Unless you take a particular flight month in month out over a number of years, and remember how the inventory changes in the days, weeks and months prior to the flight, you will have little idea how likely it is that different booking classes will fill up. And, even if you had all that data, a) you might design a different algorithm from UA's management team's algorithm, b) UA might have changed its approach to selling premium space or c) other extraneous factors might change things.
So, as a predictive tool, Expert Mode is next to useless.