First one using the timing and spacing on the graph almost to a T (save for some minus spacing differences). Approximately 13 frames to drop, then 4 frames to bounce and fall again. It looks alright, but it's far too slow. That can be rectified very easily with some simple keyframe adjustments.
In the second version, I attempted to alter the speed of the initial drop. The bounce speed remains relatively unchanged. Now that I look at it from this angle it's not as bad as I thought, but the main reason I wasn't happy with this one was the drop speed. The ball was far too slow coming out of its highest point - it hung in the air for a very long time before hitting its max speed. Of course, in the render output, you won't see the ball begin to fall at all - you'll only see it enter the frame at high speed. So that might actually be more effective - I may come back to this version's timing, but I tried a third amendment just to see whether it would be any better. Vimeo is processing it currently, so... see you in half an hour.
My current concern with this is how simple it is. I know it's not the point to make it overly complex, but I can't help but feel it might look a bit "lazy" if it's just a straight up-and-down movement. I'm seriously considering trying to get the ball to roll across the ground. I think I'll attempt it anyway, just to see how it looks, but there's no reason I have to use it in the final render if I can't get it right (but that would be giving up...)