It may be beneficial to have the inward pointing arms of the yokes (opposite side to the spindles) as long as possible. The shorter they are, the greater the prying force on the arms of the head. Remember that the greater the difference in length between the above arms and those of the rocker, the greater the operational flapping differences between the three blades. 

Wild idea: If the spider ball is located just below the teetering rocker then there might be the opportunity to have exceptionally high coning angles (slow RRPM?) force the rocker down, which in turn will attempt to help force the spider ball, and therefore the collective, down.

Consider substituting a pair of gimbaled rings for the rocker. This device would be similar to an extremely strong swashplate. This will allow the mechanism to be kept outside the mast and the mast can have a smaller diameter.

This idea appears to be better than the Central Rocker idea because it handles the rotation problem and the Negative thrust problem.

There is nothing to stop the rocker from rotating about the vertical axis. This problem is not applicable to the gimbaled rings idea