FTP is by Coggan’s definition an estimate of power at MLSS obtained in a field test. (I can look up the paper and quote him directly if you insist.) That is, you need a field test protocol to determine it and from an exercise physiology standpoint a priori you cannot directly compare numbers obtained with different test protocols.
When it became broadly used in practice, people initially identified it with the result of the 20-minute test minus 5-10 % as that was the method suggested by Coggan and popularized by Coggan & Allan.
You will get whatever workout that is most likely to lead to performance gains based on data analysis of power files and progressive performance improvement. It’s almost as though FTP is used to describe the work, not prescribe it…
And analysis of power files was low on the confidence scale. It’s arguable that advances in big data analytics and machine learning could upend that scale, but we may need to wait for proof.
The only question I have is, where on the power-duration curve is the AIFTP value coming from? That should be easily answered by the model and will help in all these confusing discussions.
Nate gave some people their 20 and 60min power in the beta thread. And the AIFTP value is somewhere in between. I guess more towards the 20min.
In the past >40min TTE (in a blood hard test) gave me the confidence that I‘m at FTP (Coggan definition).
Still I‘m happy with the workout selection and the model does its job. But for pacing my races this season I will not use the AI value (rather 10-15w lower).
Very interesting discussion. Im just updating the garmin to the AI FTP number So zones are right when outdoors. Ps Loving the AI version !!! I have had increases and then a 1% decrease. Its superb
For me it says I have 42% chance of failure. I’m pretty sure it’s closer to 100% chance of failure, given that it wants me to hold my 18 min PB for 60 min..
Mine is 58% chance of failure…and that’s AFTER I dropped my FTP 10 watts. Fun to see though.
In all seriousness, now I want to go find a workout that’s something like 40 mins at 100% and play with the number until I get to an FTP that gives me 51% chance to succeed at Max Effort or lower just to know what it thinks.
I sometimes wonder how accurate mine is since when I do Zwift races, my NP sometimes isn’t nearly as high as my FTP. I just did two iTT races in two weeks and exceeded the numbers. AI FTP is set at 291W. The last workout of my build block, I skipped my workout and was planning on doing a longer Zwift Epic race but lost motivation to continue at the pace in the race (19min with .98IF). Later in the same day I decided to try the weekly Zwift TT Club race, and held 304w average and 309w NP for 22min. Then i had recovery week which I followed until I did the next week’s race on Saturday, but this one was about twice as long. I held an average of 295w and 296w NP for 43min. Now my AI FTP prediction is saying 294w. I’d say it’s pretty damn accurate.
I 100% understand that and the whole concept that TR uses. I understand that Ai doesn’t care about ftp, only the watts/duration. Like I said I’ve been given great workouts (save one, which I can write off to overheating and being underslept) since the new rollout.
With that said, that doesn’t change the fact the number that TR says is my FTP is too high, and this causes all zones to be pushed up and creates questions/uncertainty about ctl/atl/pacing/tss….etc. It’s 20 watts higher than my all time TR best, and I’m 40 and hanging on to fitness for dear life!
There’s no reason the AI can’t give me the same/similar workouts in terms of time/power, but give an FTP that’s in line with the historical standard of an hour power +/-, or 95% of I would produce in a 20 minute test….etc. If it does that, then all my zones aren’t pushed up 3-5% and I can keep doing the quality threshold/vo2 work without my endurance rides going into tempo and sweet spot creeping into threshold.
This is all I want to know too! If TR sees something in the data that shows the best fitness increases come from suprathresh, no longer need SS (but sub for classical threshold), and pushing z2 workouts into tempo, then I’m fine with that too. Just let me know so stop second guessing the AIFTP.
This is what I’ve been asking for - it must know approx. how long I should be able to hold the AIFTP value in order to prescribe workouts AND give me the predicted effort of the workouts.
My AIFTP is 301: The max I think I could hold this for is around 12 m. This would be me falling off the bike.
I have my FTP set manually at 290: The max I think I could hold this would likely be 30-35m. This value seems to be working well for me for TR power levels that keep me in SS level 8 and thresh level 5.
Unless you’ve posted evidence of your MLSS upthread, I don’t see how you can base your position on this assertion and red herring, I really don’t.
Unless you’re an expro timetriallist, I suppose. In which case, fair dinkums.
…Your zones don’t matter anymore, not in the prescriptive sense for sure and maybe not at all. I’m not sure I can land that message any gentler. Most all zone systems are an artifact of ftp based training.
We’ll keep them as a way for humans to describe what they need to do, but really what you need is X Watts for Y repeats with Z intervals.
Why would FTP be a point on the power duration curve?
Your question makes me think your understanding of FTP is a maximal effort for a given duration, which of course it is not.
I did two TTS recently as part of two stages races. One of them was 30 minutes and the other around 20 minutes. The AI FTP was off by only 6 watts so pretty accurate to a real world max effort. As TR has noted, that small difference probably wouldn’t change the workouts I get from them or the watts I’m supposed to hit in any of the workouts (be they interval or more base workouts). My real world FTP was higher by those 6 watts.
“The 20 minute test is Hunter Allen’s baby. You should ask him how hard to go during the preceding 5-minute blow-out interval.” ~Doppelganger Coggan on Reddit.
I’d like to think my 10+ years of interval training and 10k+ miles on the bike every year for a decade count for something. I think if you ride enough and you do enough intervals and you fail enough intervals and you do enough races and you get dropped enough, and you go for enough KOMs and FKTS, and you have to lay down on the ground after intervals enough, you get to a point where you just know what you can and can’t do and you know the feeling in your legs and lungs of what is sustainable and what is not.
Do you think I can do 428 for 40+ mins at 6000ft because TR AI thinks I can? At the end of base season, nonetheless. I don’t. I’ll try it in the next few weeks when my legs feel good and schedule allows. I’d love to be able to and prove myself wrong. I’m a sucker for punishment and will load that 1hr power workout and see how long I last. With an ice vest and sodium bicarb and caffeine.
FWIW, I’m not the only one bringing this up. There’s obviously people here in the same boat who’s AI ftp is around/over their all-time 10 minute power. There’s threads on reddit of people saying the same thing. Jesse Coyle did a youtube on it and iirc his AI ftp was 10 watts higher than his all-time best 60m power. It seems like it’s a theme. Are we all just mentally weak and don’t really know our bodies and MLSS well enough?
Then the TR team need to all get on the same page! Because all I’ve heard on their podcast since it’s inception is that your easy days should be easy so that your hard days can be hard. If you go too hard on your easy days, then it makes you tired and worn down and you won’t hit your numbers on your hard days and it will stunt your training and you’ll plateau and it’ll wear you down. Meanwhile, their AI is assigning me 280-310 watts for 1-4hrs every chance it gets, which is generally 0.67+ IF (which is based on a FTP that I am certain is too high). Are we not doing easy z2 days anymore because the AI overlord says so?
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.
So zones are right when outdoors. Ps Loving the AI version !!! I have had increases and then a 1% decrease. Its superb
