As @hazzey mentioned in a comment under the OP, if nothing is going to break, then the choice is mostly arbitrary. So I'm going to assume that things do break.
A structural element can basically collapse in three ways: excessive axial force (tension or compression), shear force or bending moment. Each of these is calculated differently.
For simple things like a table, one can assume a simple calculation for the maximum supported tensile force $T_R$:
$$T_R = f_y\cdot A$$
where $f_y$ is the allowable tensile stress and $A$ is the total cross-section area.
Shear force can also be calculated in a similar fashion:
$$Q_R = f_q\cdot A$$
where $f_q$ is the allowable shear stress. For steel one often uses $f_q = 0.6f_y$.
Bending moment has a series of complications due to lateral buckling. However, since that probably won't be the limiting factor in the design of your table, I'll skip over them as well. Therefore, we can take simply that the maximum stress at any point in the beam must be less than $f_y$. The applied stress can be obtained via
$$ \sigma = \dfrac{My}{12EI}$$
where $M$ is the bending moment, $y$ is the height from the centroid of the desired point, $E$ is the material's elastic modulus, and $I$, the cross-section's second moment of area (aka, moment of inertia).
I have left compression for last because it's the one with complications we can't avoid, since they will probably be the controlling factors. For short segments, the maximum allowable compressive force is equal to the maximum tensile force $T_R$ presented above. That being said, due to buckling, this usually isn't the case. Buckling is a pain and has a bunch of complications in and of themselves, but for a ballpark value, we can use Euler's equation:
$$P_E = \dfrac{\pi^2EI}{(KL)^2}$$
where $K$ is a factor depending on the element's support conditions. $P_E$ is non-conservative, meaning the actual allowable compressive force will certainly be lower than this.
There is also another way the table can fail and that is if it doesn't suit your needs. It's possible for the table to support the desired load but to do so by deforming completely, making it unappealing and unusable. Deflections, however, don't have simple equations and depend on the current conditions.
So, now we need to find the stresses that occur in your table. This is its structural model (the partial dimensions are placeholders and change according to the selected angle and value of length $A$. The load is also a placeholder):
The balancing act that is required is due to the fact that:
- if you increase the table span by moving the bracing closer to the base, you're going to increase the bending moment and shear force on the table. The deflections are also going to increase.
- if you lengthen the bracing (up to its maximum 19", for instance), you can reduce the stresses described above on the table, but lower the buckling load for the bracing.
- if you open up the bracing (increase its angle to the vertical), you can reduce the span of the table but you reduce the bracing's effectiveness.
The optimal solution needs to balance all of these considerations and it goes beyond the scope of this answer to actually obtain it (especially given that it depends on the materials and cross-sections). To perform the structural analyses, a 2D frame analysis tool like Ftool (free) can be used. That being said, 45 degrees is usually a good bet.
