Curling is a sport in which players slide stones on a sheet of ice towards a target area which is segmented into four concentric circles. Digital Curling can simulate the trajectory of the stones and calculate the resultant states after the delivery, so it can be used for supporting the tactics of curling. Different from the conventional testbed for artificial intelligence (AI), the game of Digital Curling require the software to consider the stochastic consequences. In this paper we propose the method to provide better tness estimation with less number of internal physical simulations for a player. The proposed estimation method to find the next move, is embedded to the AI program. By using the several simulation results, the effectiveness of the proposed method is discussed.