I’ve had 6 months off the bike. Combination of poor health and losing my mojo. But now am ready to get back in the saddle.
My FTP last summer was 284.
With the lack of training I expected ftp detection to drop me loads… but it suggested 280. No way that’s right…. Pettit this morning was consequently too hard and I dropped the intensity in ride by 10%.
Post ride… no adaptations suggested???
Do I have an unfair expectation here? Or does that feel wrong to others as well?
Yes I can do a ramp test to rebase…. But I guess I assumed it would deal with this situation better than this?
There’s no magic dust here…I would be surprised if any AI fitness model could be accurate when missing 6 months worth of data and a metric ton of unknown variables (e.g., what you actually did during 6 months off - cross train or couch tater).
It’d be interesting to know what it actually does in that case - I’d suspect it would need at least a week or two of current workout data to get anywhere close to accurate.
Xert actually tries to model this through it’s “Fitness signature decay” (the fitness signatures contains more than just threshold power). I agree, that after 6 months it would probably not very precise (to say the least). But assuming that threshold was the same a 6 months before doesn’t seem right, either.
I’m coming back after time off the bike from moving and lack of mojo. My break was 2-3 months (2 months no riding, 1 month of not much riding at all). AI FTP seemed pretty good for me on my return, but there was obviously less of a gap.
To your question, I would recommend just manually drop your ftp to your best guess. With the post ride surveys, adaptive training should then pretty quickly get you into the right levels of workouts for your fitness.
I agree that a ramp test seems the right thing to do - and am a little familiar with the product so am happy to do that. But the TR guys are trying to sell AI detection as “never do a ramp test again”… so I don’t believe my expectations here are that unreasonable.
Where my thoughts were going - for clarity…
if you had 2 weeks off - it would be reasonable to expect the AI FTP prediction to reduce FTP a bit
if you had a month off - it would further be reasonable to expect a further reduction… probably more than a 4 watt drop on a 284 starting FTP
Atfer 6 months - I was only recommended a 4 watt drop - which feels too low to me… I’d have expected it to get to the end of its predictive envelope - and either stop at the max drop it had worked out (which would have been more than 4) - or to have just declared that “it cannot calculate across such a gap - please do a ramp”… It didn’t do either… didn’t throw an error. Just said - yep - we’ve looked - and we reckon you are a 280.
And then after a difficult pettit with a 10% manual in ride intensity reduction - for FTP detection to still not spot something has shifted - again seems odd to me
These seem pretty basic issues in the product - and ones that could be easily fixed… Especially having launched at New Year when I guess a lot of people are looking at getting back into a plan?
Reading between the lines a little I feel that this may be an issue that may be addressessed when outdoor rides are better accounted for by Trainerroad?
Like @Larzi said above, other platforms predict fitness decay fairly well and i’m sure Trainerroad could do the same. But I get the feeling, at the moment, they my have been forced to add a “floor” for the decay to account for people who may take 6 months off from strtuctured training with power but not from riding all together? Just a guess.
I think that AI ftp detection would sort itself out fairly quickly but I think realisitcally a ramp test would be the best course of action for you.
As for if you have an unfair expection? I think yes and no. I’m not surprised that FTP detection struggled - but if that is the case I think that you probably should have been presented with some kind of warning?
Or actually, it just could be that according to the data the AI has been trained on - most people with 6 months missing data have not actually stopped training all together? So that why your FTP hasn’t dropped much?
They’ve introduced a 28d validity thing to the AI FTP D although you dropped intensity straight away, having recently done and a AI FTP D at 280w, it won’t compute a new AI FTP D for 28d’s. If you don’t want to do a ramp, just carry one dropping the intensity by 10% or what feels right, complete you surveys and let the system dial you back in.
This sounds like an “easy” fix. If no activities are detected in the previous X days, then refuse calculation and suggest ramp test.
Unless there’s good science produced on an average of FTP decay following a given extended time off (which I’m pretty sure there isn’t as there would be way too many confounders in any decent sample size).
I think, in this case the focus should be on the software not the user. IMO TR should disable FTP detection after a certain period of inactivity. Maybe 2-3 months? Then force the user to either do a ramp test or manually enter an ftp. Only after x rides post inactivity (10-14?) should TR enable ftp detection