The problem as I see it is the extrapolation based on a limited amount of data. We're trying to make predictions based on a cycle that has completed only three times.

A simple alternative, for instance, could be a repeating A-B-C pattern. 3 part patterns are commonly seen across various markets.

Or the pattern could be more complex. Here's another alternative analysis of the (well, a similar set of) data:


Source: https://twitter.com/CryptoKaleo/status/1009156602039971840