Manually adjusting FTP

I am new to TR and have a question about how to handle FTP - the automatic detection put me at 217W based on previous rides. Which would have been fine a month ago, but having been ill since then I can see that I have no chance of doing workouts based on that anymore. So what is the best way to get the workouts down to my current level?

I can decrease intensity in the SW, but I would have to do that every time.
Or I could manually enter an FTP a bit lower than the one detected. But if I do this, how will the automatic detection work? I would like it to still detect any new FTP increase, but not anything based on old rides.

I haven’t been able to find a clear answer to this, so hope I can get some help :slight_smile:

In addition: How far back does the FTP detection look at rides? I would have to go back almost 2 months to find activities that would indicate what I was capable of before my illness. That seems like a long window (or maybe it’s just the decay of FTP that is a lot slower than my old body :rofl: )

I’m not sure exactly how far back AI FTP goes but just adjust your FTP manually to what feels right. AIFTPD up to the start of September had been coming up with a number of between 274-282w, which is flattering but for me but I feel it is getting biased by the road group riding I am giving it and I’m training for TT’s so I’ve manually adjusted it to 250w. After just under a month at my manual FTP, AIFTP seems to be suggesting an increase but one that’s more in line with my manual adjustment.


You could add a ramp test and do a new AI FTP detection. It should take into account the time off the bike and adjust. You could also just do the ramp test (or any of the ftp tests) so see where you actually are and go from there.

Manually editing also works but that would require you to know your body and capabilities very well, if you have plenty of previous training experience I don’t see a problem in that. You can always change it again based on feel.

1 Like

Welcome to TrainerRoad and to the forum! Sorry to hear that you were sick, huge bummer but so many athletes have been experiencing that lately, and luckily, Adaptive Training is designed to accommodate your current fitness as you ease back into training.

In asking if you should either decrease the intensity of your workouts, or manually lower your FTP, the answer is: neither! :wink:
You absolutely did the right thing by marking your recent Sweet Spot workout as ‘Too Intense’ when it felt that way, because it gave Adaptive Training the opportunity to do all that adjustment for you by updating your future workouts and taking off a bit of that intensity to put you at a more achievable level.

@PhydomiR is right in that “Manually editing also works but that would require you to know your body and capabilities very well”, and this is where Adaptive Training really shines and takes that guess work out for you. You’ll see this too when you start feeling like yourself again and AT recalibrates quickly to make sure you’re challenged by productive workouts that keep moving you forward.

I also want to go over some options for you in the future when you’re not feeling 100%, where you should feel free to just select an Alternate to your scheduled workout if you still want to try to target the same goal of the workout but at a different workout duration, or to just grab an endurance workout via TrainNow if you really just need to chill that day.
Adaptive Training will account for all of these changes too when looking at your future workouts, so definitely listen to your body as you’re getting back into things and don’t be afraid to make changes like this.

Finally, ‘How far back does the FTP detection look at rides?’: I hear your concern that it may not be looking far back enough or fully understand your 2 months off, and while AI FTP Detection looks at your most recent training, its definitely addressing workouts beyond just a couple months ago, don’t worry.

It’s also important to acknowledge that easing back into training after regular ‘time off’ will look different than if you are sick, which is why the post workout surveys are so important when you’re struggling. It adds that context to Adaptive Training so that it’s understood where you’re at in your fitness and how to ramp you back up appropriately; and it calibrates in that way really quickly as we’re seeing in your case!
Your bases are covered when you respond to those surveys and let AT apply adaptations that reflect your fitness. Nice work answering those promptly!

If you have any lingering questions about moving workouts, your survey responses, Adaptive Training, etc., please feel free to message either myself or @sarahlaverty, we are TrainerRoad’s Community Managers and super happy to help take a look at your calendar and workouts anytime! :sunglasses:


Thanks for the detailed reply, @IvyAudrain - I did of course notice, that AT immediately dialed down the intensity on many future workouts. It was nice to see that working :slight_smile:

However, even though the plan is now less intense, the workouts are still based off of an FTP, which is too high. My worry regarding the time the AI detection looks back was more along the lines of how long time it would take for it to ignore, that I was doing big numbers in the beginning of August and start to dial down my FTP to more realistic levels.

It might just be my thinking that is a little old-school, but it feels wrong to have workouts based on a percentage of FTP and then ignore that the FTP isn’t correct. While I’m not entirely sure, what my current FTP is I know that two months ago I held 218W for an hour on an outdoor ride and today I got cracked after 15 minutes of less than 200, so something has been lost. I know, that those quick loses will also come back faster, but it still feels wrong to have that high value now.

I guess I’m just trying to understand and learn TR - I’ll let it do it’s thing for a while with AT and see how it goes…

You can also run ai ftp detection again. It factors in time away from the bike.

I ran it less than a week ago, so I don’t think that will have changed much. That was what I was aiming for with the question about how for back it looks: If I could assume it would go down after X amount of time since before my illness. But based on Ivy’s reply, it sounds as if it looks back quite far (although it will of course factor in time away and have a natural decay, but clearly not enough for my current situation…)

1 Like

I’ve recently adjusted mine down manually about 10% because I’ve been riding outside all summer, and the AI FTP is based off that. My indoor performance is much weaker (also different power meters etc), and I’m about to go back on the turbo. Even with all my progression levels at 1.0, using that outside FTP indoors will destroy my legs and confidence. I know the theory is to trust the process and wait for the system to adjust, but frankly I don’t want to waste time failing workouts when I can bypass the robot overlords. Not ideal, but not sure what the alternative is until adaptive training becomes clever enough to factor all this in.

