Computer Learning for dice strategy

I was playing a dice game with my family, and after some thought i realized the entire game stratagem could be broken into three  or four numbers. I then realized that this problem would make an excellent foray into machine learning, hence this project.