It's the spinning of the ball. Picture the ball segmented through its axis with a plane parallel to the ground. The top half is spinning forward (forward relative to its translating motion). The forward rotation increases the drag of the ball due to the no-slip boundary condition of the air flow at the immediate surface of the ball. The bottom half is the opposite. The ball is rotating away from the translated motion and the drag is less. The resultant vector of the drag force on the ball is pointed more toward the ground.

I would suggest doing a search to get a good technical answer to this. It will have to do with changes in pressure most likely.