But they were at a 5.0…. so even though they manually did a ‘hero workout’ to get to 6.2… the recommended 5.6 is smart for long term success… then maybe next time 5.8 or 6.
However they are not really using that system anymore, its more of a carry over and to show relative efforts in systems.
Mine however… due to too much running/weights throwing off TSS i assume, does regress… almost week over week… intervals are getting shorter… not longer. Seems to mellowed out with a demanding approach and editting a response on RPE (from VH to H)
Oh, yea I said that because it prompted me with the survey why I cut the workout short. My original comment was making fun of AI because I cut out rest intervals making the workout harder, and it was too dumb to realize this and all it saw was that I stopped 13 seconds early. So it was a joke because any human with half a brain could look at that and see I didn’t fail the workout, but AI was having trouble with it.
The bigger issue was that even though it “credited” me with a 6.2, that level didn’t seem to transfer on for my next workout, which was a 5.6. And much easier from my TrainingPeaks RPE I gave it. So it seemed like AI didn’t truly give me the 6.2.
This is exactly my point. I want to know why I’m being given easier workouts when everything else says I should be getting harder, that is, progressive workouts.
What would be helpful would be to know what these three workouts are actually made up of, so we can try to assess the delta between them in terms of watts and minutes rather than the abstract 5.0, 5.6 and 6.2.
The 5.0 that turned into a 6.2 was posted above. The 5.6 TR gave me next was Glassy +1. Neither were that difficult at all. Even at the increased intensity I rated them Moderate. If I had done the stock workouts it would have probably been an Easy. The funniest part is that the 5.0 workout had a predicted difficulty of Hard 68% and Glassy +1 was over 50% predicted Hard. So both were far easier than predicted.
The PL system is definitely still a thing though, it’s shown in every workout and is still tied to the workouts difficulty.
The ai might not use it to pick workouts but it would not make sense to have, for example, a 5 be easier than a 4 and I highly doubt that to be the case.
“Your post was flagged as inappropriate : the community feels it is offensive, abusive, to be hateful conduct or a violation of our community guidelines.”
What you’re doing here is certainly an edge-case scenario, and I don’t know if we’d recommend doing workouts this way. If you’re ready for a much harder workout than what’s prescribed, we’d recommend finding one via Workout Alternates or somewhere else.
Chopping workouts up like this likely (and clearly) won’t bring as good results.
I understand that, from your perspective, you overachieved and were awarded a higher level than what the workout originally offered, and that should be taken into account. I can look into that, but something tells me that the amount that you’ve changed in the original workout is causing an issue with how we’re processing it.
Could you send me your TR username so that I can dig in a bit deeper?
I’ve been wondering if TrainerRoad AI could (should?) go through all workouts to reassess/update the Workout Levels. It does feel counterintuitive to be given workouts with a lower WL than previously achieved, and if the reason/explanation is that WL’s aren’t fit for purpose anymore, then surely it makes sense to update them?
I think the WLs are fit for purpose…. but mainly because their purpose isn’t really important.
WL are a useful tool to quickly compare workouts and to roughly guage where your recent training is at.
But the AI predicted RPE and the new NP distribution charts are far more powerful tools.
That said, most of the time where athletes are not seeing their workouts progressing to higher WLs it’s because the AI thinks they are tired…. not because of the deficiencies of the WL system.
And let’s also remember WLs are the same for the whole population, regardless of what you in particular have been doing lately, whereas the predicted RPE is specific to you and thus a much better guide.
I can easily find a Hard Start threshold workout with a lower WL than a similar length O/U, that the predicted RPE tells me I will find a lot harder - because I’ve been doing a bunch of O/U stuff lately and haven’t done any Hard Starts in well over a year.
TR has everything it needs with projected failure rates to completely revamp the WL system and make it more accurate.
But that’s not always the reason for workout regression. I did run into a situation where despite increasing the intensity on a V02 ride and only marking it hard, TR recommended the same one the following week because the next eligible V02 workout in line was much harder and was still not recommended (there was nothing in between).
