Ok but you also need to make sure you are doing level 4 and above workouts 
The problem before was that it’s sounds like you were stuck doing level 1 and 2 workouts which are a waste of time.
8 Likes
Ramp tests overestimate me by almost that much. Kolie Moore did a podcast where they discussed a study that showed the average ramp test overestimate was 7%. My guess is that AI FTP is significantly more accurate than the ramp for you.
5 Likes
I look forward to the next 6 months of training and will follow up with my observations.
Cheers
Do you expect to see a different outcome than the thread you started a couple days ago here?
Unless I am mistaken you will be doing the same thing as before which I would generally expect to cause the same issues no?
8 Likes
Why do you hunk the ramp test is more accurate?
I find AIFTP is about right most of the times when training. During the CX season I see that it gets thrown off when racing I think because of all the on/off power from the CX races and then my RPE.
It must give it bad data.
If I’m in the mood to verify my FTP the old school 20 minute test followed by the workout Marion a couple of days later. The workout should feel hard.
The workout in itself is some good feedback for the system.
Agreed. This sounds like exactly what happened last time.
It’s odd, out of all of the suggestions that will work, why are you opting for the one you’ve already tried….and shown it doesn’t work?
Joe
7 Likes
I’m at a complete loss as to why you would come to this conclusion.
Let me start by saying I’m not one to trust AI, so I have historically done the Kolie Moore test. I am one of those people who WAY over tests with a Ramp Test resulting in an FTP that is far too high.
Having said that, when I’m using TR, I have learned to let TR’s AIFTP and the associated PL’s drive my training. TR made a great point in your other thread by explaining how setting your FTP higher than their AI thinks you should be results in the system giving you EASIER workouts. That’s the opposite of “higher effort training demands” that would lead to you feeling “like I’m better equipped to compete”.
As a plan of action, you came here and raised the issue. TR and a bunch of other users responded by explaining why what you’re doing won’t lead to TR helping you get the results you seek. They also gave you clear explanation of how to fix that. Perfect!
So…how could you possibly read all of that and end up deciding to change absolutely nothing? You’re letting your ego get in the way of improvement. It’s a common trap.
17 Likes
Perfectly fair. You should do what motivates you to train more. Whatever you might think of TR, I think one thing they do really well is giving you an FTP target, that if then use their PLs with, you’ll find the right difficulty pretty quickly.
1 Like
@OliverTwist, if my reading comprehension is functional, I understand that you took a ramp test & compared the result to your AI-generated FTP & found a difference.
I think that’s awesome! Very interesting. How do you know which one is wrong? How do you know either one is correct?
Or do you mean it’s a good idea to do a monthly ramp test just to evaluate how you are responding to training? Sounds good to me! I would be curious to hear from you how changes in ramp test results track with changes in AI FTP suggestions. For sure, I would follow that thread.
As for me, I can absolutely assure you that my AI FTP is not the power I can hold for 45 minutes to an hour. Also, I can absolutely assure you that 75% of my best ramp test 1-minute power is not the power I can hold for 45 minutes to an hour. You can read my historic threads on ramp test issues and most particularly my experience with Gollnick workouts.
As the Coyle/Coggin ‘Determinants…’ paper showed, there is tremendous variability in fractional utilization among even well-trained cyclists. That is the failing of the ramp test as a predictor of FTP. But it doesn’t mean that a ramp test can’t be used to evaluate training adaptation.
2 Likes
Good point! I agree! I’m going to go one step further to argue in favor of AI FTP…not for this specific individual case but for the TrainerRoad user base as a whole: the difference in workout compliance before and after AI FTP implementation is night and day.
There is way better workout compliance after AI FTP. Previously you could find several workouts that were regularly assigned and the ride feed looked liken an apocalypse. Sometimes 30% to 40% of rides were failures. You don’t find that anymore. Anyhow, that’s what my data scraping says. Maybe TrainerRoad can confirm…
6 Likes
To confirm, the title of my post was AI FTP Detection not Accurate. I stand by that.
AI Detection FTP - 232
TR Ramp Test 249
20 Minute Outside Threshold Test 251.
Done numerous quarterly FTP tests with my pricey coach and generally were consistent and in the 250 range +/- 3 points
I’ve never seen such a dramatic FTP drop like I got with the TR AI Detection
Regards
1 Like
And what your 20min indoor test results? You can’t really compare indoor and outdoor. My 20 min best in the last (4 years) was 313w outside (corresponding to an FTP 282-297w). There’s no way I could have trained at that FTP indoors though and looking at my FTP history I got 250w from an indoor 20mins test at the time. Its been a bugbear for a lot of users that TR doesn’t let you have an indoor and outdoor FTP as they can be quite different. FWIW AI FTP D when it was introduced a year later after a few adaptions it raised my FTP to 282w.
The 20 minute test protocol in Training and Racing with a Power Meter is a 5 minute blow out effort followed by the 20 minute effort. Then you take 95%.
251 * .95 = 238 watts
I assume that you did not do the blowout effort.
232 vs 238 is the same ballpark and within the margin of error of your power meter.
I bet you cannot hold 251 watts for 45-60 minutes.
5 Likes
A better title would be “ai ftp detection does not match a ramp test” just because you want a higher ftp does not mean the lower one is incorrect.
Have you done the same protocol as your coach had you doing recently? As others said you can’t expect indoor and outdoor ftp to match so doing that inside like the ramp test would be the best way to compare.
Or do what has been repeatedly suggested and verify if the ftp you want is correct or not.
Based on his outside 20 minute effort, we are now down to six watts between that effort and ai ftp. 6 watts is nothing - 2.5%.
Pretty good for an AI to guess your FTP!
3 Likes
@OliverTwist How do you know your ramp test FTP is correct? How do you know your 20min FTP test is correct? No wrong answers…
2 Likes
I am perfectly okay with what the aiFTP is suggesting. I HATED taking the ramp test every 6-ish weeks.
3 Likes
Statistically we have two data points. Nothing says either one is accurate.
This whole debate was dropped long ago as the wrong FTP would impact your training. If you are using AIFTP and continually doing harder workouts You end up training at a level that is appropriate for you. Again there are other workouts you can do to see if you are an appropriate FTP. What level of threshold workouts can you do? This is where the actual FTP you are using becomes more important. I get too high of an FTP with a ramp test so your variance there doesnt really surprise me.
1 Like
So with so many FTP test variables maybe we dispense with all of the high tech gear and AI wizardry and simply go back to Perceived Exertion and Hearth Rate