McKinley is a 2hr30min workout with 5x1 VO2 followed by 2hrs of Sweetspot and high-Tempo. Due to all of the continuous time in high-Tempo and Sweetspot, it is classified as an 11.3 Sweetspot workout (wowza!).
Time in Zone from McKinley
Coasting: 0 (0%)
Z1: 19min30sec (13%)
Z2: 7min30sec (5%)
Z3: 85min0sec (56.7%)
Z4: 33min (22%)
Z5: 0min (0%)
Z6: 5min (3.3%)
Z7: 0min (0%)
Time in Zone from your group ride
Coasting: 14min33sec (8.3%)
Z1: 44min22sec (25.4%)
Z2: 43min55sec (25.1%)
Z3: 18min49sec (10.8%)
Z4: 9min28sec (5.4%)
Z5: 12min35sec (7.2%)
Z6: 12min41sec (7.3%)
Z7: 18min30sec (10.6%)
Although your group ride has an IF of .83 and McKinley has an IF of .84, you can see how dramatically different the composition is between your group ride and the workout. I would not have associated your group ride with this workout as they are completely different.
The implications on your Levels thereafter are, as you pointed out, not representative of your current abilities. I’m going to have our Customer Support Team reach out to you to help reverse this association so AT can keep you on track.
This is an example of how, try as we might, solving this problem in a manual way is too complicated for us to do ourselves (ourselves = us as humans, not TR specifically). We are hard at work on AT’s outside ride analysis and its ability to assign the appropriate credit from an outside ride. It’s a top priority and we’ve got brilliant people working on it.
As a builder of models for financial markets, this is a fascinating topic (to me). …As a broad categorization you can either have a model (like say XERT), or use a lot of data and “brute force it”. It seems that TR using it’s workout library as the fundamental unit of development, has decided to go the second route and hence, no wonder why is a hard problem.
Thanks for the detailed response, it’s very much appreciated.
Obviously it’s very hard to assign a workout that might match a group ride like that so I just tried to find one with similar time/IF/TSS but I completely understand your explanation.
I really look forward to AT working with outdoor rides and not having to fiddle around like this. Would you suggest to continue trying to ride match with a workout or just leave the outdoor rides as is until AT can account for them?
I’d suggest leaving them. The ride would need to be extremely well-structured for me to associate it to a structured workout, and that is extremely unlikely in a group ride/race scenario.
I have to say I’m far more ambivalent about unstructured rides with AI FTP Detection, and those unstructured rides being taken into account for that. PL can then be looked after by my survey responses, and AT doing it’s stuff from there.
As far as I can see, when it comes to endurance progression, which would be most affected by unstructured rides, you’re adding time as you progress? I’m still not going to have more than an hour and half before work in the morning!
Are you certain that TR didn’t automatically associate the ride with the workout?
I believe that can happen even if you didn’t execute a “workout” on the head unit. TR sees a ride imported, then sees a workout scheduled on the calendar and might just connect the two.
This is viewing a ride I did and what TR linked it to. I know in this case I did an actual workout, but showing this to indicate where you can double check if TR did or did not link in your case.
For me if I have a outside ride scheduled for the day, any ride I do outside will be associated with the workout and result in PLs changing. This is the case even if I don’t load the workout on my head unit or if the durations don’t match. Outside rides are just assumed successful in the TR algorithm unless you tell it different.
Also I’ve found there’s no undoing the PL changes. I’ve gone back and unassociated a ride/workout but the PLs don’t revert.
Any update on this matter? Last update dates back to August 22. Would love this feature as well since more and more riding will start to occur outside.
