PerceptBot 2.3 also has a firing strategy that is, I believe, entirely my own. He only fires in one of four scenarios, based entirely off of the enemy's velocity and his approach angle. The first strategy is the innovative one: if the enemy's velocity is between 4 and 8, I assume he is slowing down, and I assume he is going to stop and eventually reverse, and I feed those assumptions into a predictor that moves the enemy bot forward in time based on those acceleration assumptions. Then the more mundane firing strategies: if the enemy bot is moving away from or towards me at a +/- 30 degree angle or less, I fire using a linear predictor. If the bot is very close to me, I fire using a linear predictor. And the last firing strategy, added to gain an obscene amount of wins against Winkie and TrackFire, is a small random chance of firing every so often at bots whose velocity is < 2.
Even though Pandora 1.2 and PerceptualDuelist 1.5 have traded spots with PerceptBot the last three times on the PerceptualSurvivalistLeague, PerceptBot is still #1 in my mind until someone comes along and crushes him completely. ;)