In that document he also discusses other ways to estimate power at MLSS, including critical power.
Going by what he wrote (including one of his posts on that subject here), he did agree that maximal 20-minute power minus 5 % is a viable way to estimate MLSS power — on average.
Independently of what Coggan thought or thinks, for many, many years a 20-minute test was the de factor way to determine FTP in practice. Just like with any test, with time athletes gained experience and then simply corrected the numbers accordingly to get an estimate for their power at MLSS from a test result.
Importantly, TR’s AI FTP no longer wants to approximate power at MLSS, but be able to offer you the right workout. That’s subtly different.
How long you can do power at MLSS depends on you and your training. That’s always been an issue with any FTP test, any simple computation based on statistical averages will fail people as most of us are not average in that respect.
Machine Learning changes things here, if done properly, it goes beyond simple statistical relationships and instead bases its predictions off of the past performance of an athlete. That’s why I wrote AI FTP has the potential of being a much better way to gauge FTP than any test protocol.
Do they mean that you, and more importantly everyone, can intuit your/their FTP on any given day? Accurately? No, I won’t accept that without evidence.
I’m 13yrs in training, when I’m fit I’d expect my ftp to be around 250-260. But that could easily be ten watts more or less. And I’ve used the same meters for seven years, unlike most.
I don’t see the connection between this and zoning systems that you do. If a prescribed day (series of days/training plan) at X watts makes you faster, then it makes you faster. Whether that can be translated into OldSpeak or inherently vague terms like “Zone 2” really doesn’t matter.
Your underlying question of whether it really does make you faster is valid.
It also forgets one important fact: at least if you stay within the TR ecosystem, since AI FTP works with absolute power (as opposed to fractions of your “FTP”), the system should prescribe the right workouts for you.
TR’s database is huge, much better than anything researchers at a university can come up with. So TR’s solution have the potential to uncover and exploit relationships that are beyond reach for coaches and exercise physiologists.
That could lead to some breaks with orthodoxy and “established truths”. Hence, concepts like zones need not coincide with the definitions Coggan and the early contributors to power-based training came up with.
Overall, what I would look at is whether TR AI represents an improvement, i. e. whether it works better.
The funny thing is that the estimated 20/60min values coming from TR (only a few data points available) are ~105% of AIFTP for 20min and ~95% for 60min. There were some outliers but that’s the ballpark. Which means that they didn‘t reinvent the wheel in this case.
That’s true, but my comment was meant to be about 3% off of a “true” FTP is pretty close. Especially with all the caveats around if that was an actual max effort (pacing, nutrition, previous days’ workouts, cooling, motiviation, etc.)
I would say my evidence is specifically the past 7 years of training on TR. Dating back to the OG SSBHV plans and the ramp test protocols and through the og AI FTP detection my FTP has consistently come out of the offseason in October/November around 370 or so and increased to 400-408 by February/March. Typically I’m doing level 7-8 sweet spot workouts and 4-5 threshold workouts at that FTP. Without going down to sea level there’s been no evidence I can do more than 405-410 for 30+ mins at any point in my life. In March/April I either hit my genetic ceiling, I get sick and lose 10-15 watts of fitness and then race with what I can get back….and the temps warm and I struggle with cooling and hitting those same numbers when it’s 80+ degrees, or I use that fitness on ultra endurance stuff and tend to lose the interval consistency that got me to that point, or some combo of all that.
This year I’ve stuck to the same script and followed my plan fairly well and been as consistent as past years, and TRs telling me at 40 years old I’m 20 watts better than my all-time bests at the same time in the year?? Their reasoning is that the “scale* is now more accurate. So that means it was 20 watts off before? I was undertraining that whole time, despite running a 130 CTL? I underperformed at every ramp test? Their AI ftp detection was always off? I should’ve been doing Tallac +3 at 390 instead of 370? Sorry if I don’t buy it without any evidence except “trust the AI, it has lots of data”. My evidence says my all time best up my local climb is 435 for 12 minutes, and now being way older, I’m supposed to believe without evidence I can do 428 for an amount of time that would justify it being called an FTP?
44 year old me beat 35 year old me by about 15 watts. 45 year old me halfassed the year and still beat 35 year old me, did an actual hour TT at my 35 year old FTP. I’ve only been 46 for a few days but if I actually do what I plan I’ll be beating all of them…
I know that isn’t your entire point but you seem to be thinking that you are older and can’t possibly beat your younger self. Unless you were training the same and as diligently as you are now then it is very possible you will. I trained harder at swimming in my 30s than I did in HS and was approaching my HS times in the pool because I was better at training despite being 15 years older.
as you said above you’d like your 10 years of training to count for something, it is, you have likely figured out how to execute better despite being older and can in theory actually perform better than 10 year younger you still figuring stuff out.
This isn’t to say that the predicted ftp or your aiftp are right or wrong just that age isn’t necessarily a bad thing in this case.
No, what I’m saying is there’s little evidence for that. I’d love to be fitter and maybe I am a little but, but not enough to justify 428, as I keep saying. If I was doing the same workouts now at 20+ watts higher, then that would be evidence. And maybe I should go replicate a week of what I did years ago at higher wattage and see what happens.
To me those look like someone with very similar fitness. At least within the margin of power meter/kickr core error . It’s similar workout levels, similar RPE, similar power/np. To TR AI that’s a 4% difference. I don’t think either one is capable of pushing 420+ for a time duration that would justify calling it an FTP.
As a former triathlete that’d consistently get his ass kicked by much older men and women at masters swim, I know firsthand that swimming is as much technique as it is fitness. Not saying you are, but you can be less fit and be faster by moving through the water better with better technique. On a bike, watts are watts are watts.
No, but it shouldn’t call it Anaerobic if it’s VO2. Or call it VO2 if it’s Threshold. Or Call it Endurance if it’s Tempo. Or call it an Over-Under if it’s Supra-Threshold.
This isn’t the way it’s working for everyone, but it is the way it’s working for some people. And there’s a time and a place for each zone.
I did used Critical Power testing the beginning of December, and a Zwift 20 minute test and both provided the same FTP number. I trained through January and then restarted my TR subscription and the AI gave me an increase of one whole watt. Following the plan set by TR AI I would say it was accurate, and the increase I got a week ago seems to bear that out.
So far the start to this training plan has felt more challenging yet more successful than the past. It will not be perfect, but it feels like the company is very much on the right track with this.
It depends, their ai engine might be boosting “ftp” while simultaneously killing your tte. The fastest way to improve fitness doesn’t necessarily translate in to race day performance. Tr likes to push sweet spot work in to threshold, which will end up with a different physiological response.
I’ve actuality been using Gemini to help me build and interpret my previous fitness gains and race results.
I haven’t verified ftp for a bit since I’m still working back in to consistency after having another kid at 45, but the ai ftp is usually within 5 watts of my manually set ftp so I never accept the updated number. It seems silly to update the number in the middle of a block.
I can’t accept this TTE didn’t exist in isolation like that, FTP/MLSS is a good enough proxy for cycling fitness.
I think that’s just pandering to TSS trackers , there is surely no justification to ignoring fitness gains during a training block - whatever you perceive a block to be.
Not that FTP really matters in this sense, for this paradigm!
. It’s similar workout levels, similar RPE, similar power/np. To TR AI that’s a 4% difference. I don’t think either one is capable of pushing 420+ for a time duration that would justify calling it an FTP.
, there is surely no justification to ignoring fitness gains during a training block - whatever you perceive a block to be.