Ai FTP and FTP values used by the prediction calcs

TrainerRoad Software
1

Quick one which might have been answered but my search couldn’t find it.

Had AI FTP last week with a drop from 211 to 201 which I guess makes sense as I’ve had 2 weeks off hiking in the Scottish mountains. I’ve decided to stick with 211 for a bit and see how it feels as I still feel pretty strong.

My prediction is currently giving me 205w, which is quite a drop. So this makes me think it’s using the last AI FTP value even if I said not to accept it. Not sure if TR folks can confirm (or explain) the values used? I think I need to accept it from other threads I’ve read.

My calendar is open so feel free to tear me to shreds

2

I would accept the revised FTP as the new model will give you the workouts that best serve you where your fitness is now and build upon that.
On a tangential point, looking at your calendar you may find that doing another sweetspot or threshold session next week might help regain some fitness quicker since you’ve had a couple of weeks off already and don’t need an easier week of just endurance.

3

5% drop seems like a large amount for 2 ½ weeks off especially when outside rides just before adventures were hitting yearly PBs. The hour endurance that was scheduled for tomorrow I did today was super easy too.

Guess I should trust the plan really, outside rides will probably help bump it up fairly quickly too. Might swap the existing to SS or threshold, see how they feel and then accept or carry on accordingly.

4

I’d be careful with that assessment. If I spent a week hiking, I would definitely not come back with fresh legs.

5

Very true! I am not conditioned for heavy packs right now!

Hiking bit of a tangent (albeit great fun) mind you, from the initial question for tr folks.

  • Is 5% drop expected after 2 ½ weeks off when hitting PBs outside before holiday
  • Does the prediction look at my FTP as is, or what prediction says? Going by that I said to ignore for now?
  • From the above, after 1 workout after the non accepted AI FTP that I completed and marked easy why should it predict further drops?

Screenshot_20260503-222709
Screenshot_20260503-2227091080×2400 150 KB

6

That’s a good point!

7

Screenshot_20260504-121634
Screenshot_20260504-1216341080×773 62.2 KB

Now after 90 mins outside in the sun it’s going up! Which is what I’d expect as it normally has been giving me perfectly judged workouts.

I should say I’m not grumbling, the FTP has generally been great for me I’m just a bit puzzled with the speed of degradation.

8

Looking at your calendar it’s more like 3 weeks without cycling and 5 days is marked with sickness. Coupling that with a bit conflicting signals the week before, a 3.0 threshold workout is very hard, but you do 80 minutes at .97 IF two days later, I can see why lowering your FTP that much seemed reasonable.

9

True, that answers the drop (probably), just now wondering what value is used to predict the next detection values.

Still not decided to accept it mind you!

10

I’m pretty sure it’s neither. The AI knows what wattages it thinks you can hold and prescribes training and predicts FTP based on that.

11

Thanks! Actually that does sound familar.

12

There are some good points here.

Your fitness will start to fall depressingly quickly when you’re not cycling. It is possible to keep some aerobic fitness with other activities, but it’s hard for TR to apply that to your on-the-bike fitness.

Once you’re back training, we’ll quickly figure out where you’re at, though.

@Torneng is right about the fact that TR AI works by looking at the data you give it and calculating FTP detections/predictions from that.