I have been using TR for a few months now and have been quite happy with it. Particularly as it seems that the AI has improved and it’s generally giving me workouts that feel more appropriate at all times (not in the beta by the way, although would love to get access ).
I am following a plan targeting some races in spring - 5 days, 7 hours/week, with hard intervals on Tuesday, Thursday and Saturday, a 30 min easy ride on Wednesday and 1.5h endurance on Sunday.
On Tuesday, I completed Haw (VO2 Max 6.1) and rated it Very hard. No yellow days/adaptations triggered. Then Wednesday I completed the scheduled Manly (Endurance 2.6) and rated it Easy. I’ve got Antelope +2 scheduled for today (Thursday) and had Colline (Threshold 5.0) initially scheduled for Saturday. But upon completion of Manly, Main (Threshold 2.0) was suggested in place of Colline for Saturday. I do note it has a higher TSS and IF than Colline, but still not sure as to why that was considered more appropriate. My current Threshold PL is 4.3. I have accepted the adaptation regardless, but would like to understand the reasoning behind the change.
First of all I would’nt stress it too much if you have a SS ride today. It might well adapt again after that one. The overall reasoning is always that the Ai picks the best workout for you all the time, and I’m not sure any more details than that can or will be shared. Speculating a bit it might be that the Ai expects that Main will be the best workout for you based on how your Tuesday workout went, and how it expects todays workout to go.
Yes, that’s what I thought, not going to dwell on it and as you said it is likely it’ll adapt again anyway! Not really expecting to understand AI, was just surprised as previously I could always kind of see the reasoning for any adaptations, this one just felt strange.
I think there is some fine council here to just not dwell on it and move forward. But I also can’t get to the point where I have this level of trust in the system. There should be a reason that AI is suggesting a workout over another. But with some of the very minor adaptations, I find it impossible to believe that the AI suggestions could ever be proven as a “better” workout for you. It appears to me that some of the minor changes should not be needed at all. I would wager that the best coaches in the world working with the top riders are not making near as many adjustments in training plans as AI does. Are we to believe the TR’s AI would outperform those coaches?
I’m in the beta so can see that as well as “main” being predicted to be hard by AI - it also has a SS level of 7.4 - so you are definitely not been served up something too easy and unproductive.
I think it’s just confusing the PL calculation because a good portion of the intervals are not quite in the threshold zone - but it will have the same results.
Essentially main is harder than a normal PL 2 workout - and even if you are not in the beta - TR has been using the improved algorithims to prescribe workouts for a while - not just blindly stepping up workout PL.
As I see it there are two different types of adaptions here. You have the ones that the OP has been given which are quite noticable changes (going from 4x12 at 99% to 14x6 at 85-95%). I would expect coaches to also do similar adjustments if they think the athlete would benefit from more work at a lower power.
Then you have all the micro adjustments TR does that you mention. Going from a 1.2 to a 1.3 30 minute endurance workout or adapting back and forth between workouts that to most humans are deemed as identical. I totally agree that there are way to many of those adaptions. In my opinion TR would benefit from applying a filter to filter out adaptions that don’t have a meaningfull physological impact.
Why? The models been run, the energy has been used. It has to check, you mine as well get what it thinks is the best workout. It probably doesn’t have a meaningful impact one time but over a course of a year it could add up.
Because for the athlete it’s just noise. I have a plan that includes a 30 minute endurance ride on Wednesdays. More or less every day of the week there are adaptions for that workout, sometimes even multiple times a day. Usually it’s just back and forth between two almost identical workouts. Either just pick one and stick to that or at least wait closer to the day to start showing adaptions. As it is now it just makes it harder to pick up on meaningfull adaptions.
I’m currently experiencing something similar. Completed a productive sweet spot session earlier in the week, marked it as moderate, only to have my weekend under/over workout reduced from something progressive to something achievable. My previous under/over workout last weekend went well - it was a bigger step up in WL than expected, but I handled it better than expected (marked it as a hard effort). I thought it might have recalculated by now, but no. One more day to go until I find out.
At this point in time I’m happy to accept that (updated?) AI thinks that the achievable under/over session is the best thing for future fitness gains, and I simply hate under/over workouts, so will gladly take the easier option! It will be a mental blow if I wake up tomorrow and find the workout has been adapted upwards again, but as this is one of my last workouts before a recovery week, then I can’t see why it wouldn’t push me harder.
Edited to add: I’ve been using TR for many years. This is the first time I can recall that a future workout has been adapted downwards from productive to achievable, so something has changed in how these things are calculated.
Im only on a LV Masters plan but my mid week schedule is similar (Tues-Hard, Wed-Easy and Thu-Hard) and I’ve had similar adaptations in the past. My thinking is Tues has been analysed as pushing you to the edge of needing adaption but if you rested completely on Wednesday no adaption would be necessary. But you/I have done another work out (albeit easy) and after that AI has reanalysed things and your now tipped over the edge and Thursday is adapted. I can only think in the long term it doesn’t really matter.
You could pin that workout as one of the two options and it would not change on you. You could do it for the entire plan in about a minute. If that is the one that causes frustration.
I’ve noticed recently that sometimes the adaptations to a ride won’t come right away. For example I’ll do a threshold ride, say it was only a 3, then nothing changes… until two days later and the next scheduled threshold ride gets bumped up in difficulty only then. I wonder if this is actually a thing, if you’re seeing this as a delayed response to your VO2 ride. Or perhaps Main is more specific to your event type?
Main also looks to have more TSS than Colline and more total time spent above 0.85. I would not assume that you’d find it easier than Colline - PLs in general don’t correlate that well with actual difficulty, and I think that’s one reason TR seems to be moving away from them. I wouldn’t overfocus just on the PL when predicting difficulty and training effect.
Even worst case scenario where Main is the “wrong” workout for you… this is like worrying that you’ll starve to death because a few crumbs of your dinner dropped on the floor. This would have basically no measurable impact on your overall training.
).