Just to be clear I know the wahoo outside workouts work just wasn’t sure if they worked with adaptive training.
If you can’t do structured training, what benefit would you get from AT?
Like in your example, the software could note that you spent 12 min in zone 4 and then recommend you do 15 min in zone 4. If you can do 15 min in zone 4 (random example of filling the PL), then why couldn’t you do a structured workout to begin with?
Then we’re back to no.
1 Like
I can see the challenge around interpreting the adaptations/stress from a completely unstructured group ride. That’s really complex and some many variations are possible. Time is zone would be a simple place to start (and better than not counting things at all), but a set of structured v02max intervals is going to create different stress than the same time @ v02max over the course of a group ride with a bunch of other zones are being hit and the vo2max time mainly non-contiguous spikes. It seems much easier to interpret and “give credit” for free form endurance and tempo work. I’d be fine if they launched with a decent interpretation of endurance and crappy analysis of harder efforts to start (and then improve it over time). I’m sure people would complain though, even it was just tagged as beta. It’s just such a glaring hole in the system. At this point, it’s not looking at those rides at all, so even an inaccurate or incomplete approach using a would be a step in the right direction. I did a long zwift ride today and I did a decent job of finding a workout that somewhat mimicked it (and ran the TR workout in the background), but even then it did progressions primarily based on the TR ride, not for what I actually did.
It’s difficult for TR because they don’t have a model…instead they have their workouts as the fundamental unit of their system.
3 Likes
I just assumed they built some kind of model to run all their workouts through to set the levels. Hopefully that wasn’t just looking at completion success rate, but evaluating things like continuous time in zone, rest vs interval balance, etc.
Even if they did build a reasonably smart/flexible model, it’s a whole mother level of complexity to apply that to free form rides.
If I were to start from scratch, I would start by leaning heavily on time in zone and specific watts, then some kind of multiplier based on interval analysis. Example- 20 minutes of vo2 work would earn you more credit if you had more intervals of at least x minutes exceeding y watts. It would be crude to start, but you just keep adding variables and refine algorithms as things are figured out. I assume part of the challenge is figuring out a measuring stick for the results so the engine can improve. With the legacy workouts, they have a ton of historical data to measure against. For a free form workout, it happens once, so you don’t get clean history for the machine learning logic to play on. You’d have to measure the results based on patterns that you identify, but it’s still a challenge to understand what the outcome of any pattern is.
No idea how their model works but given it can’t tell when you over perform on a session (increase intensity or add another rep etc) and can only tell pass or fail, I think they have/had(?) a lot of work to do to implement it for an unstructured ride.
For me I don’t really mind if unstructured outdoor rides affect progression levels but I do want TrainerRoad to acknowledge they’ve been completed and attempt to amend future training days based on these having occurred.
5 Likes
Once you have a model, it’s trivial. Just look at Xert’s implementation.
4 Likes
It just depends on how you approached the model. Zert was probably focused on a model for unstructured rides from the start. If TR built a model around legacy structured workouts without a strong eye toward unstructured rides in the future, the task to support unstructured could almost be like starting from scratch. Let’s assume TR just built a basic model against legacy workouts that only looked at the zone label on the workout, the historical completion success rate, and the TSS. That might actually work reasonable well assessing legacy workouts, but you don’t have the tag or the completion history to use that model for unstructured rides, so the engine isn’t going to have much value. I’m not saying TR started with anything that elementary, but just using the hypothetical to make the point.
1 Like
This is a nice summary of teh difficulty of the problem and it is interesting to hear that even regular outside rides are difficult to accurately quantify.
For me, what I would want from AT and unsctructured workouts is probably a simpler problem. If I go out and dig a HUGE hole on a group ride, it would be nice if AT looked at something like total time where W’ is below zero and consider the effects on the next workout. For example, it I kill myself on a Sunday winter group ride, it would be nice if AT said “You know that SS ride on Tuesday, let’s make that an hour of endurance instead”.
4 Likes
I agree. But also I think 4-5w/kg is 90% genetics that most of us will just never get to even after a decade of structured training.
To some degree the answer is already “yes.”
I do low volume plans, and add workouts either through “train now” or simply adding Zone 2 or sweet spot workouts.
Those don’t “technically” get AT. But when I reap training benefits from those added workouts, I tend to rate the scheduled V02 Max or threashold workouts as part of the plan as “moderate” instead of “hard.”
Then AT adjusts my future workouts, making them harder, and this giving me “credit” for the added workouts.
So, while AT doesn’t directly address the added workouts, if those added workouts make you stronger, AT will “see” it on the planned workouts.
I’m here just to second the idea that outside unstructured rides don’t need to impact PLs, but I would like it to be taken into account in not serving me up a heavy Vo2 max session the day after a 300 TSS outdoor ride.
3 Likes
Maybe I don’t fully understand PL’s ( very possible) but why wouldn’t you want unstructured outdoor rides to impact PL’s? PL’s are the result of your training (both volume and type), so if you exclude some of that training from the data set, you aren’t get a wholistic view of your training.
1 Like
it certainly doesn’t hurt to have unstructured rides impact progression levels. But for someone like me, doing the equivalent of a level 9 endurance workout outside unstructured isn’t going to have an impact on the training recommended to me by adaptive training. My sweet spot workouts, at the moment and as far as I can tell, are not impacted by the amount of endurance work I do. From my reading, what folks really want is for TR to adjust intensity of structured workouts based on non-plan volume. I do think some folks have an interest in not having their progression levels decay too much as well, because then they get way easy workouts when they get back to structure, but there are easy workarounds to that if someone is self-aware enough
2 Likes
Yes, but better to have the ride considered by AT than not considered. Am I right? I want my PLs to match as close to realty as possible.
Good point. The mentality throughout TR is that outdoors riding is not training. Recognising outdoors riding has got to be the 20% that takes 80% of the effort.
TR’s fundamental question in Plan Builder of ‘how many hours a week do you train’ does not specify whether this is ‘indoor training’ or ‘indoor and outdoor’ or ‘structured and unstructured’, and so many users will be interpreting this question so many ways, and ending up on the wrong sized plan.
2 Likes
Oh, I don’t think that is true at all. They clear believe that indoor training is superior to riding outside, but not that riding outside is not “training”.
I would say they rank it as follows : Indoor training > Outdoor TR Workouts > outside riding
The key difference that is often unstated is Structured vs Unstructured. That can be applied inside or outside, solo or groups and such.
TR places Structured on top, but that’s because of their primary focus is on Training. TR knows that there is certainly value to Unstructured riding and it even has training benefits among others.
It’s about relatively, not absolute worth or waste.
4 Likes