Adaptive Training Closed Beta Update

We are back to how you rate the workout. It does impact the adaptations.

1 Like

Hypothesis: AT takes “very hard” to mean at the limit of your capabilities. So it is backing off so you don’t crash and burn. Plus a “very hard” workout could easily have been an unsuccessful if you weren’t feeling 100%.

Again, total hypothesis on my part

Response from support:

Thank you for getting in touch with the TrainerRoad Support Team!

This is intended behavior - since you rated the workout as “very hard”, the system then knocks your future workouts down a few notches to make sure that you are not overextending yourself in the future.

This is one of the key features of Adaptive Training whereas in the previous plans if you struggle through a tough workout, the next one only gets even harder. Adaptive Training simply protects you from the possibility of overcooking it.

If “Very Hard” results in next comparable workout being 1.0 below current PL, I need to think carefully about the consequences of rating a workout in addition to how it actually felt. The workout was indeed very hard but doing my next Build phase workout 1.0 lower than my PL, as proposed by AI, feels like easing off too much.

I would appreciate feedback from those who have gone through full Base-Built-Specialty program under AI choosing to trust AI each time. Did it work for you?


Adaptive Training has a sense of humour, of a kind:



I’ve had the similar experience recently, here’s the link ty my post earlier in this thread:

There’s a whole thread about this issue:

I agree with you, as do some others. The survey as it is now, is a problem IMO. It’s not meant to be how it feels (according to TR). It’s meant to be, how was the workout compared to what you or it was “expected” to be.

For example: I have been rating all of my VO2 max workouts as Very Hard. Rarely a Hard, but certainly no Moderate or Easy. Because VO2 max is hard to me. I know it’s going to be a tough workout going in. According to TR, I should be rating these as Moderate. Which sounds crazy. In no way or world would I ever consider a VO2 max workout Moderate. They’re always tough. So my VO2 max adaptations have been lower every time. Similar with Threshold and Over Unders. They are hard, and as such, I’ve never rated any of them as Moderate. So far, the only workouts that get Moderate are Z2 rides like Pettit.

So basically, unknowingly, I have been messing up the algorithm. Because I’ve been rating things based on RPE.


With rating of workouts being not based on RPE but on how difficult I expected them to be I imagine AI adaptations would be more accurate. But then “moderate” should rather be “as expected”, etc. Official guidance on this would probably be helpful as a sticker to the Beta thread…


Yeah, that’s the reason we are pushing so hard for that guidance in the thread linked above. I kicked the spurs as hard as I can (via PM and that thread) and am waiting to see what they can tell us.


Thanks for all that, @mcneese.chad; we’re on the same page here and I’ve posted a comment on the thread you refererence.

1 Like

Still waiting for guidance from the team on that one. Sorry for the delay, we know its a big point of confusion, hold tight!


I’ve adjusted my mindset and I now rate my workouts according to how much I want to see the adaption change. It initially has all my levels FAR lower than what I’ve done historically, so I’ve been trying to push them to the right levels. They weren’t moving up fast enough for my liking, so i changed my thinking to “how can i best impact my next workout?”

So, if I’m doing a VO2 workout and when I finish I think to myself, “I expected that to be way harder based on what I normally do in a VO2 ride”, then I would rate it as “easy”. Boom, the next workout will be way harder. If I’m at my limit and I don’t want to fail the next one, I would rate it as “Hard”, expecting to see a smaller adaptation.


This parallels the list I made this morning, based on the TR support comments about the Pass survey ratings:


This may not be right or what TR really intended, but from the experiences we are having and TR’s support info, this seems closer than anything else I have seen.


The only thing here is that I’d probably never choose Hard or above if I completed the workout. If I hit a workout, I want the PL jump. So if I based my survey response with this, 99% of my completed workouts would get Easy or Moderate, and I’d rarely choose Hard. But I’d probably never choose Very Hard or All Out. Which to me sounds wrong.

