When I look at my threshold workout slated for next week it says the threshold sections are at 95% of the 314 AI ftp number (which is 298, a number I’ve basically never done straight threshold intervals at). It seems the ai ftp number is still somewhat relevant in the numbers I’m supposed to be doing in these upcoming workouts. They’re all percentages of the 314 AI ftp.
I went through a similar journey, and have settled on #2 lowering my FTP. I first tried option #1, but the workouts were pushing way too close to my limits and were ‘out of zone’ for the intended energy systems of the workout: The unders in o/u, were not under. The SS were mainly threshold and z2 become tempo on many of the prescribed workouts.
If I were you, I would trust option 1 and see how it goes, but rate the survey’s as accurate as possible and don’t care if the predicted FTP goes down. Just let it happen. If you find after a block things really aren’t feeling right, then you can move to option 2.
You can also go to the ‘Training Approach’ and edit individual zones and lower your endurance rides to make them easier instead of pushing into tempo - this will at least give you a better chance of hitting your hard days.
From my own experience manually lowering my FTP: The workouts are still as hard as they should be, and are now in the right zones - the RPE is similar to the previously higher AIFTP, but now I’m now getting longer TIZ. The feels right to me. I’m getting fitter and progressing and retaining all the TR features except the prediction value (It still check it manually by accepting the FTP). I’m now doing around level 5+ threshold and 7+ SS for reference.
1 Like
I thought so too. Now Im not so sure.
Not based on my experience - they are still the right RPE, but just higher level workouts that are in the intended zones and longer TIZ.
1 Like
“Did the first 28 days of sweet spot and over/unders and was able to do all the workouts.”
Didn’t you say that you were doing SS and threshhold workouts with your ftp at 299? So why would doing threshold workouts at 95% of 314 be an issue?
At least try the workout and see if you fail. If you fail, it will adjust. I’m not really sure what the point of this whole thread is.
1 Like
Nate has said on the podcast (and in those links above) that the model is agnostic to your FTP.
The model looks and says “I think you can do threshold intervals at 298 watts”, and then displays it to you in terms of the FTP to make it more understandable. If that’s making it more confusing for you I’d just ignore the FTP percentages and look at the watts. If you’re worried about threshold intervals at 298 - compare it to the power records for that time frame - do they look reasonable? Much too high?
1 Like
I’m in complete agreement with you. OP can actually test this themself as well. Take a note of the actual wattages prescribed for a few upcoming workouts. Go and manually change your FTP +/-20 watts and look at the new workouts AI prescribes. Are the wattages more or less the same?
2 Likes
Yeah but at some point the AI has to pick a workout, the AI is not creating workouts. If the training plan calls for over/unders the AI has to pick an over/under workout which are based off of FTP. So whether the AI understands FTP or not, it’s only choices are to pick workouts based on FTP which are typically some version of 95/105%. So, if the AIFTP is high, the over/under workout by default will be high unless it’s picking from progression levels in the 1.0-2.0 range, which it won’t because AIFTP is based around 3.0. If over/unders are part of the training plan then the only solution appears to be to lower your FTP.
This really becomes an issue for over/under workouts in particular. If it were just threshold workouts it matters less, because doing threshold at 97% or 101% are likely both doable as long as the interval duration or time in zone for the workout isn’t too long.
2 Likes
My understanding is that you have this backwards. AI picks a wattage target and then runs through the workout catalogue to find a workout that fits those targets. Wattage targets come first then the workout is a result of whatever wattage AI prescribes. FTP is not directly considered when picking the workout, it’s a translation back to % of FTP after the workout is picked.
I think this is a good point. It seems like despite the AI FTP the workout selection is still somewhat 1-dimensional. There is a fixed library with a scaling factor that it chooses from. It might be nice if the scaling factor was more rigid (eg FTP from ramp test), but the AI would recommend more custom workouts or scaling per workout type. For example, maybe endurance workouts stay the same but it could recommend +/- X% for thresholds or +/- Y% for over unders.
However, I wonder if you could just manually lower intensity on over under workouts if that’s what you want. Unless you want everything lower in which case you could lower your FTP.
1 Like
It doesn’t matter, the workout catalog is based off of FTP. So the AI may not know/care about FTP, but the AI’s only options are workouts created around FTP. The AI may “want” to give the user an o/u workout with wattages that equal 93%/103% but they don’t exist or don’t fit the workout intended for that day and/or training plan.
1 Like
I think the workout catalog is large enough that even for something like over-unders there’s a big range to choose from. In this case of this thread - the original FTP debate was like +/- 5% I think. So if it’s too high they can do Kid (3x9 minutes, 2 @ 90% and 1 @ 105%) or if it’s too low they can do Sill (3x9, 1 @ 95% and 2 @ 105%) and that’s a pretty huge range - and those workouts are only 2.4 to 3.8 in PLs.
Maybe this situation of not enough workouts gets wonky if the FTP is off by like 50% - but in a case like this where we’re int he range of +/- 10% it should be no issue at all I think.
I only chose those two workouts to make the 3x9 structure match - there’s an even wider range if you want between the low end and high end (for simplicity’s sake I’m looking at 1 hour over-unders):
Low-end: Starr-1: 4x6, 2 @ 90% and 1 @ 105%
High-end: Spickard: 3x15, 2 @ 98% and 3 @ 118%
With a range like this - your set FTP likely has no input.
All o/u workouts in Base are a minimum of 1:30 - 2:00 long. Even in your 1:00 scenario, if my FTP is 5% too high, the workouts would be 95%/110%. That’s pretty brutal in Base 1. Then, how do I progress week after week and into Base 2 and 3? The workouts don’t exist because over/under type workouts were never intended to be fractional percentages. However, AI would need fractional percentages to adjust proper wattages.
1 Like
It could, but the AI would then have to ignore that the training plan is calling for Threshold. Even still, I don’t think you’re finding enough workouts to allow proper progression for everyone in every fractional percentage needed. Not to mention the IceFall-6 with a 5% over FTP estimation is 85%/95% or 100%. Which you’re not getting “over” anything. These workouts were likely created as intro type workouts into true over/unders, which is why they’re not proper progressions.
Napeequa is 2 hours, 4x6 (about as low as it gets) with 2 minutes at 90% and 1 minute at 105%. Completely doable with an FTP that’s 5% too high - even if you’re right that the system is constrained in this way.
Look at how huge the 2 hour Threshold over-under catalog is: 4x6 (88% / 102% for u/o) to some ungodly hard 5x15s (90/105). Even if you think the FTP is well off AND the system is somehow limited by this the range still seems more than sufficient.
1 Like