Yeah I think the most confusing part of understanding this is that the number used to calculate the final_price Hive cost of the HBD isn't just a median average... it's a median average of list of median averages. It's a weird two dimensional nested average which isn't explicitly stated. It's also a bit weird to think that the hourly averages get aggregated over 1200 blocks of data and then those 1200 blocks of data basically get trashed to save disk-space and RAM.