Since it looks slightly possible that there is a 24 hour effect, I'm changing the data taker to make 1 week plots with 4 hours per division, starting at 0:00 UTC today. I've added some filtering, too, since the action seems to be quite slow.