I totally get the frustration. But remember you are looking at a smooth line of erg (or that’s my assumption). The model is just dealing with the numbers it gets. I turned the erg smoothing off several years ago on my kickr. Surprised how up and down the line is even on erg. So the model is taking in all that variation. The models goal is deliver you the best workouts based on the data it gets.
As somomeobe posted above, you could always bump up the intensity by 1% if the final score is importanct to you. Or take it off erg and try to never drop below the target watts (hard to do!). In the end, I don’t think any of these tricks or worries about the final number is worth the stress unless you think there is an actual bug.
Try to look at this with curiosity and think, “wow I didn’t realize a break or power fluctuations had such an I pact. I’ll try to do better next time!”
I would look at this a different way - TR have no expectations of us.
If the data analysis shows that riders who do not pause at all when doing that workout, or series of workouts, improve a bit more than riders who do…then that’s what the predictor should show.
Like myself, who had a 53s pause after warm up before main set, if you actually did the full workout, then I would agree this would be surprising to find that AI analysis and simulation is accurately predicting lower physiological adaptations - but that might also simply be the fact of the situation. Maybe our willingness to pause to pee, or proclivity for mechanical issues does correlate to lower performance gains - it would be surprising nevertheless.
When we tell you that you can run AI FTP Detection, you can choose when to run it, so putting it off will push that window down the road, as you mentioned.
If AIFTPD is available to me now, but I want to run it on Monday, the next 28-day window should start on Monday.
That’s useful context to the single ride we were discussing.
What kind of variation in FTP have you seen in previous years, how does that compare to the current prediction and what was the drop corresponding to the ride with the 18min pause during warmup?
There are clearly better ways to get our attention.
You have very little recent training history, and your FTP went down from 258 to 256 in December. 249 does not look like a bad estimate for you, given your recent workouts.
I am not upset about the initial drop, was kind of relieved. Coming off injury and holiday break was totally understandable. What I am questioning is the Predicted FTP. After the update it predicted 30% (301 FTP) increase at the next FTP adjustment. That workout was hard, but I finished it and stayed close on the prescribed workout. After that workout FTP was dropped 10 points, bummer but understandable. Workouts since then have been labeled as prescribed and followed yet still losing predicted FTP (289 now). Am I supposed to do more on off days. Less on off days. Change my plan, and if I do to what. Ok, I lost 2 points this last ride, why? Because I marked it as moderate? Because I backspinned after a interval.
I would like to turn it off visually from the Career page, but keep it running in the background. Is this possible?
I originally thought it would be motivating, but I actually find it hugely distracting. I have followed the plan and completed every workout since being added to the Beta, and I have seen my predicted FTP jump around, up and down, by 15+ watts. I just completed 90 min threashold session of over-unders with a 100% perfect score on Workout Levels, and somehow my AI FTP Prediction dropped 2 watts.
My FTP history has a very slow, but steady increase, with small 1-3w drops at times. In the most recent two years I have gone from 272 to 307. In that time my volume has been mostly steady with a 20% increase recently. With that 20% volume increase, I have dropped my third day of intensity to help absorb the volume, but will likely turn that third day of intensity back on in a few months.
As I said, I am truly not super concerned, but I do think that these numbers bouncing all around sews doubt in their accuracy. Today, I did a 90 min Threshold effort, with 100% perfect workout level scores and afterwards my PFTP dropped another 3w. I currently have it set to 307 and since being added to the Beta I have seen it has high as 323 and it’s currently predicting 306. I have completed every single workout without failing.
Thank you! And just to clarify, when toggling it off, it’s still working in the background, just not visually seen front and center? Will it still be visible on the Calendar? If you want to turn it back on, how would you do so?
I believe that it’s just turned off until that detection day. You’ll get an AI FTP Detection on that day, and you’ll see the next prediction on your calendar, which you’d have to turn off if you don’t want to see it.
So essentially, it just turns off that specific prediction. All future predictions will need to be turned off in the same way when they become visible for the time being.
I think that makes sense. So just to clarify. It will no longer show up on the Career page and bounce around day to day, or show up on the Calendar until the next prediction. At that point, it will predict again. At that point, it will need to be toggled off again.
So, this would basically behave the same way it has with the previous version.
I’ve come across something that I’m a little confused by.
I’ve put in my schedule based on the amount of time I have available to train - 1 hour Wednesdays, 1.25 hours Fridays, and 2+ hours on Sundays.
So far, the AI has been assigning me v02 work on Wednesday, Sweet Spot on Friday, and Threshold over-unders on Sundays. However, I noticed that in coming weeks it’s starting to recommend longer Sweet Spot workouts on Fridays since I’ve maxed out the level for that duration. Here’s an upcoming week for example:
I’m a curious as to why it wouldn’t then switch the Sweet Spot to Sunday since I would then have ample time to do longer Sweet Spot workouts, and move the Threshhold to Friday since I’m not yet maxed out on that zone?
It’s not a major issue per se since I know I can always change this manually by switching the Sunday workout to Sweet Spot and vice versa (or ignore it altogether) but I’m curious about why it would default to trying to shoehorn a longer workout into a day with less time available when another option is available.
It doesn’t know anything about which day of the week you have more time on. All it is thinking about is getting you the most productive workouts on each particular day. Your sweetspot level is fairly high and it’s getting to the point where more time is needed to progress you in that zone. It can’t make it much more intense without making it more of a threshold workout rather than sweetspot.
All that said, I’m in a similar situation and would like the option to switch my default sweetspot and threshold days….
Thanks, that makes sense. I had assumed that since I had entered the time available each day in Plan Builder that would be taken into account in planning the right workouts (ie. longer workouts on the days with more time available). It might be lightly less than ideal but again, not a major issue since I can always swap them manually which seems to work well.
I know you didn’t mean me and I’d never expect that. But isn’t it funny that what is so obvious for us … shouldn’t the smart AI know also . I’d prefer the AI prediction be more conservative and therefore surprise me more often with an uptick.
I’ve lost track of when the comment was made exactly, (past few weeks) or whether it was posted in the beta thread or not, but I recall Nate commenting that pauses occurring specifically during the warmup or cooldown of a workout were going to be “downgraded” in terms of their impact on the AI’s overall assessment of workout performance. This might be relevant to some of your issue here…
I don’t know if this update is still planned, work-in-progress or has now been implemented… Maybe @eddie might comment if he has any further info?
. I’d prefer the AI prediction be more conservative and therefore surprise me more often with an uptick.