Perfect Example of why the Ai Prediction is so incredibly frustrating

1

I started with TrainerRoad around 2 months ago. I was well trained before but wanted to see if I can push myself a bit further with the Ai prescribed workouts and also wanted to try out something new. I did a Ramp Test before I signed up and ended up with an FTP of 322 Watts. Right after the workout TR told me it could assess my FTP and told me it estimated my real FTP to be 336 Watts (obviously too high). It also predicted that with the 3-Month Training plan that I selected (3 Interval Sessions), my FTP would go up to 350+ Watts (also completely unrealistic).

I essentially just ignored the numbers since I didn’t care too much and simply told myself that the Ai needed proper data through TR prescribed workouts to assess my fitness better. So I started my training, had to rate 1-2 Interval sessions as “Maximum Effort” (aka. I had to adjust the power down towards the end) and the Ai started to downgrade my predicted FTP accordingly to a point where at the end of the block it told me my new FTP was now 327 Watts.

So I ended up with a assessed FTP that was below what it initially had me noted down as. Not a problem for me, since the 350 and the 336 numbers were highly unrealistic and if I assumed my Ramp Test of 322 Watts as a baseline this would still indicated a ~5 Watts increase. Nothing crazy considering that I increased the number of Interval Workouts I did per week from 2 to 3, but hey it’s an additional 5 watts, I’ll take it. Especially since FTP as a number doesn’t really mean all that much to me.

Up to now the predictions didn’t bother me too much. The workouts were challenging and I felt I was making progress. The new prediction the Ai gave me was around 333 Watts if I remember correctly. Once again a 3 week Interval Block and 1 Endurance/Rest Week (that coincided with a trip I had planned). So far so good. Over the next few weeks the Ai slowly corrected the predicted FTP down again to 330 as I had 1-2 workouts that I rated “Very Hard” instead of “Hard”. Most of them Vo2Max Workouts where I simply didn’t think I could add another Interval Block, so according to Eddies I rated it as “Very Hard”.

Which brings us to today. The last workout of the 3 week block. Every one of the last 3 weeks followed the same pattern: 2 VO2Max Intervals + 1 Threshold Workout. After now building up some fatigue this weeks Threshold Workout was harder than the last 2 and predicted to be a 50/50 between Hard/Very Hard. I completed the workout, pushed through the last block and rated it “Very Hard” since I don’t think I could have done a full 5th interval block without having to adjust power towards the end. The Result? My Ai FTP Prediction got dropped from 330 to 328.

So for my 4 weeks of effort, in which I followed the plan pretty much religiously I’m now predicted to improve my FTP by 1 Watt from 327 to 328. At this point it just feels like the Ai is taking the piss :roll_eyes: Nothing beats completing a hard workout at the end of a training block, a workout that’s significantly harder than the ones prescribed during previous weeks and then being “punished” for it. I had a 3 Watt increase dangled in front of my face and it got basically taken away because I gave it the “wrong” RPE.

Now I don’t really care that much about FTP as a number. First of all I don’t think the TR FTP is an actual FTP but should be more marketed as a type of Fitness Score that TR assigns you. Second I am quite content with the workouts I’m being prescribed. I’m also not someone who’s self-worth is tied to his FTP. I said in the past it’s in my eyes essentially the cyclists’ version of “How much do you bench bro” and mostly relevant to design workouts (which as I said I’m fairly happy with).

BUT: TrainerRoad put this thing front and center in the app and web interface. It’s essentially the KEY element of the dashboard. It’s a carrot dangled in everyone’s face, promising big improvements to ones fitness. So when I read how frustrated people are with the way the predictions work, I can absolutely understand it. It’s is incredibly frustrating to go to the limit, nail a hard workout and then being “punished” because the Ai didn’t like the RPE rating. And “Just ignore it” can’t be the answer, not when TR makes it the central part of their platform.

Personally I really think that the Ai FTP Prediction should be rebranded or changed/removed. At the moment this isn’t really a source of motivation, it’s just a super frustrating experience in my opinion. Which is honestly a shame, because the workouts themselves are very much on point and at the end of the day that’s the most important part.

Just my 2 cents.

Page1
Page12142×1033 372 KB

Page2
Page22150×1052 265 KB

2

I think they could really benefit from just saying your AIFTP will be “current AIFTP +/- 1”. It would be better to be surprised upward at the end of the month than adjusted downward multiple times per month.

Also, get rid of this kind of marketing

IMG_1556
IMG_15562048×530 67.6 KB

3

I agree, at least for me the training itself works pretty well. What annoys me is the marketing and focus on a number that even they themselves isn’t really your FTP. If you need that carrot rebrand into a TrainerRoad Fitness Score or whatever.

The funny part is that the final interval session next week (before my trip) is one that I manually put in (to make sure I get in one last interval workout before I travel). The initial plan didn’t have that in there, but had a Zone 2 Workout instead. If I now remove that VO2Max workout again, my FTP prediction drops to 326 :rofl: Telling me that after a month of following the training plan and doing 3 Interval Workouts per week my FTP would DECREASE :man_facepalming:

4

Don’t have much to add, except to say I agree.

Note: I’ve been using TR since 2014. I’ve seen most versions and updates to the platform. I cannot be certain, but I think the latest change has led to customers leaving. No business wants that.

