My guess is they figured out who prints the lottery tickets, then gathered data on the ticket ID numbering system produced by that printer in the past(winner vs loser ticket ID's), then trained that data via deep learning to figure out probabilities of jackpot likelihood for each ticket ID # (or by range of numbers), then figured out the historic distribution patterns in relation to the ticket ID #'s, then pinpointed the specific places actually selling those ticket ID's, then drove to West Bumblefuck and bought all those tickets.