This doesn’t answer your question, but don’t forget you can manually change your FTP.
I just tried that and it has disabled AI prediction and wants me to accept the old detection value.
So now I don’t even know if it will adjust the workouts as I go as the AI is disabled. I’ve accepted the reversion back to 265 so the AI works. I’ll manually select a higher level for my next threshold and will hope the AI bumps up the workouts after that to reflect a higher fitness level.
Hey, and welcome to the forum! 
Manually adjusting your FTP will sometimes affect whether we’re able to get you an FTP Prediction, but everything else will work just fine.
As for your upcoming workouts, keep in mind that your Athlete Levels will decay if you don’t complete a workout in a specific zone after a while. 
There are a few things that are probably worth discussing with your training. Can I share some of the details here in this thread?
1 Like
Thanks for the welcome! I’ve been on TrainerRoad since Aug 2023 but not much one for the forums other than an occasional read.
As for your upcoming workouts, keep in mind that your Athlete Levels will decay if you don’t complete a workout in a specific zone in two weeks.
Yes! but was expecting to drop threshold from 4.3 to maybe 3.9-4.0 due to a recovery week. It has now dropped again to 2.9 (because I skipped a couple of endurance rides ~ maybe 3-4hrs riding across the week?).
There are a few things that are probably worth discussing with your training. Can I share some of the details here in this thread?
Yikes! Airing all my bad training habits in public… yes, that’s ok, I would love some suggestions.
In case it’s relevant:
I train for fitness rather than competition, 47yrs old, unable to sleep particularly well which impacts recovery, don’t recover particularly well, which means I can do 2 solid hard work outs per week (Tue/Sat) or 2 solid hard work outs per week (Tue/Thu) then fail the Saturday threshold ride. So, I settled on a master’s plan with the with sweet spot, threshold and VO2Max set to demanding.
I rode a lot of volume last year (~540hrs) chasing a km target which wasn’t ideal to build power. My gains plateaued later in the year, and every threshold ride was at least hard or worse. My lower quads and around my knees were constantly strained and always suffering from training fatigue
After achieving the goal, I took 2 weeks completely off the bike and was then extremely surprised but just how much fitness I lost (12 months’ worth of gains). I pushed my seat back to shift the load backwards and that has helped heaps.
This year I’ve dropped the volume, repeated base a couple of times to try and build muscular endurance via sweet spot rides but begin to plateau. I changed up the plan, so I am now doing VO2 Max / Threshold plus endurance rides. Due to other commitments and whatever else, I have skipped a couple of rides and bumped a couple back. I wouldn’t think I’ve been terribly inconsistent, but it could have been better. The last block I managed to do all the hard rides even though I bumped the last one back a day.
Most of my sweetspot/threshold/vo2max rides are now coming in at Moderate (if I’m rating them correctly).
I fuel my rides by eating beforehand (rice) and during the hard rides (dextrose as a drink) or cliff bars during long endurance rides.
My goal is simply to get fitter and be able to spin up hills rather than grinding up because I’m cooked.
Thanks for all that extra info! It sounds like you’ve got good goals and are working toward them in a way that works for you. 
Two solid hard workouts a week is great and plenty enough to keep things moving in the right direction. The most important (and often difficult) part of building a solid training plan is finding the amount of stimulus that actually works for you long term, and it sounds like you’ve done a good job of being honest with yourself and picking a solid path to follow. 
I do think that your weekly overall TSS has been pretty good since the start of the year. There were a few weeks in February and March where you had little dips, but that happens from time to time. Where I think you have room to improve, though, is your consistency from week to week with those key workouts. Some weeks you do one hard workout, some you do two, and others you do three.. I think getting on a consistent schedule that works for you will help a lot. You’re doing a lot of other riding (it looks like commutes or something short), and I’m not sure if that affects your workout schedule, or if it’s recovery or other life things that get in the way, but long blocks of steady stimulus are the best way to improve. 
Also, I just noticed that your upcoming threshold workout, Asta, is pinned to your calendar. I’d unpin it to see what we put in it’s place. I bet it’s a higher-level workout than 2.9.
One thing I’ll mention about spinning vs grinding up hills is that it can be a somewhat arbitrary goal for many athletes. The power it may require to spin at a specific cadence on a specific hill may be unreasonably high, making that goal unrealistic. There are hills in the world that challenge everyone! My recommendation here is always to find gearing for your bike that helps you manage hills as best as possible based on your current fitness. You can still set goals to get stronger, but having a gear that keeps you out of the red on the climbs is a really important tool. Don’t try to overpower the hill; just bring the right tool to keep your effort in check. 
Let me know if this helps!
2 Likes
Hey y’all! 
I just wanted to follow up on this one and a few other threads to let you know that the team just tracked down and fixed a bug that was causing AI FTP Prediction volatility. In some cases, an upcoming workout in your plan could swap to the next best workout (instead of the first best). If your workouts and AI FTP shifted without anything changing on your calendar, you were affected by this bug. The fix has already taken effect so you don’t need to do anything on your end for your workouts to be set correctly. 
As always, stay in touch and let me know if you run into issues moving forward, but we think this is going to play a major role in eliminating FTP Prediction volatility moving forward.
12 Likes
Please can you add my missing 7w to my AI FTP 
Was that bug only for detection or did it actually select the second best workout to perform?
I believe that there were some cases in which athletes did complete that second-best workout, but in many cases, things were recalculated before they were able to load it up. We check for changes to your plan often with TR AI, so in most cases, I believe this was more of a ping-pong back and forth between two workouts.
Does this bug explain why my prediction started respectable, then dropped, and stayed dropped? It was fluctuating but in a downward trajectory.
I think this bug could certainly be playing a role, yes.
I do think this is separate from the outside workouts bug that we’ve been talking about for a while, though, so if you’re doing workouts outside (sorry, I can’t remember everyone at this point), then that’s also playing a role with your predictions.
We are making good progress on that issue, though!
No outdoor workouts for me … just a high FTP prediction which gradually dropped to -2w , and 1w gain in 6 months.
I’m 6 days out from my next detection and most of the projected gains have eroded away. For me, the projected FTP has only created frustration. I’m getting pretty close to just going back to riding my bike.
I keep hoping Nate and Jonathan will address these issues in a podcast.
Just think it’s worth confirming that this was a bug in Prediction, not Detection.
Yes. This was an FTP Prediction volatility bug.
FYI - I have noticed I get less weird changes to prediction and future workout changes - if set a workout time for all of my workouts. it knowing I have 24 or 48hrs locked in between seems to stabilize things… or its my imagination.
Interesting. Rest and the right time seems key to the prediction. I swapped two upcoming workouts (one a 30m z2, the other 90m over-unders) because of time conflict. This resulted in a drop in predicted FTP 6 days from now. I swapped them back and predicted FTP returned to prior value. It really wanted the z2 ride first.
I know this thread has absorbed multiple issues, some of which have been solved, but my OP question remains … why does an FTP prediction drop, and stay dropped, despite following the plan. I’m paying TR to increase my FTP… after 6 months it has gone up 1w … was the plan not intensive enough? Have I done something wrong? Don’t say sleep/fuel unless you can show me in my activities where my performance was lacking.
Are people still doing this? I was stoked when Nate initially showed us how to do this, and moved things around to bump a watt or two in the initial prediction, but once I realized that the prediction swings drastically and the final adaptation is nowhere near the initial prediction, I completely lost faith in tweaking the settings or believing it actually makes any difference.