In simulations, are all the pairs of same angles differences contributing for shaping the good result? Do the computed correlations depend on the used angles set ? What would be the effect of lacks?
In real experiments, it is not possible to adjust the polarizers rotations at high speed over dozens of values. It is already a nice performance to switch at this controlled speed. There are only 2 possible positions per arm, but different on the 2 arms. The values are adjusted in such a way to find interesting ratios for the Bell inequalities.
The simulation takes all the integer rotations in degrees from -90 to 90 ° and then sums the outcomes by integer differences, then from -180 to 180°. Thus, a given difference is constructed by many different pairs.
After verification, the answer to the question is no. We took the angles set of a known experiment and randomized the simulation rotations only on these few values. The results are of the same quality as when using the entire range from -90 to 90 °.
Analytically, we can easily observe that the draft correlations function depends mainly on the h distribution.
I remind the little interest of new communication loophole refutations between the two polarizers. Even if the experiment becomes very slow, it is better to measure more differences for a denser curve. The current technology allows it.