AI FTP Detection Update

I know :smirk:. I agree, it would cause endless confusion as it wouldn’t be able to be used as an acronym. It was tongue in cheek, and honestly I think it’s an accurate way to describe it.

You’re right a new name is needed. Back to the drawing board.

1 Like

Two nanoseconds after TRP comes out there will be an equation published to convert TRP to FTP.

Just like 4mMol lactate, we be stuck with FTP 4-eva



1 Like

I don’t think there are that many outliers anymore.

TrainerRoad has talked about putting AI Detection to the test by giving it users data then get it to predict the FTP increase. Then comparing that to the actual result. I am not saying there won’t be outliers just that they would be very extreme cases.

I would be more interested in seeing how it handles those that have hit their theoretical ceiling.

1 Like

I think you have to remember there are people who will get anxious when AIFTPD says 235 and they test and get 233 or 237.


I agree that there aren’t that many outliers anymore, but it could be that those athletes are no longer using TR because it didn’t work for them. I don’t use TR anymore because the model isn’t optimal for me, but I’m still a subscriber because I want TR to continue improving with user data.


I must start by saying that if anyone reading this suffers from this anxiety. I feel for you and wish you did not have to go through this.

FTP is not a perfect science it is what you were able to do on that day at that moment based on 100s of variables. If you were to use AI FTP D and not do the ramp test afterwards you avoid this situation? I am guilty of blindly using AI detection since I accept the detection will be close enough. (modelled off thousands of tests)

1 Like

Definitely NOT me. But let’s face it, there are definitely people who think Tempo ends on an exact number and SS starts on an exact number, etc.


As a satisfied user of a Stages left only power meter, I’ve long since accepted ‘close enough’. :grinning:


I agree to a point, but depending on your definition of FTP, there are certainly tests that very accurately determine the defined number in a repeatable way. I’ve been able to do FTP and V02 lab testing at the local university a couple times and there isn’t too much debate on the results assuming the test was properly administered (at least for the FTP portion). I’m not talking about FTP in the context of setting a number for training zones, but FTP as the physiological point where your lactate production and processing start to get upside down. To be fair, even the lab testing isn’t going to get you to an exact number because they aren’t pricking your ear every 30 seconds and there is also some lag between what you are doing and the blood markers, but it’s gonna get you pretty close and it’s highly repeatable. For those of us that geek out on this stuff, I’d highly recommend trying to get into a lab at least once. Try to find a a lab run by sports scientists who aren’t trying to sell you something (not a health club who bought a machine and may not have the expertise in sports science). The universities also run studies where testing is part of the study and it’s an easy way to get free testing if you fit their target population.

For all the “non-lab” tests, I agree that comparison isn’t that valuable and you really just want a test that is highly repeatable/consistent for whatever objective you are using the number for (like setting TR training zones). If just using an FTP number to set training zones, a “wrong” FTP might be a better number to use for most of your zones vs. the “right” number, so it’s not that critical if it’s right or not as long as it’s consistent.

If I’m actually wanting to determine my physiological FTP (and don’t have easy access to a lab), the most effective test I’ve found is over/unders. I know this isn’t typically mentioned with all the common FTP tests, but I find it’s a better gauge and more consistent because it’s not some crazy effort where you have to be highly motivated and bleed from your eyeballs to get a result. You still need to go in with a good idea of what wattage you should be straddling for the test, but the actual testing isn’t something I dread and it’s a higher quality training session compared to a ramp test or 20 minute max effort (in my opinion). It’s also the one place where I find heart rate indispensable. Not the absolute value, but what it’s doing in the context of the over/under intervals. With a properly set FTP, I’ll see HR crawling up slowly throughout the “over” intervals and steady or slightly dropping on the “under’s”.

1 Like

I’d be very wary if AI FTP was the only option and manual wasn’t. That’s because of my experience with it. In April I spent a month off the bike due to a family/work trip in Europe, and according to AI FTP my FTP reduced from something like 335 W to 315ish W. I did a ramp test and got 327 W. I verified that number with a bunch of workouts and a brisk outdoor ride. For some reason my reduction in fitness did not express itself with a reduced FTP, but atrocious endurance — my power bar did not change too much, but my life bar was much smaller than before.


