OrâŚ.âHow do I manipulate my training plan to get it to predict a 300W FTP in six months?â
![]()
OrâŚ.âHow do I manipulate my training plan to get it to predict a 300W FTP in six months?â
![]()
Iâve clicked the link to sign up for TR AI which took me into the App, but that was it and I canât find it in there, about where to sign up for it?
Regards
Mike Drake
That wording/link is confusing.
They are saying if you sign up now (or currently are a member) that youâll get the feature upon its release. There is no current sign up for this feature, it is not even something that is in the current beta.
Yep.
Itâs coming soon and when it arrives everyone will have it.
Pretty sure it will only attempt to predict your FTP within a 4 week window.
I.e. your next AIFTP detectionâŚ.
Definitely could be wrong - just a guess.
I just canât imagine it being super accurate much further ahead?
Honestly Iâm tired of seeing these âadsâ and ânoticesââŚfeels like Iâm being bombarded with a fake news itemâŚjust do it or stop these endless âcoming soonâ adsâŚ
Hopefully they do some consulting with Nate Silver to figure out how to explain probabilities and confidence intervals to the masses.
I have criticized TR for not discussing features and missed Nate on the podcast talking TR roadmap stuff. So I for myself enjoy those feature teasers and after years that felt more like stagnation finally some concrete insight is very welcome.
Lightning up the holidays with some pre produced teasers to me feels much nicer than another 2 weeks silence and as a result more frustration in the speculation thread.
Looking forward for the next podcast.
Yeah this is a no win situation, you donât update people and they get mad you donât update. And if you do update people they get mad that they are just teasing. Neither side is right or wrong both are annoying depending on what side of it you are on.
The ads though on the forum for those of us being paid users is a little annoying. I did have a chuckle and almost screenshot one between 2 posts by Nate in the Beta thread, like if people are testing stuff they clearly are users and donât need ads.
Nope!
FTP can mean different things to different people.
When we do it, weâre getting you the right FTP to put you at the right progression levels to get faster.
Weâre not predicting your 60 minute power. We think that would put you training too low even at high PLs; or not give you enough room to increase without constantly increasing your PLs.
The cool news is our AI doesnât know what FTP is or what workout scores are. Itâs just looking at raw power and HR. We then back translate it to an FTP that gets you at the right insertion point in progression levels.
So, if you happen to put your FTP higher or lower than what we say, the AI is still going to prescribe you (basically) the same wattage.
You just might be doing 95% of threshold instead of 100% of threshold; but under the hood itâs still 250 watts.
Weâve wanted to go to an âFTP-lessâ world where you just train based on watts but we think thatâs too big of a paradigm shift for athletes at the moment.
Why even post this? Weâve never posted âsoonâ in marketing materials and not had something come out soon after.
Why is it dodgy?
Good, FTP is not 60m power, I donât think any knowledgeable coach or physiologist defines it as such thankfully but too many amateur athletes seem to think FTP = hour power.
Thanks!!
Because these are predictions, thereâs many branching paths it could take. IE if you perform differently on a workout than we predicted, the following workouts are all impacted.
The good news is we update it as soon as that happens so you can see!
Right now we predict the most likely path through you training. In the future, weâll do monte carlo simulations where we run the simulation many times to predict a range of FTP based on your training. Think of that one like your retirement portfolio stock prediction.
We built a proof of concept for that, but as of now itâs too resource intensive and slow to be released.
Weâve also tested out doing monte carlo simulations with different training plans adjustments (think different time, days, workouts, etc). Those are interesting but they can take hours and isnât feasible for production launch.
Our AI team is going to work on a smarter way to achieve that after the dust settles with this launch.
The really cool thing about this feature that I like is that itâs super motivating to see your predicted FTP. You can also see what would happen if you skip a workout, or a weekâŚ.that motivates me to get on the bike because I donât want to see my FTP drop.
I also saw that I shouldnât be doing an âAggressiveâ training approach. That actually overtrains me and I should be doing balanced. Thatâs also something we want to automate in the future but for this launch youâll just need to try the different approaches and see what happens. Whole process is like 2-3 mins.
The guts of this AI has been worked on for three years, and the entire company has been working on the rest of the integration for about 10 months. Weâre so excited to get this out and show everyone!
Completely different
Youâll see workouts the AI picks in your simulation window which is 4 weeks long
After that, youâll see zone/workout profile/duration on your calendar for workouts more than a month out.
As workouts get into the simulation window, they will be picked by TrainerRoad AI.
How many people are going to see a (predicted) decrease at some future point, and freak out? And then start changing around workouts to try to get to a (predicted) increase? It will definitely be interesting to see how this all plays out.
This is one of the really cool parts! If you donât workout, you can see how much your FTP is going to go down.
I think you might be thinking of other platforms where if you schedule more, your FTP will be higher.
Thatâs not the case with us. If I put myself to an âAggressiveâ training approach my FTP is about 15 watts lower in a month than if Iâm on a balanced training approach.
Aggressive progresses me faster at first but then the AI has to pull me back due to increased recovery demand.
So if you try to load up your training with more than you can take, youâll get a lower FTP than with less training.
In our experience, weâve seen a LOT of athletes select a plan thatâs more than we suggestion, and then add extra outdoor workouts without replacing their intense workouts.
This mitigates this a lot, because the AI will keep changing your workouts, intensity, and adding rest when you do too much. And then you can see how that impacts your FTP before you do it and make intelligent choices.
As Iâve said before, the AI only looks at watts and HR in your workouts; it doesnât know your FTP or progression levels.
The AI just picks the best watts for you to do; then we translate it back to FTP for athletes current mental model. Of course when your watts go up, your FTP goes up; but itâs not just focusing on FTP; itâs focusing on the type of power needed in your plan to make you faster and achieve your goals.
So I think both of your statements that itâs toxic and overvalues FTP is as a metric are false. It should reduce people biting off more than they can chew and it doesnât even know about FTP.
PLs are still there, and you should see progressions like this in workouts if you donât have too much rest between workouts or too much intensity. Itâs usually more complex than just 3x12, 3x15, 3x20 though.
As soon as possible. Team has a list of critical launch bugs that they are working through. I want to say a date but there are always unknowns that happen.