Hi all,
As the flaming orb of warmth is slowly returning, I’m looking for some insights into how outside workouts are going with the new AI. Indoor I noticed it to be very strict on the intervals. Even a few seconds off might already reduce the score.
Will there be a similar impact for the outside or is it going to be a bit more lenient due to the existence of inconveniences like traffic lights, other people, …?
Thanks
Maybe a dumb question, but where do I see the workout score? Do you mean like the athlete level changes under a completed workout? I never look at them because I am not sure what to make of them
I don’t have an answer to your question, but being honest, if it’s “more lenient”, then it’s not properly rating the workout imo. They’ve been telling us for years that outdoor workouts are less productive for all the reasons you listed, so I’d be very surprised if they built something into the ratings to have it “lie” to us just because we were outside.
You can find the workout levels, or score as I called it mistakenly, if you look at any completed workout. It’ll list the levels expected from the workout and what you managed to achieve. E.g. in the example from the screenshot, i got a little more endurance as I rode for a little while longer in endurance zone.
I get that it would possibly falsify the output of the algorithm, but I believe it was also said repeatedly that it’s ok the do a small stop e.g. for a traffic light during the lower power sections of a workout - for example in-between intervals or during longer endurance intervals - as that should not impact the outcome of the session.
Also the outside versions of workouts tend to be a bit less strict at least on power requirements, so I wonder what the AI makes of these ranges
Just for clarification for @sepiasaucer , the first number is what you achieved and the second number is the target. This confused some of us.
In terms of Workout Levels, outside workouts won’t get the same detailed workout scores you get indoors. There are just too many variables, so your completed outside workout will receive the level the workout had planned for you. In other words, if you do a Sweet Spot 5.0 outside, you’ll receive a level 5 for it.
However, this is just the visual Workout Level, and isn’t taken into account by the AI. Exactly as with indoor riding, the AI looks at the totality and details of what you did—time at power, recovery, etc. If you do half of what you had planned, the AI will adapt accordingly. The same is true if you go above target or add extra intervals—the AI sees this and responds.
Perfection is impossible outside. Don’t stress over the occasional stop sign, coast down a hill, or other normal variables. However, it’s definitely a good idea to choose a route that suits your workout. If you have 20 minute intervals scheduled and encounter long stoplights every 3 minutes, that’s going to impact the quality of your workout and the AI will recognize this. This isn’t the AI penalizing you gratuitiously; it’s a reflection that you aren’t getting the stimulus the workout was designed to acheive.
I think a good way to think of this both indoors and outside is that the AI is never trying to catch you doing something “wrong”. It’s just looking at what you do, assessing its physiological impact on you, and deciding what that means going forward your calendar. Outside riding is more variable than inside, and that’s not necessarily a bad or good thing—it’s just how it is.
Thank you for the clarification, @SeanHurley . From what I understand, the outside workouts will pretty much continue to work like both outside and inside used to work before the recent AI update.
Just one follow-up question, if you don’t mind: If I understand you right, the app will give me a „5“ for the completed workout of the same level, but the AI will basically ignore this and not adapt the training/proposed workouts based on this achieved level, but purely on the data provided during the workout. With that in mind, would it make sense to already account for later detrimental effects during the ride by picking a harder than initially suggested by the algorithm? Purely from a numbers game, this might then end up at the intensity initially prescribed.
I don’t think I’d try to game the system this way, personally. Speaking from experience, I usually find myself chasing the top of the power range shown on my head unit. The end result is I have more of a tendency to end up with average power above my workout’s indoor target than below, and extra fatigue is more of an issue for me when training outside than undershooting my workout target is. That might not be the case for you, but the AI will adjust to whatever you do regardless. If I were you, I’d follow the recommendations for a while, see how that works, and then adjust accordingly if you need to (though I doubt you will).
I’ve done half a dozen or so outdoor interval workouts since the AI launched. Overall compliance as scored by Garmin computer at the end of the ride isn’t great (typically 50-60 out of 100) but the workout level credit given by TR has always been exactly as planned or slightly higher. FWIW I’ll only do the intervals themselves on roads and terrain where I’m very confident of being uninterrupted by lights, pedestrians, traffic, downhill or technical sections where I can’t maintain power, etc. Usually a quiet climb somewhere. And I tend to aim for the top end of the power range (can’t help myself!). But my warm up, recoveries, any extended Z2, cool down are all over the place and often bear very little resemblance to what’s prescribed.
Realise this is already solved, and I know my outdoor workouts never hit the mark but I do like doing 9-12min threshold intervals on hills and would like to again.
@cartsman I’ve had no problem with aiftp prediction by converting 1hr endurance rides to Easy Solo rides so far, can you share screenshots of what you’ve seen with other types?
Have you @SeanHurley or anyone seen the effect of outdoor workouts, particularly Threshold, on your predicted FTP?
