TrainerRoad AI FTP Prediction FAQs

You wont have a whole month of easy workouts - all you have to do is one or two level 3 or 4 workouts that the AI will plan in for you - if they are easy then rate them easy and the next workouts will be at a higher workout level. You don’t have to wait for the FTP to “catch up” in order for the correct difficulty of workouts to be prescribed.

2 Likes

Ramp test. Singular. One.

have you contacted TR Support about this? What did they say?

There is an interesting point here though - we currently don’t have any way of changing the AI’s mind about our current fitness.

Changing the FTP doesn’t do anything - the AI will just adjust the workouts down if you put a high FTP and up if you put in a low one. Its made its mind up about your fitness and will try to work around the FTP setting.

We don’t have a way to tell it it’s wrong other than doing some workouts.

In the case of athletes where they know there is some bad data going in then it would be nice to be able to just point he AI in the direction of 10 or so rides with good data and reset its assessment.

I’m not sure this is going to prove to be an issue though because TrainerRoad seem sure that the system will dial you in very quickly if you just accept the current FTP and do a few workouts - even with all the “my FTP is too low” complaints I’m yet to hear a case where an athlete has followed this advice and it not worked.

3 Likes

@JoeX can you link DM me and link me to the post you’re referring to?

1 Like

Didn’t Nate say on one of the podcasts that an interruption to a workout did cause TR to mark down that workout.

It seems odd to me because I often pause a workout to top up my water bottle or something else and that shouldn’t impact anything if done during a recovery interval.

2 Likes

What you’re saying though is “I stopped riding and I want TR to pretend I did not”. Why? Why wouldn’t you want it to reflect the work you actually did?

5 Likes

I contacted support but they have no solution

So when you said, “the AI will not generate any workouts on my plan”, they replied, “sucks for you”? They didn’t try to solve the problem? What did they say?

1 Like

Isn’t the solution just to accept the new FTP and do a few workouts? Have you done that?

RPE has been the same as predicted for all but one.

The recommended “workout” doesn’t raise my heartrate over 100. I don’t want to waste my time. TR really needs to find a solution to this.

If you’re doing 3 minute VO2 Max intervals with 2 minute recoveries that would *definitely* have an impact.

3 Likes

I don’t think this is an acceptable solution though. Nate has been telling people to modify their FTP in some cases, so I don’t believe “just accept the AI one” is the right answer unless there is some unique situation.

Having said that, I feel like we’re not getting the whole story here and that’s why I asked what they said.

3 Likes

Sorry I wasn’t trying to be being mean to AI. (or anyone else). It was meant more as a title to get attention. The FTP prediction was hyped up to be great and was really looking forward to seeing if I followed the plan though just what could I accomplish. Here are my last two weeks.

Perhaps I was calling things the wrong thing. The predictive FTP dropped and work out levels showed 4.4/4.7. This was a 2.5 hour endurance ride, I paused within the first few minutes of the warm up to fix a technical issue, and did not pause at all during the rest of the entire 2.5 hour training session. I feel as though nothing should have been docked in the slightest. And if I knew it would have been, I would have simply started over again. The AI should be able to recognize that this would have nothing to do with fatigue.

1 Like

Based on what I can see from your heart rate graph, you did die twice during the workout. Maybe that had an impact as well?

8 Likes

I think this is the key to so many of the people who are concerned about this. The math doesn’t “feel”. You learned a lesson though, and as you said, next time you’ll just restart the workout. I personally think it’s fantastic that the AI is doing the math and giving us a true representation of what we did. It reminds me of people finding out their KICKR has been smoothing the power on their interval graphs. (And yes, I do think it’s funny that I’m saying “I feel like this is better” :grinning_face:)

2 Likes

No, HRM just dropped for a few seconds. It is clear from my power that I didn’t stop pedaling. I get that the AI has to calculate their measurement from something. I would simply wish that anomalies, such as no HRM data for a few second while power is sustained, would be considered into the calculation.

Thanks for sharing.

18mins out of a 150min ride is more than ten percent, so I can see that being significant - how significant is a different question.

I’m looking at my prediction fluctuating a watt or two and contrasting that with my historic view that 249 and 250 are essentially the same thing.

1 Like