Adaptive Training Closed Beta Update

Something I am still kinda confused on is, how much does the level really matter? If you have your FTP set correctly, and you do a .95 IF workout but you level down and end up with a .94 IF workout, do you really think you are losing fitness? What if the workout AFTER that .94 ends up being a .96 with a lower level? DO you still feel you are losing out on fitness gains?

I definitely agree that the survey needs to be described better, and so should what the levels mean. I don’t really care what level it gives me. If I am struggling on a VO2 workout, but get another VO2 workout that is “easier” I still expect to struggle as long as the FTP setting is the same.


Maybe this has been addressed, but I wonder what AT does when you adjust the workout intensity. I did a VO2 max workout that was too easy. I ended up increasing the intensity by +5%. Should I rate how the workout felt at that +5%, or how it felt at its original intensity?

How the workout you completed felt.

That includes any modifications you made via Workout Intensity adjustment, or exceeding / missing power targets in Res or Std modes.

1 Like

Hey. Glad to have been added to the beta. I’ve been using TR for c. 8 months solid and the progress I’ve seen has been great. Apologies for the potentially newbie questions, but some observations of my experience thus far using IOS:

  1. added a sustained build LV plan;
  2. did a ramp test got my levels reset, as expected;
  3. no adaptations - so all of my current workouts are either stretch or breakthrough as they seem to be linked to my old FTP estimate;
  4. skipped a workout in favour of an outdoor unstructured ride yesterday - no adaptation.

On 3) makes sense I guess as my FTP only changed about 5W, so what’s tagged as breakthrough, etc., should be achievable;

On 4) I estimated the TSS from my HR date (very cool).

Should I have seen any adaptations based on the above?

Was the plan added via Plan Builder or manually? If the latter, AFAIK you won’t get adaptations.

1 Like

@PotsieA - I think that’s the issue. I deleted the SB block and went through plan builder. The IOS app now checks for adaptations. Thank you.

1 Like

But this was the problem I and others were having. My VO2 max workouts usually (>90% of them) get a Very Hard rating. That’s how I feel they are. But they’re not digging me into a hole, that’s just how they feel to me. So, if AT was more automatic and didn’t prompt a survey, and I was nailing workouts and progressing with harder and harder VO2 max over the course of the plan, then great. But as is, I rate an Achievable or Productive workout as Very Hard, then get a lower workout next time, which I then hit and rate Very Hard again (likely because as I said, VO2 max is very hard to me). So now I’m just getting lower and lower workouts instead of actually progressing.


I think you can work around this by creating a PL index like FTP*PL.


Yup, this is it for me too. Plus, sometimes just getting on the bike feels hard, that doesn’t mean that it needs to suddenly make the next week easier… I’m not training for fun, I’m sure as sh*t not expecting it to be enjoyable.

This was why I gave up on Adaptive Training during my last block. It was insistent on reducing my output / progression when I knew that I could (and did) do more. I’ve got it back on now, and it’s already trying to make my sessions easier… despite the fact that it’s started me at level 1.0 for most things (due to the over the top decay they have coded in).

I’ve gotten to the point where I feel that if I use Adaptive Training I need to lie in my survey responses. For me now “very hard” doesn’t exist, and “All Out” is for when I feel like I failed a session (no matter what the app thinks).

1 Like

Interesting. For me, I’ve had very few Very Hard and only 1 All Out workouts. Would be interesting to know if it is a perception or workout selection difference.

Either way, it sounds like the survey isn’t working as intended for you.

I think that’s kind of what is being described as the expected classification. “All Out” is those sessions where you get the success survey, but your interval outputs were declining/you struggle to get them to the target.

i.e. success, but only just.

It sounds like it’s just a recalibration of terminology, which is what a lot of the discussion has been about recently.

Just my tuppence.

Sorry, I was trying to address how you could show progression with PL given changes in FTP (150w-300w). My thought is if you created an index of FTP at the time x PL, then you can show progressions as FTP changes.

Sure, but in my experience “very hard” usually results in the next sessions of that type being reduced in intensity, so at that point they might as well be all out anyway.

I think there’s an “off by one” error in the survey - the middle option should be the centre of the bell curve but it isn’t, “Moderate” is. Also if you go through the thread linked to by Chad then it’s apparent that the easy workouts like Pettit won’t normally get any response other than 1 or 2. Similarly the hard workouts like Leconte will normally get responses 2, 3 & 4. I.e. there’s (at least) two separate scales embedded in the 1-5 scale of the survey.

AT knows how a workout “should” feel and has the power & HR data to make that assessment but what it needs is how your current state of fitness found it, basically an automated version of a coach going: “You should be able to do this” and your response being: “That buried me.” so the coach eases things off a bit. In theory AT should learn that you rate hard workouts as “very hard” and adjust its suggestions accordingly.

If you do any VO2max workout properly it’s going to feel very hard on whatever assessment scale you use but it seems that the correct response to completing such a workout is “Moderate”! :man_shrugging:

Somewhere on one of the podcasts (possibly the one introducing AT) it was mentioned that the internal testers struggled with the system backing off on the intensity and then slowly building back up. They continually felt as if they were being given workouts that were too easy.


That chimes with my thinking: roughly speaking, select a response that’s one level lower than how you actually felt, and perhaps AT will adjust your plan more appropriately.

This wouldn’t be required so much if AT was, as others have discussed already, adjusting its interpretation of your responses on a personalised basis (ie. learning that you often responded “Very Hard” but still were able to complete the w/o OK and thus your present plan progression was maintainable), but anecdotally at least it doesn’t seem as if it’s doing that, and is instead taking a standard/fixed approach for everyone, hence the workaround tactic.

Hmmm, here is the updated info on rating a workout:

My coach prescribes vo2max workouts that are either controlled or max repeatable. He is expecting me to say it was hard or very hard, which maps to:

From where I sit - the ratings are NOT supposed to have the middle option (3) be the center of the bell curve. If AT is thinking like a coach, most of the vo2max workout responses should be 4 with some responding 3 (could have done another interval) or 5 (couldn’t finish the intervals as prescribed). :man_shrugging:


Yup, a ‘middle’ only makes sense of they were aiming to include ‘expectations’ in the mix (like a 3 for expecting a VO2 to be painful and anything on either side of the middle to be easier or harder than expected), which they have clearly stated is NOT the goal.

So, I agree that a properly set FTP, and a VO2 workout aligned with the rider’s current Progression Level should feel like a 4 in most cases, maybe a 3 if you are ripping and a 5 if you are off pace for any reason.

lol, did a Sweet Spot with a stretch rating, marked it as hard (keep in mind, all my levels were reset post FTP test)… so A.T. has made my session next Sunday notably easier than what I just did.

…aaaaaaand I’m out again. I’ll pick my sessions by hand.


I think this might be more a feature of the new plans than of AT itself. All of the new MV plans I’ve looked at have much easier Sunday SS workouts than previously, so they are not “meant” to be hard. That said, if you want harder, the alternates function is always there.

1 Like