1 Like

Yeah, to a point this leads me to what I used in my own personal 3-point “effort” rating system that I used for years before AT.

1 - Low: This was a rating for many Endurance workouts under 2 hours, and any workout that felt easier than expected. In some cases, I might have adjusted my planned workouts to be a tad harder if I was expecting a 2 - Medium.

2 - Med: This was my typical rating for a challenging workout, but one I completed fully with no “cheats”. This usually meant a “hard” workout (like Thresh, VO2, Anaerobic) went as planned, so no change in planning.

3 - High: In my system, I used this for super hard workouts or “fails”. This likely lead me to consider adjusting my plan.

  • TR’s system uses it (5-All Out) in the Pass survey, and has a separate “Struggle” one.
  • So when I use “5 All Out” now, it is because the workout smashed me and left me in a heap afterwards.

In a way, a simpler system with 3 basic outcomes:

  1. No change in planning, or a slight increase in difficulty if a “hard” workout was “too easy”.
  2. No changes, steady as she goes.
  3. Likely changes to decrease difficulty in the near term.

What TR seems to have now is a lopsided bias (2 as the middle in their scale), that may well be leading to limited or slowed adaptations for people that are judging workouts based on “feel” given the rather short words/definitions we were given (and a potential 3 middle). The additional “gray areas” that the 5 point system added from my 3 point version only muddy the waters IMO.


I think hard is still OK, though. The plan will bump up PL the following week. If you always pick moderate and bump the PLs for workouts that feel hard, it seems like that increase plus the PL progression built into the plan will soon lead to failed workouts.

Am I thinking about this right?

1 Like

Possibly. We are guessing here, but essentially the survey response is used to leverage how much (if at all) to alter our Progression Levels, which then leads to potential adaptations to the plan workouts on our calendar.

Effectively, it seems possible to choose your progression level. Pick “1-Easy” for everything, and you may well be in trouble in a week or even a few days depending on the related PL and pending workouts in that week.

Likewise, those of us that used 3-Hard or 4-Very Hard with any frequency, may have had delayed or limited progression in workout difficulty, and potentially “de-rated” plans that were below our actual abilities.

Ironically, it’s almost like we have a 3 or 4-point system since the top 1 or two are potentially “negative” in ways that are not fully intended.

  • Personally, the 5-All Out might as well be a “Struggle” since you get that survey at the end anyway. It’s essentially the fact that you barely survived enough to avoid triggering the actual “Struggle”, but darn near failed anyway.

Thanks. I got in last week, and haven’t had much chance to experience the joys and frustrations of AT.

I’ve settled somewhat on using clock watching as an additional input into how I rate workouts:

  • If I don’t look at the clock, or barely look at the clock the ride is easy or moderate
  • if I’m clocking watching, but only for the end of intervals, it moderate or hard
  • if I’m only getting through the interval by playing time games (e.g., just get to 5 minutes, get to 3 minutes, etc.) then it’s very hard

This is why we badly need objective guidance on this…

Hard or Very Hard are my default responses to pretty much all my interval days - surely any interval session (z4 and above) should be hard to achieve or you’re essentially doing the wrong session completely??? For me ‘Hard’ is about the need for focus and effort to get it done, while ‘Very Hard’ means there were moments I wondered if I’d get through it or would need to take a cheat break…but I didn’t. ‘All out’ would mean falling off the rollers at the end, needing serious recovery time before I even got off the bike itself and probably needing a lie down on the floor before I even contemplated rejoining the human race (benchmarked by the memory of a parachute regiment medic crouching over me offering first aid as I lay on the side of the airfield perimter track, still clipped into my TT bike after finishing a 10 mile TT…). So now I’m thinking I’m doing it all wrong :rofl: :rofl:


Agreed. I’ve never in my life done a VO2 max workout that would be rated less than a hard, and I’d say most would be Very Hard. Which messes with the current system apparently. Which is why we’re trying to get clarification.