5

In my opinion TR did themselves a great disservice when they introduced the FTP prediction. Anyone paying attention in statistics class knows that it’s near impossible to extrapolate a trend and get it within a reasonable margin of error. Heck, even a formal FTP test isn’t that reliable day to day.

My advice is to just do the the work and not focus to much on specific numbers. If you get faster the plan is working.

Btw, you could actually get slower from doing 3 hard sessions a week rather than just 2.

6

This is the way.

I’ve concluded that if I follow the plan and properly execute the workouts, fuel, rest, and adapt then I’m winning. The FTP prediction has become just that. It’s not a promise or proof of my ftp - proof of FTP would be executing the actual test.

I get that it’s frustrating when the prediction doesn’t bear the fruit you were hoping for, but you did so the work so the reality is that you did indeed accomplish something.

Do feel stronger, is your power curve showing improvement? I think it’s important to look at other metrics to assess increases in fitness over time. Too much weight on the FTP number can be discouraging.

7

Did you rate the workout accurately? If so then how is this an example of TR “punishing” you. This “TR punished me” logic breaks my brain. TR is responding to your biology. RESPONDING to YOUR results. If you want to be annoyed at something be annoyed at yourself. Or your parents. Maybe you’re at your genetic limit. But you found the workout harder than the prediction thought you would and it adjusted its expectations. It’s not “punishment” for heaven’s sake, its just describing the reality inside your body as YOU reported it.

Few other thoughts:

  1. you said you are new using TR. Accuracy will naturally increase with more data

  2. I don’t know how old you are or anything about you really but being in the 330s is fairly advanced for a non-pro. Perhaps you’re nearing your genetic limit where progress moves much much slower if it moves at all. You are well-trained, you already know this.

  3. maybe try dropping one intensity day? Your schedule is pretty wall-to-wall and is impressive from a consistency standpoint, kudos on that :+1:t3: But if you can’t recover that quickly then its kind of all for nothing. Personally, I was stuck until i dropped an intensity day and came down to 5 days/wk from 6 and got more recovery. Then the progress turned upwards again.

  4. turn off the prediction. You said you wanted it “removed”. Remove it.

EDIT:
5) also, not for nothing, ramp tests are well known to skew high. If that was true in your case then you may have improved more than you think

8

Why? Its working for a lot of riders.

To be fair it does say “TRY” it now :laughing:

9

Because it isn’t working for some, and their answer to some of those people is “our training won’t always raise your ftp”

The “try” placement is funny though :rofl:

10

I really think you haven’t read my post completely :sweat_smile: Or if you did you are ignoring some pretty important parts of it :squinting_face_with_tongue:

11

Maybe you should give my post another read? Certainly seems like you either didn’t finish it or didn’t understand it :man_shrugging:

12

I have read it twice. I fail to see what I didn’t address.
I’m happy to respond to anything you feel like bringing to my attention.

As far as I can tell this is the only paragraph I didn’t directly address.

Also, as I pointed out in my edit above… ramp tests often skew high. If that happened here then you had a pretty good 2 months IMO. Especially for someone up in the 330s.

But TBH, most of this entire conversation lives well within the margin for error across the board. I understand why its frustrating though. Its a frustrating sport a lot of the time.

13

At first I thought these are a lot of words for someone who “doesn’t care” :wink: but ultimately agree. If they just called it what it is, a TR training score, there’d be less pushback.

In the podcast with Nate it does seem they care and really are working to get it right. So I’ll give them that.

14

True, I just wanted to make sure I properly explain it. As I said: I like the workouts I get. And I know I’m improving.

It’s just the marketing that really has me triggered and the de-motivating effect that the whole Ai FTP Prediction mumbo jumbo has at times. If they didn’t make it front and center everywhere, I wouldn’t care at all.

15

Then I’d suggest reading it a third time. Especially the final paragraphs. You also make a lot of assumptions about me, which I find interesting. I’ve honestly met quite a few people like you in forums and don’t really see the point in arguing with you about this. I thought I made my key points quite clear, but hey, maybe I’m wrong :man_shrugging: Let’s agree to disagree I guess

16

You know you can turn off AiFTP prediction, right? Easy to do and will eliminate a lot of frustration. I turned it off two months ago.

17

lol took the words right out of my mouth. I’m glad we finally arrived at a point of agreement.

:+1:t3:

I’m sorry TR isn’t working for you.

18

During Base Phase or if only doing TrainNow workouts I don’t mind the AIFTP prediction and feel it has been a valuable tool when I have to rearrange training days.
During Build and Specialty Phase I turn it off, I don’t want it influencing my RPE rating.
I find that it feels pretty good not having that number front and center when opening that app.

19

Oh I like that approach. I find myself being both critical of the current implementation, and yet still making adjustments that align with whatever makes the prediction go up.

FTP isn’t the only metric. If I choose a build/specialty that puts more anaerobic focus it isn’t well represented by predicted FTP. Yet it’s still the metric that sits there judging me.

20

I was having the problem where in the goals that I set for the season I want to hit a set Watt per Kg for CX season looking at both weight and FTP.
Once I get some good gains and maybe a Hard workout was more like Very Hard that day, that little knucklehead voice starts talking to me. :joy:
“Come on was it really Very Hard” just not wanting to take the AIFTP hit but not thinking about what else I have gained in the Build/Speciality Phases.
That had got me in the past.