Purged CV arithmetic

Ten years of daily data ($N = 2520$), labels are 10-day forward returns ($t_1 = t_0 + 10$, clipped), 6-fold purged CV with a 1% embargo. (a) How many training samples does each fold have? Show the count for a middle fold. (b) What is the correlation between adjacent labels, and roughly how many independent labels are there? (c) With ordinary unpurged k-fold, how many training samples on each side of a middle test fold have label intervals overlapping the fold?