Has anyone attempted to make a bot?

I hope there are no secret bots playing here now, but I wonder how good one would be. My guess is that from sometime around Move 33, the bot could outplay any human. But getting it to equal a human in the earlier part of the game would take more work based on positional considerations than any one programmer would likely put in.

nanobrain-ai played in the top league of StreetSoccer Championships 19 and 20, so I guess they're pretty good.

I don't see the significance of Move 33. Unlike, say, Amazons or Dvonn, StreetSoccer doesn't have a distinct midgame and endgame.

I think there are slight differences between good squares for scoring the next goal (both offense and defense) and good squares for outscoring your opponent in the longer run.

I agree that there is a difference, for example if you are behind you should take more risk I think. Maybe take a spot where you have 1/2 to score and 1/6 to get a goal instead of 1/6 to score and 0 to get a goal the next move.

There is also a threshold: When the game tree becomes small enough to fully explore, then your ai can play perfect. But the additional time makes it more complicated. I doubt it's already move 33.

If I remember correctly there were also few positions analyzed somewhere in the forum (maybe on german). And there was a sentence like "the evaluation says that moves X and Y are the best", which made me think that this sounds like it was evaluated by a bot.