If it's a single element, how in the world will you ever measure it
accurately enough to KNOW what the up-or-down tilt is? That is, the
directionality of a simple half-wave dipole just isn't enough to matter
when you get down to a couple of degrees. In freespace you MIGHT be
able to measure things that closely if you were extremely careful, but
over practical ground, I don't believe you have any hope.

If it's multiple vertically-stacked elements, then arrange the feed so
their currents are shifted in phase a small amount from element to
element. A back-of-the-envelope picture and some simple trig should
tell you just how much phase shift you should have between adjacent
elements.

Cheers,
Tom