I’ve seen Alex’s results increasing over the years but had no clue he used TrainerRoad. One of our engineers is a friend of his and mentioned that Alex used TR this year to handle his training leading into world champs and I was thrilled to hear Alex was game to record a podcast about it!
We’re a bootstrapped company without the budget to pay the biggest names in cycling to use our product, but we do know that some of the worlds’ best use TR (some just as an training tool, and some as their coach), so it’s really motivational for all of us here at TR when we find out about these things.
Kudos to Alex for recording a podcast in a non-native language and for being so transparent about his training! What an incredible athlete.
Asked this question on YouTube about 30 seconds ago, but will ask here now I found this post.
Does the TR AI consider the non pedaling power (pumping) output/work of someone riding enduro/DH/BMX style riding, it seems to only take in consideration pedaling power output, as when I ride with a power meter my calory burn and TSS is quite low compared to the HR only based energy consumption estimation?
Seems enduro racing would be a good case for AI to smartly analyze both of these metrics and choose to intelligently take some TSS/Cal from either source where appropriate.
e.g. when pedaling up the hill utilize the power meter for TSS/Cal and then during the descent look at the speed combined with continued elevated HR data to make the decision to use HR for energy consumption then combine both to get a more accurate representation of the demands of Enduro/DH/BMX.
As of now it isn’t taking into account that non-pedaling stress as there is no objective way to measure that. HR is tricky as well, as if you just rely on that you are not guaranteed to increase the accuracy of a guess.
Would be great to get to the point where that stress could be quantified better!
To start with maybe just add an option (when completing the survey response) to ignore Power Meter data and use TSS based on HR Estimation.
When tracking my BMX open gates sessions the power based TSS rarely exceeds 20, where the same sessions with the power meter unpaired from my Garmin the TSS ranges from 60 up to 100 depending on what we are doing in the session. It might be flat sprints, gates, pumping laps or full track race laps.
I could keep disconnecting my power meter, but then I would lose the ability to track if my peak power is improving.
What we really need is the load sensors on the pedals to combine with grip load sensors to measure the forces being put through the bike, run this through an algorithm to get a true indication of how hard DH/Enduro/BMX athletes are working.
Something I’ve noticed about HR-based estimates by my watch is that they are not more accurate, but rather just a guess. The problem this presents is when you bring in a ton of data that is just noisy like that, it doesn’t improve things like our brains think it would.
I’ve found a more accurate approach to be me entering in a manual activity and manually entering the TSS based on what I felt.
Would be great to have a way to calculate all the other type of stress accumulated with non-pedaling efforts on the bike, but that sounds like a really complex hardware/software problem to solve!