I believe you’ll get the same choice. If we don’t do it on launch it will be a follow on. I’m asking now to see where it’s scheduled.


Yes, so you eluded to the issue.

It gave you a very low FTP because it didn’t have much data at all, and 9 weeks of no rides before it was run. We’re going to prevent it from being available in this case when we launch. But I also know the engineers are trying to work out specific examples like these so the model is more accurate for them.



That’s a good point, aerobic fitness is not just threshold power, it’s also how long you can hold that power. I used to think that because I could complete 3x15’ at FTP that my FTP was set correctly, even though I was gasping for air all three intervals and had severe tunnel-vision on the last one, 10/10 RPE. 3 months later my FTP hasn’t really gone up by much, but I can hold 3x15’ at the same power at 7/10 RPE.

1 Like

What if we did this:

Change FTP branding to tFTP (Training FTP).

Then put a 60-minute power approximation after every workout that would be based on your threshold level.

You could watch small changes from 200 → 202 if your threshold level went up.

Then everyone can rest easy because it’s tFTP plus you’d get the added benefit/motivation of having your 60-minute power change.

Also, there’s a greater chance of multiple zones being impacted with the new PLs. And VO2 or Anaerobic might influence threshold PLs too.


Yeah, it was very confusing, because I expected a much larger drop in FTP. So I went on an outdoor ride with a several known climbs. I did one above threshold (which too about 14 minutes) and another, shorter one at FTP. I did my second-best time on the VO2max climb and was able to do the FTP climb without any issues. In fact, I had to hold myself back.

But after the last climb my pacing plan was to take it easy. However, rather than averaging about 200 W, I limped home at 160 W. (I could have pushed harder, but that wasn’t the point of the ride.)

The next workout is harder.

kJs is not a good way to measure the difficulty of a workout, even in the same zone.

And we validate this with workout outcomes and we’ve tweaked the adjustment a few times. We want people to have productive workouts to get faster and achieve their goals.

We know this method works by the amount of success our athletes have had (large fitness increases, national championships, world championships, personal goals achieved, etc.).

The good news is we don’t force anyone to do it. So if you’d like to do your own testing protocol, get an FTP increase, and keep your threshold at the same level then go for it!


That seems like the wrong solution, because IMHO you haven’t identified the actual problem in your post.

TR’s FTP does aim to approximate the lactate threshold, because the purpose of this value is to set your training zones — using Coggan’s model.

The actual problem as I see it is that people use TR’s FTP as the main metric for fitness. Rather than renaming things, adding to confusion, I’d invest in creating and surfacing better performance metric and make those explicit to users. These should depend on the plan and your goals. E. g. if an athlete chooses a crit plan, then absolute short term power and repeatability are metrics. How often can you go above threshold and recover?

This would also help you tailor plans to the needs of the user. E. g. I have added a polarized block to my training after my drop in endurance, because my problem in May when I restarted training was not power, but endurance. And in my experience, a polarized block improves my endurance but doesn’t raise my FTP much.

I realize this is a HARD problem, but renaming FTP to tFTP does nothing to address it and just causes further confusion once someone digs deeper. (What is the difference between FTP and tFTP? But if tFTP isn’t the same as “Coggan’s FTP”, why does TR use Coggan’s zone model with tFTP = FTP? What is FTP really?)

TR and its athletes need better metrics. Too many of us are chasing power. Even though I am aware that FTP is not the only performance dimension that matters, it is really hard to quantify progress in the other dimensions. E. g. last year I did a power PR (117 % FTP for almost 7 minutes and an average heart rate of 159 bpm — and I had energy to spare). But how do I quantify that? Should I repeat the climb to assess certain aspects of my power?

1 Like

I would like this, it would help to give the athlete an idea on how their aerobic fitness is progressing. Depending on how your body adapts to a given block, you might see a big jump in your tFTP but your actually 60-minute power might not change very much.

1 Like

Family Feud GIF - Family Feud Good - Discover & Share GIFs