really strong summer (best race results I had in many years), then when I signed up and got my AI FTP few days after my race, I was given an FTP that put that gravel race at an IF of .98 for over 4 hour, as well as a ride a few weeks earlier when I held 102%ish of that FTP comfortably for an hour at my sweet spot heart rate (I.e. it was a 60 minute SST workout). I checked the forums, was confidently told that was correct, so I immediately pulled the plug and tried Fascat’s app, which was … fine?
I just re-subbed to TR to try the new model and am playing around with things until it kicks in for me, but there are a few reasons I’m eager to try the new one:
- The model seems less complex if it’s primarily looking at relationships of power and heart rate over time; based on everything I’ve been reading over the past few years, the current model seems too sensitive to noise and jumps around for people too much in addition to being reliably inaccurate for me based on real world testing and performance.
- This is potentially a big one I haven’t really seen talked about, but if the new model only cares about power, then it should effectively uncouple VO2Max work from FTP in TR, which has historically been a problem for a lot of people. If it does work out this way, no-headache iLevels based on some form of PDC could be pretty handy. That said, I’d love to hear some thoughts from @Nate_Pearson and CO on this part of things.
It seems like the app should in general be more flexible and less locked in than the current plan, which is another problem I’ve had with TR historically.
1 Like
The new beta appears to be doing that. I’ve had it suggest a 1:45 SS ride when there was a 90 minute option but it appeared to be higher intensity.
1 Like
What plan are you following? That would determine the power that is being trained.
I always thought that you aren’t a 40 minute person or a 70 minute person. Your FTP power and your TTE are what is relevant and dependent on your training at a given time.
AI FTP - It’s an estimate of a measurement. The old system was an estimate of a measurement. It could be more accurate to the real world than the old one and lead to different values. It could both be more accurate to a kollie Moore test and give a better number for its use in picking workouts and still give some people wildly different values. I think in the couple of examples of large changes the individuals were not accepting changes because they wanted longer TIZ. That doesn’t tell us which number is correct.
I guess if you trusted TR before and have no data about how your numbers will change, why do you not trust them now?TTE change based on the type of training you are doing. That is true with a ramp test, the current AI FTP number, etc.
Yes it did that for me too, prior to aiftp detection. Then the massive ftp bump took care of that. So while the software can deal with it, it seems to prefer to get you back on schedule with intensity increase through ftp bump.
Seems there’s just one question. Do you want to get fitter or are you looking for an appendage to wave at others. If the former, then no issue. If the latter, then fine, take the hump.
No, the actual question is
“Do you want to get fitter utilising only TR to guide your training and as the main resource’? If so then the TRAiFTP is perfect for you.
If, however, you work with a coach, design your own training, or base your race/big day efforts off a previous figure calculated in a different way, then the TRAiFTP may not be the most appropriate number for you.
Nate has made it clear that the TR platform is not designed to work optimally with other coaches
For me, in my current situation, working with a coach, TR no longer has much of a place. That’s ok. When I am skint and have to return to self coaching, the new system looks absolutely amazing and it would be a strong contender if I want a computer to decide my training for me in future.
I do wonder whether this new FTP optimisation is partly to do with working with people’s anaerobic capabilities better. So for some the system has whacked up FTP thus any shorter efforts are harder, and it will instead then focus on increasing TiZ for threshold, sweetspot etc. The biggest problem with the previous FTP was indoor workouts for VO2 max being too easy unless you take it out of erg mode and ignore the prescribed power. Once I put all my training outside I realised how little I had been working rhe VO2 zone and above. However it takes knowledge and practice to be able to pitch the effort of, say, 5 hard efforts without blowing up. Not everyone has that, wants that or needs that. Some people literally just want to be told what to do, which is totally cool. Maybe the new TRAiFTP thing will help with that?
1 Like
For me, in this discussion, use of TrainerRoad appeared implicit.
Use of the TR software, and use of TR to drive all your training, are not one and the same
There are a lot of people here who use the software but coach themselves or use it alongside a real time coach, rather than relying on computer selected workouts.
2 Likes
I think TR could have their cake and eat it too if they wanted to.
They can set the “TRAIFTP” behind the scenes, they’ve already said that they look at wattages and such and then calculate / show that to the user. Replace it with a “scaling factor” that is just 100%. Lower it to effectively lower your TR Training FTP and focus on TiZ, TTE and higher progression levels. Increase it if you want higher power workouts at a lower PL or something similar. Show the user the PL Impact as they move the slider. There’s no reason they have to show an FTP that isn’t really FTP unless they’re using it for Marketing / Advertising purposes to make people feel better about themselves.
