Hi, this is my first HN submission so please forgive the rough edges.<p>For this project gathered chess positions from some well-known chess players with a lot of games. (Usually ~25k games and 1M+ positions. The most for any player was GM Aman Hambleton with 1.8M positions indexed.)<p>With each board position I create a probability distribution of moves the player has made from that position. Then, I simply draw from that distribution (with sampling temperature) to choose a move. As long as there's at least one game with a given position, you can keep playing.<p>You'll see that it's pretty easy to get "out of book", where you've reached a novel position. It is possible to indirectly model a player's behavior throughout the game, so we can guess what moves a player might make in that position. But that's a bit more than a weekend project and a bit more involved than a lookup table.