It might not be immediately obvious how we can model rolling averages as observables. What we need to keep in mind is that pretty much anything we can think of as a sequence of values, we can probably model as an observable sequence.
Rolling averages are no different. Let's forget for a moment that the prices are coming from a network call wrapped in an observable. Let's imagine we have all of the values we care about in a Clojure vector:
(def values (range 10))
What we need is a way to process these values in partitions or buffers—of size 5—in such a way that only a single value is dropped at each interaction. In Clojure, we can use the partition function for this purpose:
(doseq [buffer (partition 5 1 values)] ...
