The simulation window should be rolling. As it stands, the more time goes by, the less utility the window has. It doesn’t seem right that today, when I’ve just gotten a new FTP, I have a lovely 30 days to plan and adjust for fatigue, but in the last week I’ve just about lost the functionality.

[image] Metrics on new AI FTP Detection + TrainerRoad AI Announcements I think the best solution is to have two detections: The first would be as far out as the system feels is sensible (at the moment I guess this would be four weeks) and that moves on a daily…

yeah, that would more or less work. I can’t believe that this wasn’t a topic for discussion as they were designing this. The decreasing usefulness with each day that goes by is just odd. To put it another way, it’s like you’re building up to a point (next prediction) that has nothing to do with you…

The simulation rolls, but the prediction does not. Most of us are concerned with what our FTP will be at a date that is farther out. A few of many challenges we’d run into with this are: potentially greater sensibly to current workouts and the prediction could bounce around even more Greater …

I dont know if there would be a cost difference, their model is refreshing every time I log into the platform, this would be similar, just with a different target date. I agree, i’d like to see the constant “FTP in 30 days” type of box. As I got closer to my detection(tomorrow) it went from growing…

It’s a compounding calculation that is done each time because it has to calculate what each workout will be to run the simulation. Right now the system has to calculate 14 days on average. If a prediction rolls at a steady 2 months out, that calculation moves from 14 days each time to 60.

My FTP prediction is off but historically I know the range of increase for the end of the 28 day cycle. I also would find a prediction more useful to be at a set point in time. In my case this would be for before my cycling trip. Thats my focus but it is in June. Adjusting work to optimize a 28 …

[image] jamesdrichey: I dont know if there would be a cost difference, their model is refreshing every time I log into the platform I noticed that too; not sure why it’s needed if no activities have been completed since? Might be a quick easy win to turn that off and only update on new activ…

I misspoke about the fatigue window. I think a 30 day rolling for prediction would be good, and then 60 and 90 great too with accompanying caveats that they’ve got the potential to shift around.

AI Trainining Simulation: Suggestion

TrainerRoad Software

Helvellyn February 6, 2026, 11:20am 2

Topic		Replies	Views
Look Ahead: See How TrainerRoad AI Helps Today’s Choices Make You Faster Tomorrow Announcements	22	1952	December 29, 2025
FTP Prediction Question TrainerRoad Software	0	53	February 9, 2026
How to Use the New AI FTP Prediction (Without Letting It Drive You Crazy) TrainerRoad Software	114	4234	May 1, 2026
Suggestion for FTP Prediction TrainerRoad Software	40	960	March 20, 2026
How to Disable FTP Prediction & AI Training Simulation when NOT Using TR Plans? TrainerRoad Software	17	881	March 8, 2026

AI Trainining Simulation: Suggestion

Related topics