I wrote a solver that's just brute-forcing solutions. Specifically, it's doing a depth-first search through a tree where every node is a move with four child mo

I wrote a solver that's just brute-forcing solutions. Specifically, it's doing a depth-first search through a tree where every node is a move with four child moves (left, up, right, and down). From what I can tell, the strategy is mostly what other people are doing by hand: go to one side, delay there to gather red dots for as long as possible, then move to an adjacent side. Repeat. The solver has a few tactical tricks as well, involving knowing how to get as close to red dots as possible without getting hit, and also how to move such that the dots group together. I just posted a game with a score of 346 where you can see this quite clearly. It's able to actually get in more than two full laps around.