I published a new version of my model for estimating doubling cube actions in backgammon by approximating the game evolution as a series of jumps in cubeless win probability.

For anyone interested in backgammon doubling cube theory: I'm working on a new model for calculating cube decision points. It models the probability of winning a game as a stochastic process that jumps from turn to turn, extending earlier models where that probability diffuses continuously. Posted on arxiv.org.

http://arxiv.org/abs/1203.5692