Save the date!
Next Tuesday, November 2nd at 8:00am PST, tune in for a special edition of the Ask a Cycling Coach Podcast.
Nate and I will be live on YouTube with a Q&A session taking all of your questions on Adaptive Training.
So what do you want to know? Let us know in this thread or on our Instagram!
From how it can make you faster, to why it works the way it does, to our future plans for machine learning-powered training, now’s your time to ask!
See you all live, Tuesday at 8:00 PST.
Longtime podcast listener, recent user of TranerRoad (11 weeks into SSBLVII and feeling pretty good - should have started this years ago). I have been using AT, and I am one of those who is loving it. The machine learning feels to me as if someone if managing my plan based on my performance, and my survey response. I feel as if the TSS and IF is racheting up, to drive me even harder and build a rock solid base before I enter the build phase (SPB next).
I have been including outdoor social rides and the odd additional workout by looking at the SSBMV and HV plans, and seeing what would have been prescribed on that day. My question is, does AT take TR workouts added manually outside of the build planner into account when it adapts the plan?
I do sometimes second guess my survey response, but I try to answer the survey straight away after a workout, to prevent gaming AT. I feel this immediate response is a true reflection of how I felt. 5 stars.
I’m curious on how to best answer survey responses. Some intervals are brutal = very hard, but after a rest interval I’m ready for the next. Does that mean the entire workout was very hard? Would it be best to take the average feeling of the workout or the pain level after the final interval? Or something else?
Any preliminary data to share? AT vs Non-AT? Polarized vs Pyramidal? Any big wins you’ve seen on individual users? Any surprises in the data, for instance sustained power work having more of an impact on Vo2 than expected?
What is the team most stoked on from the initial results?
Adaptive training is working really well for me for cycling, but as a triathlete I’d love to hear about the plans to include swimming and running into adaptive training. Assuming this is part of the plan the sort of questions I have are: what you think you will use as the equivalent of FTP for swimming and running, will we need heart rate data from swimming, how will you deal with gradient when running, is there anything I need to be doing about data feeds now to prime the system for when adaptive training is extended to swim/run?
The reason I am interested in the running gradient issue is that I have a Garmin 935 with a broken altimeter and have given up getting it fixed as every time I get a replacement the altimeter goes in about 3-6 months. This DC Rainmaker post suggests that there is a fundamental design issue that has been fixed in the 945 with a software change, but not in the 935. The Fascinating Reason Why The Garmin FR945 & Fenix 6 No Longer Shows Pool Temperature | DC Rainmaker. Everything else works on the 935 so I don’t want to fork out for an upgrade!
I would also like to know about the interaction of AT and triathlon but mostly in terms of how the adaptations have manifested themselves. In the Sprint Triathlon plans there are a number of workouts which are somewhat akin to “race simulation” efforts (these used to be Carter and its variants but have changed in the new plans) and are followed by brick runs. I went through a whole cycle of the Mid Volume plan with AT and found that these efforts were adapting away from consistent paced efforts to intervals, which to me defeats the point somewhat.
Is there a plan to have AT look at the logic behind a specific session and to (if available) choose another effort with the same purpose?
My other question is around duaration of the adaptations. I have had a number of times when my plan has adapted a workout and input one with a wildly different duration (often up to an hour and a half from an hour). This has caused me to search through the Alternates tab and I almost always find a workout of equivalent duration to the original. Should AT look at the duration of the workout and choose one (again if available) with the same time parameter?
I’d like to hear more about the interaction between AT and the experimental polarized plans.
More information about AT and using it to estimate and update FTP would also be good.
Why is Haw Knob like no other workout? (There are no alternates for it )
Also, I have no worries about Ramp Tests anymore, but I am curious about predictive FTP.
What rider scenarios (see examples below) should AT produce useful adaptations / better plan than just following the standard plan without adaptations.
Example rider scenarios (not exhaustive):
- Exclusively follow a TR plans using the TR app, all indoors
- Exclusively follow a TR plans, but use a mix of TR app / head unit, combining indoor and outdoor riding
- Follow a TR plan using the TR app indoors, but supplement with “free” outside rides. E.g., LV plan + 1 or 2 long outdoor rides on the weekends
- Follow a TR plan using a combination of the TR app + head-unit (i.e., mix of TR workouts indoors & outside), but supplement with “free” outside rides. E.g., LV plan + 1 or 2 long outdoor rides on the weekends
- Not following a TR plan, but do rides using the TR app
Same issues w my 935. I’m on my 4th warranty replacement unit and other than that it works great for me. I’d love if the Tri plans had more useful run and swim options but I remember a podcast where they mentioned that is a ways off.
