I'm not sure there is any point to that.
As an exercise I implemented a Gaussian-distributed random number generator (RNG) in ladder, which included polynomial approximations for sine, cosine and natural logarithm accurate to six significant decimal digits or better, and which random numbers pass the Anderson-Darling normality test.
I challenge you to prove that has any practical value other than to see if it can be done, compared to loading an array of externally-generated random numbers and writing a trivial program to loop through them, or writing the same algorithms, but simpler, and far more easily understood, in ST. If you look through this or other PLC fora for requests for an RNG, the most common response is "Why would you want one?" I'm not saying there is never going to be an application (e.g. the Simon game, or a whack-a-mole game), but I doubt there will be a need for something that robust, and if there was it would be much easier to do it in ST. Does that make ST better? Of course not.
The question isn't whether "can X be done in ladder;" it's "what language is best to do X in context Y," where the context could include customer directives or preferences, who is going to look at and/or support the code, etc.
I think the answer is that context is King and drives the answer sometimes to ladder, sometimes to ST, and sometimes to summat else.