I'm also guilty of spending lots of time daydreaming about robocode but never releasing a bot. Have you spent much time thinking about mini, micro or nano bots? I found the lack of space to over engineer frameworks really helped push me from daydreaming to releasing my first bot.
I should do something smaller first, but a lot of my ideas are probably larger than a minibot could handle. Admittedly it's probably a good way to test some of the weirder concepts out, though!