I have embarked on a Low Volume Criterium plan as it’s my first real attempt at indoor training. On top of the training plan, I typically join 2 weekday group rides (approx 90tss each, as assigned by TR, yes I have a power meter) and 1 longer weekend ride around 150tss. When I signed up for this plan I put a fake A race in December. I don’t actually plan to race but figured I’d be in top shape for my birthday
I see myself riding indoors more in the winter so might want to change to a medium volume plan. Can I start a new plan now without losing the “learnings” that AT has already made while I’ve embarked on this journey? Does AT keep a history on me so I can do this? Or does it not even matter because AT will adapt my plan based on my next few rides?
Biggest question is how to answer the survey questions. On a three interval, over/under if I can make the first set and feel “moderate” the second set is “hard” and the third set I need to drop 5-10%. Is that hard, very hard or all out. In other words do I average my perceived effort for the entire ride or use the last interval as the guide? I applaud TR for bringing AI to the training process. For TR next enhancement it seems like TR could monitor my heart rate (is is low, moderate or over heating vs. my past rides of the same style, then the software could suggest a difficulty ranking for the survey.
The general consensus is: rate the workout. Some workouts, like Pettit, are never going to get a rating other than easy or moderate unless you are ill. Similarly, unless your FTP is way off, a workout like Mary Austin is always going to be hard or very hard.
In one of the many long AT threads @mcneese.chad posted a table of how a workout should feel for a given response. For interval type workouts I’ll give it -
moderate if I think I could do most of the workout again.
hard if I think I could do one set again.
very hard if I could complete the workout but wouldn’t want to do another set/interval.
I’m very endurance based, a workout like Baxter for me is very easy, other than simple discomfort of sitting on a trainer for long periods (and the boredom) I could do it all day and my HR would probably max out in mid Z2 - I’d do it regularly and the aim was to keep my HR in Z1, the best I managed was just 23 seconds in Z2!
HR is a tricky thing to monitor in a consistent manner as it’s affected by lots of things. In addition you have effects like HR drift - over the course of an hour’s workout doing intervals my HR will rise by about 10-15bpm.
Does powematch negatively interact with AT at all?
Does one sided power negatively interact with AT?
Does AT recognize differences in power meters? I.e. pedal based outdoor but smart trainer indoor? Or one sided outdoor but dual sided indoor?
I have tested that my imbalance measures close to 10 pct but is not stable through a power curve. Moves from 5pct to 10pct. Do these kinds of power imbalances negatively impact AT?
Forgive me if I’m not asking the right type of question for this thread, or if I’m asking in the wrong place, but I’m wondering about manually changing workouts.
I want to add a V02 block to my training sometime next spring (A races are end of May and end of July). Steady State TTs are my jam, and I need to work more on my weakness.
What happens to my Plan Builder/AT plan and levels if I pick a three week block of my training plan and replace the steady state workouts with V02? Will AT get messed up since I’m not doing what was planned, or will it go off of what I’m doing, not necessarily what Plan Builder says I ‘should’ be doing? Or, will it work like a dream and suggest progressively more difficult V02 the following week as my levels increase?
I don’t want to mess up an entire season’s plan just to get some extra V02, but I know that such a block is necessary for continuing improvements.
Is AT now modifying “stock” plans (not part of plan builder)?
I had been away from TR for most of the spring & summer and came back recently. After I activated AT and did a ramp test, all my PL’s were at 1.0, which I understand is what happens, since it had been so long since I did a structured workout. I started picking some workouts just to get used to being on the trainer again, and looking at what I thought were achievable workouts so that when I added a Plan Builder plan, AT wouldn’t have to start working me up from 1.0’s.
I was referencing the “stock” SSBLV1 plan as reference, but I noticed that the plan it showed to me in the overview (Log In to TrainerRoad) changed as my PL’s changed. When it was all 1.0’s, the first 2 workouts were 4.0 SS and Thresh (Tray Mountain -2 and Starlight -2)
After I did a 4.0 Threshold workout (Thresh 4.0 / SS 2.8 secondary), the first 2 workouts were Carson -3 (SS3.1) and Temple (Thr 4.2). After I did a 4.5 SS workout (Geiger), the first 2 workouts are Eichorn (SS 4.7) and Temple. Again, these were just workouts I picked out of the library to try and get a sense of where I was at, in terms of capability. And I’m only looking at the plan summary. I’ve not added any plans to my calendar yet, though I plan on doing that today (Sunday).
So I come back to the original question: Is AT now adapting base plans that are not part of a Plan Builder plan?
Thanks a lot for the podcast & I’m looking forward to seeing if AT will help me not hit a brick wall in the 3rd week of sustained power build LV like I have done in the past.
Excellent insight, thanks.
Yes, it was in the latest softwares release notes (if you have ios or android should be easy to find)