When I’ve workouts on the calendar that are ‘Not Recommended’ it drops my AI FTP Prediction by quite a bit. Is that because the system is assuming I’ll fail them or lower the level considerably?
1 Like
Or its assuming you’ll pass but it won’t let your body recover enough to get the vital adaptions to enable the previously predicted AI FTP increase.
2 Likes
No it’ll be because it is too much stimulus whether you complete or not.
1 Like
Can you explain what you mean by this?
I think this is something worth digging into, both in training philosophy and physiologically. I am firmly in the “leave one in the chamber” camp, but I also think it is ok to periodically challenge yourself with a hero workout. Of course this is dependent on age, training history, race goals, etc. but a difficult workout, as long as you recovery properly from it, can be more valuable than just constantly training in a zone where you never really challenge your limits.
1 Like
That’s the key unknown, just completing something doesn’t mean you’ll recover well enough for the rest of the week, it could put you in a hole for a long time.
I would assume the ai has simulated the outcomes and seen that this is the case for this kind of athlete.
As pro athletes know, only a small amount of really hard work is needed to stimulate gains.
Shouldn’t adaptive training adapt the workout to something else?
I’m seeing “not recommended” on all of my supra long thresholds.
Yes. Is there a pin icon next to the workout?
1 Like
None are pinned.
I thought this was a bug earlier. Maybe @eddie remembers it.
I thought it still gave you the option to ride, just they were non reccomended and it was up to you.
I’m also curious about this. I’ve been self-selecting “Not recommended” sweet spot rides of late, but I can’t really see any issue with them. It’s normal progression, e.g. from a hard Penygadair to a very hard Jemez 5 days later. The Jemez had close to 5% failure prediction which was way off. If anything, I did it on a bad day and still didn’t fail. My next sweet spot is, again, not recommended, Saint Marys. Still normal progression when you look at the workout. That’s on demanding sweet spot apprach. Doesn’t trigger yellow days, too. If anything, when I had AI FTP, it would push me to more intense sweet spot workouts which actually felt closer to failure and were hard to recover from, yet they were recommended. I can’t get the hang of the philosophy behind this
If the data shows the AI that one route has better outcomes than another, then is it worth abandoning philosophy and following the better route?
I suppose you could think of it like SatNav, when it offers a faster route.
You’ve maybe better described my question. I’m seeing ‘Not Recommended’ and they aren’t followed by yellow/red and they aren’t significantly harder, so doing those workouts actually cuts in half my prediction.
In the screenshots, one is doing a 3.7 today (not recommended) and then other is a 3.6 today. As far as I can tell, nothing else changes. My guy tells me that the system is applying some lesser credit to the future hard workouts, like the quality won’t be as good with the ‘not recommended’ tag.
3 Likes
That’s a wild difference in AI RPE prediction you’re seeing there… Maybe TR staff should take a look at these screenshots, they don’t make sense to me. @Caro.Gomez-Villafane
Every ‘not recommended’ workout you have is a pinned one. Unpin them and let the model give you something it does recommend.
2 Likes
That’s not the point. They’re pinned precisely so that the model doesn’t change them. The observation is that they aren’t so much different than the recommended ones (speaking of the first threshold), but the adverse effect on the FTP prediction is huge. Also, the first workout affects the RPE predictions on the following pinned ones in a dubious manner.
2 Likes
Exactly. I’m not actually sure if I pinned them anyway.
Ah, OK I see now - sorry, hadn’t read back properly.
But - isn’t the big change in FTP prediction just a compounding effect of doing four ‘not recommended’ workouts before then? As in ‘this workout is not recommended as it will cause too much fatigue’….. then you repeat and repeat, so the model thinks (rightly or wrongly) that each one will become even more ‘not recommended’?
Personally I think the WL understate the difficulty delta between Blackcap and Cloudripper -3, given that the ‘overs’ in the former are at 110% rather than 105%, but it’s worth getting clarification as you say.
They are not “not recommended” with the original threshold workout, and there are no extra yellow days (or red ones). In any case you’re doing Fishers 4 days after either Cloudripper or Blackcap. The RPE shouldn’t be dramatically different, even more so for the other workouts that follow.
I would agree here if the yellow/red was more obviously different, but it seems like it would be pretty minor, or especially if it was over a few workouts made harder, but not quite to the level of showing additional yellow days.