Well, my modifications for the EvilTwiki bot in last week's RPA run were good enough for a 3rd place finish, which I consider a pretty good showing.
I based my changes on an idea that my kids suggested early in the competition: figure out what the other bots are doing and then stop that. Seems simple enough, but it went right over my head until later in the week when I noticed that I could check the TeamName property of the opponent's IPlayer interface. That meant I could craft a different strategy for the bots that did something special or tricky, and in effect, play the man (bot) not the cards.
So, I wrote one basic strategy that could beat about 4 out of 5 bots, and the more sophisticated ones got more specific strategies that countered what they were doing... Not as elegant as writing an optimal rock-paper-scissors-dynamite-water balloon algorithm, but still pretty effective!