Is there a time decay curve applied to training history for relevancy in the current model of TR AI?
I ask because my power data has been skewed for the last year on outdoor rides due to having the wrong crank length set for my Assioma pedals. I’ve complained about the discrepancy of my pedals and my wahoo trainer in the past… problem solved.
I know this is within a margin of a few percent and likely to not have an impact on my structured training. It did however get me thinking about how the software performs and implements decay of older fitness in favor of more recent results.