Love TrainerRoad, been using it for years, I love the idea of AI FTP detection to avoid tests and or having a guess myself.
I travel allot and have joined a gym in a different city with power based spin classes I update the data and regularly do an IF of 1.05+ when I do a TrainerRoad recommended workout I regularly set it to 105-110%. I suspect trainer road is not increasing my FTP because to much variation in input data like external riding without power data and gym work but would prefer to trust the AI.
Is there something TrainerRoad should do to factor in more external data when doing AI FTP detection?
I don’t know what spin bike you’re using in your gym but the wattage on bikes in gyms can and should be ignored. They are not calibrated to any strict standard. (Peloton bikes have been known to read as much as 80 watts higher than reality.) Best you can do in gyms is use RPE or heart rate.
Mixing external rides without power and spin classes can definitely throw off the algorithms a bit. Splitting data or manually adjusting might help keep things aligned.
It doesn’t really matter but note that an IF of 1.05 is a very hard workout (e.g. short anaerobic), while increasing the intensity of a workout by 5-10% will bump up its intensity by that much and, for example, might turn sweetspot into threshold.
You really need to fill in the RPE surveys for all activities – especially those with power data, like your gym sessions. Those workouts make up the majority of your training some weeks, and it’s really important for us to know how they feel.
Also, your last three hard TR workouts you completed (8/8, 7/25, 6/26) were all pinned by you, so they weren’t workouts that we prescribed.
The combination of you doing almost all of your hard workouts at the gym with a different power meter and not filling out those RPE surveys, alongside rarely doing hard TR workouts and hand-picking ones that are likely too easy, is likely the problem here.
I’d go back into your calendar and add RPE surveys for any of the recent gym workouts that you remember how the effort felt. Then let us pick the workouts you do in TR. You should get a solid FTP prediction after that.
Thanks, I have gone back and added the RPE, this has definitely helped lifting my predicted FTP. I haven’t been following the AI suggested workouts as, they have been too easy or not what I wanted to do based on resent training/health/sleep. This morning I did a ride that was not recommended at 105%, it was hard but I comfortable got through it.
It does’t surprise me this is hard to predict given my training varies between different indoor/outdoor equipment with different data points. Sometimes I just get bust and don’t sync all the data.
Great product and thanks for your help, I will persist with uploading data then follow recommended workouts as the FTP aligns to where it feels it should be.
