This will be a very difficult (if not impossible) task. The speed at which the ball rotates and the relative accelerations of the ball will mean that any approximations of the change in velocity (dv=a*dt) will grow increasingly inaccurate. If the ball was just translating, not rotating you might be able to get this kind of crude approximation to work, but when you include rotation then the approximations made by this approach become too significant.