Then at the same time, have an FTP model that is their AI-based estimate of FTP that is useful / trustworthy outside of TR, as well as an AI-informed Power Distribution Curve. (TBD how far of their new FTP actually is though)
Add all those and you’re catering to a pretty wide group of their user base. Probably expanding their base.
5 Likes
Stuff could be done “behind the scenes”, with two different-meaning “FTP” numbers in play, but the existing product is engineered around a single number and presenting workouts to users in the form of %s of that single number (FTP), so to do as you suggest would involve re-engineering a significant chunk of the existing product (esp the UI) and re-educating users about a significantly changed approach.
Currently, the AI is doing its thing behind the scenes, and its output is effectively being back-translated into “cyclist-readable format” based around FTP and %s of FTP - this stuff would all need changing, and would probably mean generating individual one-off workouts, specified in absolute watts, for each workout undertaken, rather than mapping to and offering up existing workouts from the library. Some of this may well be where the TR product is headed but it’s a very sizeable undertaking. Who wants to wait a year or two?
I really don’t buy that closing line of yours, as it doesn’t pass detailed scrutiny. If you systematically inflated FTP numbers to feed egos and boost custom, then within just a few AIFTP cycles someone would end up with an ever-sillier number increasingly far from their true abilities. That sort of scam is getting rumbled within 6 months and then there goes your repeat business….
Or, we have a few existing very experienced TR users who’ve seen material steps up in their (new system) AIFTP numbers before they’ve even done any training guided by the AI… if TR had knowingly inflated these numbers to boost egos they wouldn’t be very smart, since those experienced users now just have a higher baseline from which to gauge any possible fitness improvements the AI helps them achieve. If TR were cunning and knowingly playing people, rather than inflate these users’ initial (new system) AIFTP numbers they’ve have understated them so that any improvements these guys achieve under the AI flatter TR!
I think rather than reaching for cynical or simplistic explanations of what’s going on, maybe there are other reasons that are a little more nuanced?
2 Likes
Agree. There are going to be larger changes when you are changing measurement systems. I wouldn’t expect that to always be the case once you are in a new system.
Don’t get me wrong, I don’t think there’s no work to it. But they’ve already done the first part of it by not needing FTP as part of their system (Per Nate they kept it because of it being a large bridge to cross). That part becomes mostly a UI challenge. I just think moving in this direction would make it a better overall product.
And I don’t know why people get bent out of shape when someone suggests companies do things to paint themselves in a positive light or add a little “spin” from time to time. I work with vendors every day in my day job, and literally 100% do this in some way, and TR certainly isn’t immune to it, especially when Nate’s said at least some users are going to see increases.
I’ll reserve my judgment until I have full access to the new model, but it’s always been dead on for me in the past and hope it doesn’t change.
2 Likes
I’d call it Workout Baseline Power. It’s a figure used to calculate how much power you need to exert for easy/moderate/tough workouts. If it’s set too high you fail workouts, if it’s set too low you don’t progress.
That would still have the problems that come with just using ftp to base workouts off of though. It seems like tr is moving past that with how they have talked about the update which is a huge improvement imo.
We updated it yesterday to model your fitness a bit differently so you’ll have a lower FTP than before. We ran this by the beta group and the the FTP felt right for most people.
We’re also going to predict your 20 and 60 minute power after this release settles down.
That should help you see the relationship between the three.
I said this in another thread, but we’ve had this debate every time we release a new test: 20-minute test, ramp test, AI FTP Detection, new AI FTP Detection. Each time, the previous test is great, and the new test is bad 
.
The AI is still stepping you incrementally forward when you get an FTP change.
It’s not like if you go from 300 to 320, you’re going from 2x20 @300 to 2x20 @320. For those who don’t like it, please report back the wattage jump for a similar interval length that you don’t like. I think that puts a clear picture of it.
Also, we’ve seen lots of people not think they can do their new ftp + PL combo, but then they can!
And finally, if you don’t like your FTP, you can lower it, and the AI will still pick similar workouts (probably longer intervals and/or less rest between intervals), and you’ll probably have to do longer workouts.
12 Likes
I have always agreed with TrainerRoad that the discussion “is TRs FTP the REAL FTP” is useless.
I am totally on board with “The FTP value which makes TR to select the right workout is the correct one”
That being said, today was the first time I declined a FTP adjustment (300→324).
I need a number which lets TR select the best workout AND which gives me a feeling how my fitness is improving. The last use case for FTP is not fulfilled any longer if you have such arbitrary changes!
this number clearly says nothing about my improvement. I did clearly not improve from 280 to 324 in one month
3 Likes