At it's heart, Pomeroy is extremely simple. It comes down to points scored per offensive possession and points allowed per defensive possession. That's it. He doesn't care a fig about FG% or any other stat, except in how they affect points per possession. Points can come off a 3, a 2 or a FT, but to Pomeroy a point is simply a point, no matter how obtained. He reports a whole bunch of stats, but they're all outputs of the model, not inputs used to calculate his ratings. To me, that's the absolute beauty of it.
The sophistication of the model is reserved for how to adjust these basic points per possession to reflect the strength of the individual opponents and that's where the heft of mathematical statistics is employed.
As to trends, X, to the extent they are real, they are usually discernible only after they occur. It's as true for basketball as it is for the markets. There are lots of "random walks" masquerading as trends out there. Not telling you anything you don't know, of course...