Why is my AI FTP Detection lower?

TrainerRoad Software
89

I am 56 using TR in 10 years and I guess I also will be TR user the next 10 years. My FTP has been quite consistant the last couple of years, which in my age i guess is fine. Now its not about getting stronger but about not getting weaker :grinning_face:

In november 24 I hired a private coach and took a lactate test. It gave My 8 % lower FTP than TRs old FTP detection tool gave me. Nov 25 I startet a TR plan and TR FTP detection again gave me a higher FTP about 8% higher than the lactate test. Now the New AI tool brings me back in line with the FTP the lactate test showed.

I guess many of us tend to train to hard at to hard intensities given an overestimated FTP. If the New tool brings it down a little, may be its not a bad thing. I have had many Over under sessions being totally burried in the last minutes.

91

Feb 1, got on the trainer today and AI FTP 8-Ball, random spin detection dropped my FTP almost 30%.

Yes, I’ve been off the bike for 10 weeks or so, I have been hiking, lifting, running and swimming 6 days a week, not dormant - seriously WTAH.

92

Try the FTP you think your at and see how you get on?

93

I ignored it. My FTP guess is as good as TR’s guess at this point. If I’d have accepted, a threshold today would have been easy Z2 3 months ago :roll_eyes:

94

You haven’t given it any cycling data for 10 weeks and you’re upset that it’s not very accurate? It’s not magic.

95

Agreed, 30% though…come on!!!

I haven’t been sedentary: 10k runs, 3-4k swims etc 6 days a week…same as past years, just a massive drop vs other years. Just sharing.

96

Oh, I don’t think anyone is saying you’re not fit at all. Sounds like you had a great winter off the bike. I just think that you’re probably going to have to do a week or two of good work to teach the model. Good luck!

97

Having just had three months off the bike, I came back using an FTP 33% down from last year. Definitely about right for me. If you were fit before the break, 30% is definitely possible. I’m back up about 10% from my low base after a month’s block.

For transparency - I don’t use TR these days, so it’s not AIFTP

98

I agree, so far there has been no scientific rigour applied. Nate has previously hinted at an interest for doing science. This would be a great opportunity to showcase it.

99

I dropped from 241 to.206….lol…makes you want to give up.

100

Don’t sweat it! I ended up accepting my new AI ftp and my workouts have been just fine since. Plus ftp prediction has me well above my old ftp in just a few weeks. #TrustTheProcess :grinning_face_with_smiling_eyes:

101

I accepted it and it’s started going back up already. 214 is predicted after 11 workouts. Maybe it was too high as I was struggling with some workouts , it’ll find the right level soon.

102

I had a similar situation when I first hopped on TR AI.

I had been strength training and doing structured runs each week, often hitting my max HR during hill strides.

I don’t know if we’re able to translate that data into bike fitness yet, hence the drop.

Once I got my FTP figured out, things were spot on, and I’m really enjoying how things are working now.

I shared my case with the team, and they are aware that there are cases like this, so we’ll find a better solution moving forward, I imagine.

103

Keep in mind that most athletes don’t regularly test lactate, so it makes that data somewhat less valuable.

While your FTP might be set slightly higher than what your lactate data showed, you still have Athlete Levels in each training zone (including threshold) that allow you to throttle workout intensity in each individually.

I’d try using the FTP we gave you to see how that feels. :+1:

Let me know what you find!

104

So what’s more accurate, AI or ramp test?

105

What “ecosystem” are you in?

106

What are you attempting to accurately measure, what are you wanting to use the result for?

107

I’m just curious because I’ve followed a TrainerRoad program before and I can almost guarantee that the current AI suggestion of my FTP is probably maybe as much as 20 W less than I would produce in a ramp test.

108

“Accurate” depends on what you are using it for.

109

A rifle could be accurate or inaccurate, regardless of whether it’s used for serious target shooting or it’s a safe queen and never leaves the house.

I guess I will answer my own question and do a ramp test and compare it against the AI generated FTP and then compare it based on the projected efforts in my training program see if there’s an accurate correlation. I thought it was a simple question, but I guess it wasn’t. My sincere apologies..