Wow that seems really strange. I would say that it seems the AI doesn’t ‘like’ outside rides. However, I did a 5 hr ride outside yesterday, and it bumped me up 2w. But maybe that was because I had 3hr group ride on the calendar and did more. Who knows.
-6 after a ride like that, seems something is off. Did you say you asked the help desk?
What is frustrating many of us is that following the plan perfectly (no failed workouts, no extra TSS) and rating as ‘expected’ still does not produce the AI prediction. There are several posts showing this to be the case.
I can tell you anecdotally that if you want to see the number go up:
Will you be stronger? No. But if you want to see a bigger FTP number with the beta AI model, that will show it on the screen.
I wonder what kind of ovelap there is between these two groups:
1.those that are having success with the new AI FTP system…and 2.those that have a history of turning down FTP increases under the “old system”
meaning this: under the old system, I would decline FTP increases from time to time…like when going from base phase to build phase. I felt like I knew what my body was capable of, and knew if an FTP increase was needed, reasonable, and within my best interest.
AI FTP is taking the reins, and “forcing” riders to train within their limits with the intention of avoiding overloading the athlete, and still maximizing time in the saddle.
to be clear, I’m not really a fan of the AI FTP, but then again, its working for me so far.
Update, think im wrong. Before my workout today, my next ftp within 4 weeks was 259.. did the workout perfectly. Spot on, at the upside of the bandwidth and rated hard as was suggested, my ftp furure number dropped 2 watts from 259 to 257. 
Did any workouts change btwn today and the Prediction date?
No nothing, it is the number just before the workout (259) en just after saving to TR (257)…very strange don’t you think?
You need to rate your workouts as they feel.
Rating everything as easy or moderate won’t simply bring you to the initial predicted FTP. That will mess up your training. I’ve seen it happen a few times recently.
Rating things as they have been predicted would likely get you closer to the initial prediction, but that doesn’t need to be the ultimate goal (although it seems like it is for many)..
The goal should be to follow your plan and knock out your workouts as best as you can. Roll with the punches and deal with what life throws at you as wisely as you can. The initial prediction is often an optimistic one that assumes there won’t be any changes in your training over the next four weeks and all of your prescribed workouts will feel perfectly appropriate. That’s not too often how things play out, which is why we change the prediction when we feel we should.
I think many people are now numb to any gains they do see each month if they aren’t as high as what was initially predicted. I totally get that, and in those cases, I think the prediction can be a distraction. I’d recommend always focusing on doing the work as best as you can first, and watching the prediction as a secondary activity. It’s just a prediction, after all..
I know there are some issues with it right now, and it’s our top priority to get that sorted out, but hang in there if you can!
When you guys sort it out it would be great to have Nate record some videos or go on the podcast with what was learned, how that influences development going forward, etc.
This is strange, you said the workout was prescribed? Did you wear a hr monitor?
What’s strange is that if I run a prediction for you after today’s workout I see a 1 watt increase from what we would have predicted yesterday…
Tomorrow is now a yellow day. What was on your schedule for tomorrow before it was yellow?
It looks like your VO2s are usually on Tuesdays. Did you pull tomorrow’s workout to do today?
It helps to know what you’ve changed with your plan to understand what may have caused downstream effects on your prediction..
Let me know your thoughts.
Yes, i did, and decoupling was within 10%.. thx for thinking with me to help…
As saturday was a long day, took Sunday as restday (some stretching), so today the vo2m training. I did the swob and to make sure, I did the check and after this, the 259 was still there, so noticed that it would have no effect. After the training vo2, (which I loved and completed at the upper range), i got the 257.
If you run a prediction (think only administrator can), and you see a 1 w increase, it is very strange that I get a minus 2. The positive thing is, that there is a difference between yours en mine, so the solution can be found (I hope 
).. thx Eddie./
Sorry should have turned on the /sarcasm filter.
Of course you should rate them accurately. I was pointing out that if the number on a screen is what matters rather than real world results then you can defeat the system in the aforementioned ways. You could also put in a lighter rider weight or modify the slope of the power meter, edit your ride file in excel with a multiplier for the power column. Hint: you can get Strava KOMs with that method by editing the time points and inflating the power to something that can realistically achieve that speed. If success is measured by a number on a LCD screen the ways are endless
Ok sarcasm off
Would be surprised if he would provide much detail. This algorithm is IP after all.
I’d think the moat is the user data they have. And they can share what they learned without giving up IP.
I can’t imagine the predictions behavior for 10% of the users was expected. Or maybe what they are seeing is the case here where the it appears the user was moving things around and maybe remembered the wrong number.
What I’m referencing is what we would have predicted your FTP to be before and after your workout, not what your prediction on May 2nd would have been.
I am seeing that you changed your training approach on Monday as well. It looks like that may have affected your future training, which would likely be why your prediction changed..
When did you make that change?
Oke, you can see just before and after the workout ftp change.
Did not change my training approach last Monday, as the ftp plan for me was not good (some weeks with just 5 hours), that when I created the Traka - plan, I set the approach a level higher.
As said, I did this to get more hours as the Traka is not a vacation ride, but a long gravel race and so, you need longer rides too (and the Traka is May 3rd and You want a taper too).
Hope this explains…
I’m a bit confused. When did you create the Traka plan?
You didn’t do anything to your plan on the 6th other than complete that workout. Is that right?
The Traka plan created some time ago so that after completing the ftp plan I could jump in to train to my tripple A event “the Traka”. Sounds logical [Or don’t you think not?]
Monday the 6th I followed the TR plan “Wild Snow”/ Vo2m training which I loved and completed add the upper side of the bandwidth. This is also why I’m not excepting the ftp proposal of 244 [my ftp plan started at 255 if I remember wel].
The ftp plan was set att moderate level as suggested by TR and discussed in the podcasts. The Traka plan I’ve set a level higher as I need more hours, ftp plan was 5-7 hours and thats not enough (I normally train 10-15 - depends on the season - hours) to gain , but I thought stick to the plan, but the result was as you see negative.
You know, I’m BY FAR not negative regarding TR, but you also see my results, I have to do something, or stop or use the new plan and hope for the best. I want to give it a go and train with TR to Traka with the plan.
Hope this gives you information. If you want we can also move on direct to net load the forum that much. Kr Jan
Ahh! But your Traka 100 plan began on the 6th, which is probably why some of your workouts were re-pulled and resulted in a prediction change.
I can dig into what exactly changed and why to let you know more details, but I do think that’s likely part of what we’re seeing here.
The fact that your prediction comes on the very last day before your A race at the end of a taper likely has an impact as well…