It looks like a few people are having similar or different aberrant % grade data recorded. When going over familiar roads with steep pitches, I do expect the 830 to lag in the % grade calculation. It does eventually come close to what I know actual slope is for sustained grades but may miss (often does) short pitches. This I expect for how the 830 collects data. What is a bit irritating is the data uploaded may shave 4-5% off the peak % grade. Perhaps there is some internal averaging tossing this out? I can share files collected but how could this be checked or baselined against a road section that a remote user cannot validate? Just bee classing this phenomenon as "missed data".