I’m aware that FTP in TR is a prediction and the “8D model” is primarily using watts, but getting the FTP prediction as accurate as possible is important especially if using collaborative tools like “Best Bike Split”.
Prior to moving to the new TR AI, my FTP was 250. As soon as the TR AI push, my FTP dropped to 235 with a 30 day FTP predication back to an FTP of 250.
I have a lot of workouts, including recent workouts, so the model should have plenty of data to properly seed the model. Really, the FTP should be 250, and appears TR is slowly ramping the predicted FTP from 235 to the better prediction of 250. I don’t think a 15W improvement in 30 days is real for someone that has been training, but does make the training plan powered by AI look good .
Many of my workouts are outdoors, which may impact the prediction, but seems like an underdamped FTP prediction?
I think the fact that you’re rating very low-level threshold workouts (~1.0) as “hard” tells us that your FTP (250) is likely set a bit too high.
Our workouts go from 1.0 to 10.0, so if the very easiest threshold workouts are challenging for you, you’d certainly be better served by lowering your FTP and taking on some higher-level workouts in the 3.0+ range.
I’ve only taken a quick look, but if you have workout data that would suggest 250 is a better fit for you than 235, please share it. I’d be happy to take a look and ensure that what we’re recommending is accurate.
My primary concern is the TR FTP prediction is accurate so it can be used with Best Bike Split.
I’ve been using TR for years but only for zone 2 indoor rides. All other rides were outdoors and unstructured and loaded into TR from Strava. For the past month, I’ve been using a TR training plan. I didn’t put much thought into RPE, but seems to play into the into the algorithm, and lately have calibrated my RPE responses.
There was enough data in my TR account to make a prediction, but maybe since most of the rides were outdoors, the 30 day AI window may have had difficulty with the outdoor rides. Possibly the window for the initial calculation needs to be more than 30 days.
The 235 FTP workouts have been okay. Indoors is always harder than outdoors. In 9 days the predicted FTP is 257, which is still a very large jump in 30 days.
On Jan 28, I did an outdoor 20 minute FTP test as a reference. The 20 minute power was 276W, the 95% power calculation is 262. I understand this is just a ballpark value, where it could be 93%, 94%, etc.
I’d say that’s a pretty good example of AIFTP being darn close!
My power is ALWAYS better outside than inside, so for you to get 262 from an outside test while AI FTP Prediction currently estimates 262 is definitely reassuring. Let me know what we give you when FTP detection is unlocked!
.