Don’t know if this has been answered already…almost 4000 posts are hard to go through
How does AT know with the surveys if a person is just a whiner (lets say passing workouts but always saying it was too hard) or at the other end of the spectrum, somebody who says everything is easy but just barely passes the workouts?
How does one “calibrate” perceived exertion to guide the AI correctly? For example: a good sweetspot workout should be hard or moderate or very hard?
By the way I also got negative progression on a passed workout too.
Hopefully the machine learning learns this. So if AT thinks a workout should be “3. Hard” and someone rates it “1. Easy” and does so consistently then it should be adjusting the input accordingly.
There’s a post from @IvyAudrain either in this thread or the Easy vs Moderate vs Hard thread about what each level should indicate.
It really sounds like TR doesn’t have an internal FAQ / knowledge base for its support team on AT. So there isn’t a common understanding for how AT should work.
I work for a SaaS company, and with every release / introduction of a new product, our product managers do TOIs (transfer of information) & KBs for our support team, so support understands new product / features and correctly and consistently can respond to customer questions / issues.
Hmm, just did Tunnabora - Log In to TrainerRoad, should have been tomorrow but decided to do it now. Got the “pass” survey, gave “Easy” as the answer, well because it was. Was spot on with the power figures and was pretty consistent - that road was resurfaced recently and there’s a “join” at the top - I was within two bike lengths of that on each interval.
The Calendar shows -1.7 for Sweet Spot PL but there’s nothing on the workout page linked to.
No workout scoring starting March 1st as I disabled TP-to-TR workout sync on Feb 27th. Re-enabled on April 27th and workout scores reappear but no PL for custom workouts except those in Jan/Feb. So I suspect that generic outside workout support is getting closer.
This is also a self correcting system to a certain extent. If you say a hard workout is “too easy,” and AT progresses you to the next workout, the chances are higher you will struggle or fail. Once you struggle or fail, it will lower your progression hopefully to where it should be.
If you are sandbagging and saying easy workouts are “too hard” then obviously your progression will stop or go backwards and you only be hurting yourself. Hopefully in your boredom or the easiness of the workouts, you’ll change the way you respond to get the progression going again.
I can confirm, when viewing the ride analysis page, you may see a false decrease in Progression Levels if you completed a ride below your current level. To clarify, completing a ride below your progression level should not actually drop your progression levels.
Those false decreases can be seen on the ride analysis page, but are also displaying on workout cards on the Career and Calendar pages, not just on the Past Ride Details page.
This false decrease display won’t actually change your Progression Levels, and you should still see accurate PLs on your Career page.
We definitely agree, and recognize that misinformation was given out in this scenario. Our team is documenting items daily to educate everyone internally, support team included, on the nuances of Adapting Training, bugs, how to speak to them, and better create consistent messaging for each AT item.
This current report has now been distributed to the team, and feedback like this gives us great examples of where we can continue to improve.
Fixing that bug should fix those false progression displays of all kinds. Until then: your Career Page is the best, most accurate place to view your Progression Levels.
Thanks @IvyAudrain. I’ll log this with support also, but it appears there are now lots of negatives showing on the calendar:
This is happening in both the web-client and also on my Mac.
Yep, that aligns with what we’re showing on athlete’s calendars as false decreases. Ignore those until that fix is released! So sorry for the confusion.
Does increasing the percentage of a workout from 100% to 115% effect adaptive training at all to do with a larger progression? Or should a person just choose an alternate that has a higher progression if in fact the rider can manage it?
When I look back through history in my calendar, I see completed workouts labeled achievable, productive, not recommended, etc. Are these labeled in the context of my current fitness, or in context of the fitness AT thinks I had at that time?
In other words, when I see that 3 months back I completed Huxley, which was part of my training plan at the time, but AT now says “not recommended”, is it telling me “the old plan recommended this, but we now know that wasn’t a good idea” or instead “just because you did it then, don’t think you can do it now”?
I’ve been following the AT forum firehose of updates since release, but only got into the beta a few weeks back. I may’ve missed if this has been covered before when I had less context to recognize what was being discussed, so apologies in advance if I’m asking something already answered.
From my calendar it appears to be based on your PL at the time of the workout. This is what should be expected I think as well - it makes sense to represent the workout at the time, not what you can do now…
Thanks. I agree that’s what I’d expect, but I could also imagine it as a current-state label applied to everything, so I wasn’t sure and thought I’d ask.
Since I wasn’t in AT at the time, it definitely didn’t set it when I rode it. I can imagine a few ways to test it to try to figure it out, which I may try if no one is able to definitively say one way or another.
I’m very certain from my history that it reflects the PL at the time of the workout. Would be good for somebody else to confirm, but I’m 99% sure that is what is intended.
