Anyone else find that TR’s workout levels can be pretty inconsistent?
Today I rode South Twin +1, which is VO2max 6.5. I had to backpedal 30 seconds during the last interval, so after my workout I went to Workout Creator and modified South Twin +1 by adding a 30 second break at 10% FTP 2 minutes into the 4th interval, to try to get a sense for what level the workout really was for me since I had to break up the last interval. And TR is rating the modified workout at VO2max 6.8, even though it’s a hair easier.
Likewise, Phoneix -3 is Endurance 5.5, but Gibraltar -3, which is basically the same workout but 15 minutes longer, is Endurance 5.1.
Seems like the AI that creates workout levels needs more tweaking.
User created workouts are known to give mixed results of Workout Level. Some are reasonably accurate while others are very misleading and incorrect. I would not reference user created workouts in the way used here. TR knows about and plans to address this issue at some point.
For other direct TR workout comparison, you can notify TR directly if you find ones that seem incorrect. They have fixed issues like that when some errors are identifiied.
The level for Phoenix -3 was too high. We updated the workout, which dropped the level from 5.5 to 4.7.
It’s now 0.4 points lower than the 15 minute longer Gibraltar -3, which makes far more sense.
Thanks for bringing this up! We appreciate so much when athletes help us in our vision of constant improvement.
If you see more of these, be sure to let me know so we can take a look. Cheers!
@IvyAudrain The endurance time space continuum is still messed up. Just start with looking at all workouts endurance level 4 and 5 with duration 1:15 1:30 and 1:45. Sort by level and you’ll see some misalignments.
I have noticed some workouts seem have have PLs that seem out of place. For instance Muir seems low as its a endurance 3.3 a 60 minute workout with 50 minutes at 70%ftp , but Bald is a 3.6 with 2 × ~15 minutes at 70& and 15 at 65%
Hunter -4 & Hunter -5 are almost identical workouts with the difference being that -4 has 60 seconds of rest and -5 has 75 seconds. You would expect that -5 would have a lower PL but it is 4.8 while Hunter -4 is a Sweet Spot 4.7.
I have seen a couple others recently that have seemed out of place but cant remember what they were
Additional errors pointed out in that thread with some open errors here:
Tagging @IvyAudrain or @SarahLaverty to the rescue for this and the other thread (edit: all now merged into this thread). The more I see those problems the less I’m confident in the actual WLv1. Definitely time for WLv2 release.
The reason for these inconsistencies is due to how warmups and cooldowns are considered compared to the rest of the workout. This is more commonly noticed with workouts that have ramping or unorthodox warm up or cool down structures, but in most cases, the differences aren’t significantly impactful in what workouts get recommended to you.
All of that said, WLv2 will handle this differently, so we can expect that to change when it is released.
Gibraltar has a higher TSS & IF and is .2 points lower, but then Berkshire is a .5 points lower at 4.8
Maybe there is an argument that the level of Virginia could/should be higher than the level of Gibraltar -3 , since you do longer intervals @ 75% (I don’t necessarily agree, I think Gibraltar -3 is a much harder workout). But Virginia and Berkshire seem very similar to each other; Berkshire even has you working at 75% ftp for longer than Virginia does, yet Berkshire is .5 levels lower.
Maybe this means that virginia and gibraltar -3 should have their levels flipped, or perhaps it means that virginia should sit in between Gibraltar -3 & Berkshire…i dunno – but definitely caught my eye when i had these two rides scheduled for me in my plan b2b this week!
Hello! More to-do about Phoenix -3, which is on my calendar for Sunday. I noticed that the INSIDE workout has the correction that was mentioned here to be adjusted to 4.7, but the OUTSIDE workout is still at 5.5. I’m guessing that’s not supposed to be the case.