How do people feel about the accuracy of FTP AI detection?

I would think not, but I also wonder how many people actually see their max heart rate.

Assuming a typical seasonal structure and a break of about a week probably 1-2x/year immediately after their break :p. I will hit HR numbers I never see after the off season if I do any high intensity work.

1 Like

Completed a 15 minute test this morning. Had a few weaknesses I’ll need to address, largely due to the set up. Since I don’t have a matching cassette, trying to adjust power to the trainer via my Garmin was a bit wonky. Also, since I do all but 3-4 workouts a month on the trainer using TR in ERG mode, I think I need to spend more time not in ERG mode and get use to getting to the target power, adjusting via the Garmin head unit, and not always being in my ‘perfect’ cadence. And finally, my last excuse is that I’m still getting a bit used to the e-flex plus and was a bit considered about falling off the trainer at the end of the ride as I push pretty deep into the hurt zone.

Now that all of my excuses are out of the way, I was only able to manage a 15 minute effort. I shot for the stars and it was definitely a little bit high. Instead of just completely falling off at the 15 minute mark and skewing my effort, I stopped it at 15 minutes.

Intervals.icu still automatically calculated a new FTP. I held 295 watts for 15 minutes. TrainerRoad AI currently pegged my AIFTP at 261 watts. intervals.icu calculated a 275 watt FTP. Based on prior FTP tests, AI calculations, and perceived current fitness levels, I would have pegged it around 270ish, which I hit during an FTP test last June and had my AI FTP calculated there around last October/November.

I’m aware that 15 minutes is a bit shorter than the typical 20 minute minimum for FTP calculations, so I’m debating where to set my FTP. I may just do the 275 and see how the workouts feel. Spending some time not in ERG mode I think will be beneficial in some aspects as well once I figure out how to better control resistance. I found that it was best to try and focus on cadence number once I set the manual resistance vs looking at the 3 second average power. At resistance 4 and 88 RPM’s I was right around the 295-300 watt range, and resistance 3 at around 98 RPM’s was about the same power. Intervals.icu chart below - you can see that I adjusted the cadence (power change) part way through the interval session.

There’s a workout people used to recommend for validating your FTP called Lamarck, which I think is 4x10@FTP. You could try that with your 270 and see how it goes.

4 Likes

Thanks for the suggestion! The 2 minute breaks between intervals looks brutal, however, I should technically be able to (under excellent conditions and rest) be able to hold my FTP for 4x10! I’ll need to mentally remind myself that todays effort was a 15 minute power, not FTP, as there’s no way I could hold todays power for 4x10.

A bit cooked for this type intensity, but may give that a go in the next couple weeks.

1 Like

It sounds like your fitness is moving up in the direction you want it to!

Keep in mind that there are more aspects to your fitness than your FTP alone – if you’re hitting PRs, especially on your longer rides, that’s a sign that you’re getting stronger overall. :muscle:

I checked out your TR Calendar, and, as you said, November and December were a bit lower than what you normally seem to do in terms of volume/intensity, while in January, you really got back after it. Remember that it takes time and consistency to increase your FTP – it looks like you’re off to an awesome (re)start after January, and I think if you keep it up, you’re likely to see your FTP trending upward as you keep training!

1 Like

My first AI result was 2 watts lower than my first ramp test. :man_shrugging: I’ve trained with HR monitor and by feel for decades, so this PM stuff is new for me.

I’m still working up the PLs, but generally the workouts feel appropriate as far as RPE and what I think I know about my HR zones. The endurance zone starts at a lower HR than I would have thought, but sweet spot and TH feel about right. My VO2 max workouts so far have me just sniffing my anaerobic threshold HR. However, do 30/30s give the HR enough time to react? I think when they start asking for 2+ minutes at those watts the HR will get there.

So far I like the progression, as I would probably err on the side of too much too soon for the higher intensity stuff. My usual method for VO2 max intervals would be to ramp up to a higher tempo HR before starting the interval, then starting the timer and adding more effort so that most of the interval would be above LTHR. The PM method is much different. I get “credit” for time in zone right away, before my body even starts complaining. :grinning:

2 Likes

You could even just try doing Lamarck at the 261. I agree with you, rest a bit since you just tried to do a test, but if you can do it at 261 without feeling like you’re about to die, you’ll know you’re at least close!

1 Like

There is no benefit to setting your FTP a bit high. You are far better off with it being set a bit low. This allows you to nail all your workouts and even choose harder workouts when you’re feeling great. However, a FTP a bit high has you struggling and can be demoralizing.

5 Likes

This really only applies when your ftp is slightly off, if it’s 10-20w too low you won’t be getting the right stimulus to improve.

Wrong.

Self-esteem, peer jealousy, attraction to the opposite sex…the list of benefits to a higher FTP is nearly endless.

:stuck_out_tongue_winking_eye:

7 Likes

I don’t disagree, but do you think that’s going to be the case in this situation?

I mean… 275 is what intervals.icu calculated based on today’s effort. I would consider it accurate, and probably not low or high.

Many could probably argue that intervals.icu evaluation of a 15 minute test might be more accurate than the AIFTP.

That being said, I do agree with your overall statement and as of now, am planning to set it at 270 and see how it goes. I have 4 races in the next 10 weeks, so probably won’t do another proper test for a while.

From what they have posted I would say 270 is good, it might be a bit higher but I would want to see more data before pushing it up too much.

1 Like

Yeah, I was more thinking about looking for break points, where your HR is relatively steady and then goes up or down as a way to identify/verify thresholds.

1 Like

Isn’t intervals just taking the best power over a time period and multiplying by 0.95 or something. It’s nothing magic.

1 Like

The multiplier depends on the duration being evaluated. For 20 minutes I believe the calculation is .95. I was unable to find a formula that calculated other durations, but then I saw intervals.icu did it automatically.

No, it uses a pdc. I personally don’t trust it for durations under ~30m for the same reasons as ramp tests or any shorter efforts.

1 Like