I coded an AI for Tic-Tac-Toe on my Ti-86 that never loses. But instead of giving you its strategy, I'll give you the suboptimal strategy: http://plaza.ufl.edu/aycock/ttt.pdf There is a 100%-effective not-losing strategy, and if your opponent doesn't know it, it translates into about a 70% win rate.
I have played one game of "Ultimate Tic Tac Toe" in my life. And I'll never play it again. My opponent was fierce. The game lasted for 45 minutes. My head was killing me at the end. Even though I was once very into chess, I was not prepared for this game. Essentially, it's a huge Tic Tac Toe board (9 squares.) And in each square is a smaller Tic Tac Toe board. Where your opponent moves on a smaller board is where you need to play your next move. So, if he (in the top left square of the big board) marked the center square, your next turn has to be played in the center square. The object of the game is to have either a Tic Tac Toe on the big board, or more wins of the smaller boards. If you direct your opponent to a square on the big board that's already been completed, he has to option of going anywhere he wants. If you play it, you'll see how complicated it can get.
