Pattern Recognition - now you're talking my specialty. :)

However, I do everything in assembly language.

Min/Max is really trivial. You need to first cosider how you plan to deal with ambiguous states. For example, what do you intend to do when two or more places have the same min or max magnitude?

If its merely a characterization of the sample and not LOCATING that sample point for min/max, no ambiguity exists. The algorithm is a simple one pass operation. As you pass once through the table, maintain a RAM variable for the current MAX and MIN variable encountered. Intialize them to the opposite extremes or the first live data point. As you traverse the sample data, replace the MAX if the new value is higher than the last MAX stored and replace the MIN if the new value is lower that the last MIN stored.

As I said, this is really trivial even in BASIC.

aka j