Sorry if this has been covered a bunch already but had a couple quick questions to avoid having to go through this whole thread.
First, does it have to be a Ramp Test that is scheduled or can it be a 20 minute test?
And second, does the “Use FTP Detection” button only show up on the day of the scheduled Test?
Thanks
Not sure about the 20-minute test. But you can load one up for today and as long as you have selected the early access feature you can see. Take a couple of minutes to figure it out one way or the other.
The Use FTP Detection does currently only show up on the day you have the test scheduled - you can add a test to any day you want (ramp fur sure, not sure about 20-minute).
I don’t know if anyone else has encountered this, but I seem to have hit a glitch with the AI FTP detection and my progression levels. I used the AI FTP earlier this week, got an increase from 223 to 231 and got a drop in my progression levels, which I should have, but my future hard workouts look like they are based on my old progression levels and are all labelled as “Stretch”. I looked at the workouts, and I’m not too sure I can actually finish them without failure. I’ve already contacted the support team, but I just wanted to see if others are having the same problem and that people might want to be extra vigilant to make sure their workout levels get adjusted if their FTPs increase. IDK, maybe I’m overreacting and this is how the build phase is supposed to be (I’m in General Build 2).
I’ve had exactly the same problem, so what I did was use alternates for a more appropriate workout and it then adapts the subsequent weeks accordingly 
Until last year I didn’t even realise Strava offered an FTP estimation, it’s buried pretty deep in their web app. I kept an eye on it last season and found it curiously accurate while I was racing… but not so much over winter. It seems to need big supra-threshold efforts. I presume it’s more sophisticated than .95 x 20min but maybe I’m giving them too much credit 
Anyway, got the chance to use TR AI FTP yesterday, 311. Strava’s current FTP estimate? 311 
That’s a good suggestion. I was looking into the eye of the Spanish Needle, and I knew I was in deep trouble 
I forgot about that, and just had a look (if I am checking the right place).
TR AID FTP = 248w as of Monday (As posted above, this aligns VERY well with my current work and expectations.)
Strava eFTP = 223w right now
Considering I am just completing SSB2, with no real “maximal efforts”, I wonder what Strava is using?
Almost certainly either CP modeling from your MMP on Strava or .95xbest 20 min power.
I think I read they use a lot of stuff consistent with Skiba, so I would guess CP modeling. Either way, it is from data on your MMP curve. Which is almost always a conservative estimate.
Its not that, looking at my Strava and GC feeds anyway. GC says my 20mins max this year is just 235w and Strava estimates 232w. Whether it influences it 
That would indicate CP modeling then. Pretty easy to check - convert MMP from Strava to work for 2 or 3 efforts that represent your hardest during a given duration less than say 30 min. Slope of that line would be FTP (actually CP) estimate.
Sounds right to me based on watching my Strava FTP estimate last year. Well paced TT for a 40 min power PR? it wasn’t interested. Tactically questionable crit with plenty of wasteful efforts off the front? 
No criticism of Strava but I expect TR’s new tool is on another level. Looking forward to following along.
Strava only has the benefit of the MMP - what you actually do. It has no context of if you are doing those efforts as part of sweet spot efforts, VO2 work, tempo, etc. Trainer Road has the benefit of “labeled data” in this regard, which is always makes the machine learning task easier…
If you do a few truly capacitive efforts a CP model will give amazing predictive accuracy for what you can do right now against a wide range of durations (at least 2-30 min). Trainer road with AT is telling you how you should train. Either can be used for the same purposes, but really are two sides of the same coin versus directly comparable things.
Historically training protocols have been based on capacitive efforts which originate from research studies, and/or coaches/self-trained athletes who are really good at balancing the art with science. Trainer road has a ton of data to work from that neither aspect has. All of these approaches have strengths and weaknesses. But I like the TR approach for long term generalization of training strategies that are much more tailored to individuals than classical training methods.
No problem!
Right now, it has to be a Ramp Test that is scheduled. We are building this for 20-Minute and 8-Minutes tests, too, but those won’t be out for a little bit.
The Use FTP Detection button will show up on your Career page when the Ramp Test is your next scheduled workout (might be the day before it is scheduled if you have no workouts that day).
Hope this helps!
Some folks have been surprised that the AI FTP Detection gave them a higher FTP than they thought or got based on a test. If you’re someone who typically struggles with testing, this might be the case for you.
We particularly like the AI FTP Detection option, because it replaces the Ramp Test with a full, focused workout. Less stress, more training!
That said, it’s up to you! You can click on Use FTP Detection to view your detected FTP. At that point, you still have the option to either accept that new FTP or not. If you don’t accept it, you can still take the Ramp Test. Clicking the Use FTP Detection button doesn’t force you into accepting your detected FTP.
Hi Amber, hate to bug you, but the “Processing Data” mode bug has been ongoing for me since Wednesday morning. I’ve tried to reach support three times but the only response so far (to try it again) was sent over 42 hours ago. I saw there was an announced 15-minute pause to update a database last night and hoped that was a fix, but sadly not. Thanks!
Thank you @ambermalika.
Guess Monday was the last ramp test Ill ever take then. I never hated them, happy to skip them but it would depend on how hard the replacement workout is! 
I think its Goddard at the moment which doesnt look G0d thats Hard (well from the sofa anyway) 
That’s what mine was from the start, but then it adapted to Priceless. I can’t say that was easier on me than a ramp test.
After 5 days, my FTP estimate finally came through (-6 Watts) which sounds about right. First offered Goddard -4, but then adapted to Siretta, Sweet spot at 6.0, a Stretch, which raised PL by 2.4.
Can someone explain AI logic giving me a Stretch SS w/o, when I’m starting a Polarized training plan that should be no (or very little) tempo or SS. Should be mostly endurance and VO2 Max. @ambermalika Some feedback for the team.
One more anecdote for you. I have found the AI-FTP to be near spot on so far (haven’t gone deep in the well yet).
I don’t yet know if the AI-FTP is more accurate (my suspicion and hope), or whether it literally is predicting my Ramp Result and I either overtest on the Ramp vs expectation or am stronger. Previously when I trained at my Ramp Test FTP I’d invariably end up cooked by the time the Threshold sessions got long or over 100%.
Time will tell, but I’m super pleased. I’m also extra, super excited about some of the analysis and prediction features inbound @Nate_Pearson 