How does the Robot Overlords react to that? Does the AI FTP detection still run and adjust you up if you improve? And only if you improve, but not based on historical data? Or is it turned off after the manual setting?

Hey! I can with some pointers but I wanted to check:

  1. Is there a reason you don’t have PowerMatch enabled for your indoor workouts? (Possible you just weren’t aware of this feature yet?)
  2. Do you feel like there’s anything in specific that’s holding you back from expressing your outside FTP indoors? What are you experiencing when you first switch from outdoors to inside?

Thank you for taking the time to reply. I wasn’t expecting that and it is appreciated!
Bike A is on the turbo all the time, power readings come from the Wahoo Kickr
Bike B goes outside and has Favero Assioma pedals.
So I believe this rules out power match unless I’m prepared to swap pedals every ride. (too much faff). As an experiment I did ride with the pedals on the Kickr once and found they were reading about 5W higher than the Kickr.

I’ve seen and understand all the advice about cooling, bike set up, nutrition, RPE etc. For me, the turbo feels like there is resistance all the way round the pedal stroke, whereas outside it feels like mostly on the downstroke. I know people will jump in here with thoughts about pedalling technique, but ultimate indoors just doesn’t feel like outdoors. I can live with and accept that, it would just make my life easier if the system was able to recognise and take that into account.

Thanks for that extra context of your equipment. It makes sense if you’re doing both indoor and outdoor rides frequently why you wouldn’t want to change your pedals every ride.
If you think you’ll be doing mostly indoor workouts for the winter, the best outcome for you would result from bringing your Favero Assioma pedals inside in conjunction with the Wahoo Kickr, with PowerMatch enabled. I get not wanting to switch pedals over often, though.

The good news is: while you think the difference is something around 5 watts, this is minimal enough that could be addressed simply by working on your pedal stroke.
I know ‘indoors doesn’t feel like outdoors’ and you didn’t want this to be a pedal stroke thing, but there is a ton of benefit to the resistance experience indoors in helping athletes form a well-rounded pedal stroke. This isn’t a bad thing for you (and all of us!) to work on, as it will only help translate into better riding outside, too! More efficiency, more comprehensive power, and of course going faster. :wink:

There’s a good clip from the Ask A Cycling Coach Podcast here that will help, but please don’t hesitate to follow up if you have any questions, and rest assured that Adaptive Training handling two different power devices is something we’re working on!

So, I just wanted to get back to the original topic here and give a status, because it wasn’t as easy and automatic as I was told :wink:

Quick recap: I had been off the bike for a while, AI FTP detected me at 217W, but I was nowhere near that. Advice was to just let Adaptive Training handle it. That was not good advice, I think.

Adaptive Training is too slow and makes too small changes to adjust for a difference as big as I had in my actual FTP versus what AI FTP-detection gave me. After 3 weeks, I gave up, changed my training plan and started with a ramp test…

This ramp test done after 3 weeks of workouts gave me an FTP of 199. Save to assume directly after my time off the bike it was probably a few watts lower, so the original AI FTP estimate was appr. 10% off. So even though AT gave me easier workouts, the intervals were still set at a level where I could hardly finish them

So my take after trying this out:

  • Adaptive Training can adjust for the development during a plan and small changes as well, but it can not compensate for an FTP difference of 10%
  • AI FTP Detection probably works fine while you’re training consistently, but it very much underestimates how much power a 49 year old can lose by being sick for some weeks. Putting me at 217 when my actual number is under 200 is a very large difference.

Some more notes on AI FTP: I just tried running it again after the new announcement and it now has me down to 204, so it’s getting there, and a difference of this size would probably not have been a problem. Also, I have all my numbers going to as well - they have had FTP detection for a long time, and at the time I started my training plan here, they had me at 216W, so it’s not only the TR algorithm that underestimates, how much I can lose :wink:

tl;dr: AI FTP is probably fine for people with a recent history on the bike, but if you’ve been off the bike, do a ramp test or it will overestimate your FTP a lot! And Adaptive training can not adjust for such large differences.

1 Like

Hey, I’m looking into your calendar and adaptations to provide clarification. Stand by!

There’s really no need for that, the numbers speak their own language. The AT algorithm makes some small changes to the plan, but still gives me over/unders with the under-part being at 105% because the FTP is wrong.

So the FTP has to be more precise than 10% otherwise over/unders doesn’t make sense. And the AI FTP detection only does that, if you have been consistent in the data it looks at.

1 Like

It’s no trouble at all! It’s really worth looking into, especially to address your greatest concern here that “AI FTP is probably fine for people with a recent history on the bike, but if you’ve been off the bike”.

Since you first used AI FTP Detection, we’ve released an update to AI FTP Detection that will change the minimum recent ride requirements to get an AI FTP Detection value. That will give us more data on how AI FTP Detection works with athletes in your situation and would have likely improved your experience and AI FTP Detection result initially. Apologies we couldn’t get it released a few days sooner! :wink:

I’m going to send a DM to cover the Adaptive Training adjustment concerns, and clarify the ways in which AT is locking you in. We’re happy to help!

TBH, I think it’s always going to struggle a bit with the first estimate after sickness, because sick =/= sick. You could be ill and lose 40W, while the next person has a different illness, or even the same illness but a different response, and only loses 4W. The AI has no way of knowing what the impact of your specific illness will be until it’s had a chance to collect some more data on what your performance looks like post-illness. So the advice should probably be, after illness, run it more frequently to see if it’ picked up a decrease.

1 Like