I worked on a project related to this for the state of California, back when they were originally trying to justify peak hour pricing (which happened at random hours; generally when it was super hot out and everyone turned the AC on) and smart grid stuff. The result published was that all of this sort of thing miraculously worked and people optimized their bills, theoretically allowing the power company to retire the expensive jet engine generators they used for peak days. If you looked more carefully, you'd see there was high compliance the first peak day, and pretty much none after that. Of course when they published their result, it looked great; I assume they justified dropping some of those non compliant days somehow. Anyway, there were billions in grants and investment dollars at stake, which is probably reason enough to fudge the stats.

Hindsight it was super obvious what was happening if you had been to those parts of the state, and looking at the temperatures. Megastores were turning off their airconditioning when it was hot out in crazy desert places -at least the first time, after installing the fancy smart grid hardware. This was not, as they say, sustainable, so they didn't persist.

Anyway, your doing an hours long data science exercise to get best use of your new power plan made me think of this. I mean, it's obvious you're a motivated user with investments in your own personal power grid. I'm trying to imagine my grandmother doing this.